TWI887126B - Method of estimating state of charge of battery and system thereof - Google Patents

Abstract

The present invention discloses a method of estimating a state of charge (SOC) of a battery and a system thereof. The method of estimating the SOC of the battery includes following steps: calculating a voltage difference (ΔV) using a voltaic gauge based on a battery voltage (VBAT) and an open-circuit voltage (OCV); adaptively adjusting a gain (K) using a gain control engine based on a battery current (IBAT) and a full charged capacity (FCC); generating a present state of charge difference (ΔSOC_T) using the voltaic gauge based on the voltage difference (ΔV) and the adjusted gain (K); and generating a next state of charge (SOC_T+1) using an accumulator based on a present state of charge (SOC_T) and the present state of charge difference (ΔSOC_T).

Description

Method and system for estimating state of charge of a battery

The present invention relates to a method and system for estimating the state of charge of a battery, and more particularly to a method and system for estimating the state of charge of a battery with adaptively adjustable gain.

For users of portable electronic devices, the state of charge (SOC) is essential information. For a fully charged battery, the SOC is displayed as 100%. For a fully discharged battery, the SOC is displayed as 0%. It is urgent to use algorithms embedded in portable electronic devices to estimate the SOC. Previous technologies often use current coulomb integrators to progressively calculate the capacity of the battery when it is charged or discharged, and then combine it with the total battery capacity to calculate the battery's SOC. However, current coulomb integrators can produce cumulative errors due to inaccurate design or external noise, thereby estimating inaccurate battery SOC.

Among the battery management systems currently on the market, pure voltage battery gauges measure the battery voltage and estimate the battery SOC based on the correspondence between the voltage and the battery state of charge (SOC). The advantage of this method is that it is simple and easy to implement, and can refer to the open circuit voltage (OCV) curve to achieve stable SOC convergence and avoid divergence.

However, pure voltage battery gauges also have some obvious disadvantages listed below:

First, the SOC trend is wrong when the current changes dramatically: when the direction of charge and discharge remains unchanged, but the current changes dramatically, the pure voltage battery gauge may provide an erroneous SOC trend and fail to correctly reflect the actual state of charge of the battery.

Second. Inaccurate SOC change rate under different environmental conditions: Under different loads, different temperatures, different battery capacities or different aging conditions, pure voltage type fuel gauges may cause inaccurate SOC change rates and cannot reliably estimate the actual state of the battery.

Third, it cannot effectively cope with changing operating environments: Although pure voltage battery gauges have the advantage of relying on the OCV curve for stable convergence, they cannot fully cope with changes in different loads, temperatures, battery capacities, and aging levels, and cannot provide sufficient compensation to ensure the accuracy of SOC.

Therefore, existing pure voltage battery gauges will produce inaccurate SOC estimation results under certain extreme conditions, which makes it necessary to improve this type of technology to solve the above problems.

In view of this, the present invention aims at the above-mentioned deficiencies of the prior art and proposes a method and system for estimating the state of charge of a battery with adaptively adjustable gain.

In one aspect, the present invention provides a method for estimating the state of charge (SOC) of a battery, comprising: (a) using a battery gauge to calculate a voltage difference (△V) according to a battery voltage (VBAT) and an open circuit voltage (OCV); (b) using a gain control engine to adaptively adjust a gain according to a battery current and a fully charged capacity (FCC); (c) the battery gauge generates a current charge change (△SOC_T) according to the voltage difference (△V) and the adjusted gain; and (d) using an accumulator to generate the next moment charge state (SOC_T+1) according to the current state of charge (SOC_T) and the current charge change (△SOC_T).

In a preferred embodiment, the step (b) further includes: the gain control engine further adaptively adjusts the gain according to the gain (K) and the voltage difference (△V) before adjustment.

In a preferred implementation, step (a), step (b), step (c) and step (d) are performed in sequence, and after step (d), the state of charge at the next moment is used as the current state of charge, and the process returns to step (a).

In a preferred embodiment, in step (b), the gain control engine adjusts the gain by increasing or decreasing a fixed gain difference.

In a preferred embodiment, in step (b), the gain control engine further adjusts the gain according to a current difference between a battery current (IBAT_T-1) at a previous moment and the battery current (IBAT).

In a preferred embodiment, in step (b), when the voltage difference does not exceed a preset voltage difference or the battery current does not exceed a preset current, it is defined as a light load condition, and the gain control engine adjusts and maintains the gain as a fixed light load gain.

In a preferred embodiment, the step (b) includes: using the gain control engine to compare a charge change in the previous moment (△SOC_T-1) with a change target value (△SOCI), and adaptively adjusting the gain, wherein the change target value is related to the battery current and the fully charged capacity (FCC)

In a preferred embodiment, the full charged capacity (FCC) is related to a load condition, a battery temperature and/or a battery aging degree.

In a preferred embodiment, the step of generating a charge variation according to the voltage difference and the gain includes: estimating a weight according to the battery voltage using a voltage weight model, wherein the voltage weight model is a first preset relationship between the battery voltage and the weight in a battery information collected during charging, discharging and relaxing of the battery; estimating a fuzzy voltage difference according to the voltage difference using a voltage difference model, wherein the voltage difference model is a second preset relationship between the voltage difference and the fuzzy voltage difference; and generating the charge variation according to the weight, the fuzzy voltage difference and the gain.

In a preferred embodiment, the method for estimating the state of charge of a battery further includes: collecting the battery information at different charge/discharge currents before generating the state of charge at the next moment.

In a preferred embodiment, the method for estimating the state of charge of a battery further includes: establishing the voltage weight model and the voltage difference model by estimating the current state of charge and the battery voltage at different charge/discharge currents.

In a preferred embodiment, the method for estimating the state of charge of a battery further includes: calculating the weight according to the difference between the battery voltage at the charge/discharge current and at different charge/discharge currents to establish the voltage weight model.

In a preferred embodiment, the method for estimating the state of charge of the battery further includes: calculating the fuzzy voltage difference according to the charge/discharge current to establish the voltage difference model.

In a preferred embodiment, the method for estimating the state of charge of a battery further includes: querying a preset state of charge and open circuit voltage relationship table according to the current state of charge to generate the open circuit voltage.

From another perspective, the present invention provides a system for estimating the state of charge of a battery, comprising: a battery gauge (Voltaic Gauge), used to calculate a voltage difference (△V) according to a battery voltage (VBAT) and an open circuit voltage (OCV); a gain control engine (Gain Control Engine), used to adaptively adjust a gain according to a battery current (IBAT) and a fully charged capacity (FCC); and an accumulator, used to generate the state of charge (SOC_T+1) at the next moment according to the current state of charge (SOC_T) and a current charge change (△SOC_T); wherein the battery gauge generates the current charge change (△SOC_T) according to the voltage difference (△V) and the adjusted gain.

In a preferred embodiment, the gain control engine further adaptively adjusts the gain based on the gain (K) and the voltage difference (△V) before adjustment.

In a preferred embodiment, the gain control engine adjusts the gain by increasing or decreasing a fixed gain difference in an iterative manner.

In a preferred embodiment, the gain control engine further adjusts the fixed gain difference according to a current difference between a battery current (IBAT_T-1) at a previous moment and the battery current (IBAT).

In a preferred embodiment, the gain control engine defines a light load condition when the voltage difference does not exceed a preset voltage difference or the battery current does not exceed a preset current, and the gain control engine adjusts and maintains the gain to a fixed light load gain.

In a preferred embodiment, the gain control engine compares the charge variation with a variation target value and adaptively adjusts the gain, wherein the variation target value is related to a reference battery current and a fully charged capacity (FCC).

In a preferred embodiment, the battery meter includes: a weighted fuzzifier for estimating a weight based on the battery voltage using a voltage weight model, wherein the voltage weight model is a first preset relationship between the battery voltage and the weight in battery information collected during charging, discharging, and relaxing of the battery; a voltage difference fuzzifier for estimating a weight based on the battery voltage using a voltage weight model; A voltage difference is estimated using a voltage difference model, wherein the voltage difference model is a second preset relationship between the voltage difference and the fuzzy voltage difference; a multiplier is used to perform a multiplication operation on the weight and the fuzzy voltage difference to obtain the product of the weight and the fuzzy voltage difference; and a compensator is used to compensate or calibrate the product of the weight and the fuzzy voltage difference according to the gain to generate the charge change.

In a preferred implementation form, the weighted fuzzifier and the voltage difference fuzzifier establish the voltage weight model and the voltage difference model according to the corresponding different current states of charge and different battery voltages under different charging/discharging currents.

In a preferred embodiment, the weighted fuzzifier calculates the weight based on the difference between the battery voltages at different charging/discharging currents to establish the voltage weight model.

In a preferred embodiment, the voltage difference fuzzifier calculates the fuzzy voltage difference based on the difference between the voltage differences at different charge/discharge currents to establish the voltage difference model.

In a preferred embodiment, the battery meter further includes a lookup table for looking up a preset state of charge and open circuit voltage relationship table according to the current state of charge to generate the open circuit voltage.

In a preferred embodiment, the gain control engine further compensates the gain based on a fully charged capacity (FCC), a battery temperature (TBAT) and/or a battery aging degree.

In a preferred embodiment, the accumulator uses the inverse Z transformation of the current state of charge to accumulate at least one previous charge change to generate the next moment's state of charge.

The present invention is superior to the prior art in at least the following aspects:

1. In both charging and discharging conditions, the present invention can provide the correct state of charge even when the charge/discharge current changes dramatically.

2. Compared with the prior art, the present invention provides a more accurate state of charge (SOC) under different battery loads, different battery temperatures, different battery capacities and/or different battery aging levels.

3. Compared with the previous technology, the present invention has better response under extreme current conditions.

4. Compared with the prior art, the present invention provides a simpler and more convenient method to adjust the gain according to the test results to improve the accuracy of the state of charge (SOC).

The following detailed description is based on specific embodiments, which will make it easier to understand the purpose, technical content, features and effects of the present invention.

100: System for estimating the state of charge of a battery

110:Battery meter

111: Weighted Fuzzer

112: Pressure difference fuzzer

113:Multiplier

114: Compensator

115:Optimizer

116: Table Lookup

117: Subtraction Device

120: Gain control engine

130: Accumulator

310,320,410,420,610,620:Table

330,430,630:Relationship diagram

340,440: Standardized curve graph

510: Pressure difference model, combined with Figure 340 and Figure 440

640: Voltage weight model curve

710:Data table

810: Algorithm, similar to system 100 for estimating battery state of charge

812: Least mean square optimization function block

816: optimizer, connected to the least mean square optimization function block 812

820: Figure showing the battery voltage VBAT corresponding to algorithm 810

830: Figure showing the state of charge SOC corresponding to algorithm 810

910: Figure showing the error in state of charge SOC at a standard charge/discharge rate of 0.5C

920: Figure showing the error in state of charge SOC at a standard charge/discharge rate of 0.25C

930: Figure showing the error in state of charge SOC at a partial charge/discharge rate of 0.5C

1000: Method, method for estimating battery state of charge

1002,1004,1006,1008: Steps

AGS: Battery aging degree

FCC: Fully charged capacity

IBAT:Battery current

IBATth: default current

K: Gain

OCV: Open Circuit Voltage

SOC: State of charge

SOC_T: Current state of charge

SOC_T+1: charging status at the next moment

TBAT:Battery temperature

VBAT: battery voltage

△V: voltage difference

△SOC_T: Current charge change

W: Weight

f△V: fuzzy pressure difference

Wf△V: product of weight and fuzzy pressure difference

Z-1: inverse Z transformation

FIG1 shows a hardware block diagram of a system for estimating the state of charge of a battery according to an embodiment of the present invention.

Figure 2A shows a schematic diagram of the relationship between battery current IBAT and time.

Figure 2B shows a schematic diagram of the relationship between battery voltage VBAT, open circuit voltage OCV and time.

FIG3 shows a block diagram of a hardware system for estimating the state of charge of a battery according to another embodiment of the present invention.

FIG4 shows a hardware block diagram of a system for estimating the state of charge of a battery according to an embodiment of the present invention.

Figures 5A and 5B are schematic diagrams showing the related results of the weighted fuzzifier 111 and the pressure difference fuzzifier 112 in the algorithm for estimating the state of charge of the battery.

FIG6 shows an embodiment of modeling a portion of the pressure difference fuzzifier 112.

FIG. 7 shows an embodiment of establishing another part of the model of the pressure difference fuzzifier 112 shown in FIG. 4 .

FIG8 shows an embodiment of model building of the pressure difference fuzzifier 112 shown in FIG4.

FIG9 shows an embodiment of model building of the weighted fuzzifier 111 shown in FIG4 .

FIG10 shows a hardware block diagram and data table required for a system for estimating the state of charge of a battery according to an embodiment of the present invention.

FIG11 shows the experimental results obtained after applying the algorithm described in an embodiment of the present invention to the least mean square optimization.

FIG12 shows the experimental results obtained by estimating the state of charge of a battery using the system 100 for estimating the state of charge of a battery in the embodiment shown in FIG4 .

FIG13 shows a flow chart of a method for estimating the state of charge of a battery according to an embodiment of the present invention.

The diagrams in this invention are schematic, and are mainly intended to show the coupling relationship between the circuits and the relationship between the signal waveforms. The circuits, signal waveforms and frequencies are not drawn to scale.

The present invention relates to a method and system for estimating the state of charge (SOC) of a battery when the battery is in at least one of the following states: charging, discharging, and relaxing. The present invention is a method and system for estimating the state of charge of a battery using battery voltage (VBAT) and battery current (IBAT).

FIG1 shows a block diagram of a hardware system for estimating the state of charge of a battery according to an embodiment of the present invention. As shown in FIG1 , the present embodiment provides a system 100 for estimating the state of charge of a battery. The system 100 for estimating the state of charge of a battery includes a battery meter 110, a gain control engine 120, and an accumulator 130. The battery meter 110 is used to calculate the voltage difference △V according to the battery voltage VBAT and the open circuit voltage OCV. The gain control engine 120 is used to adaptively adjust the gain K according to the battery current IBAT and the fully charged capacity FCC to provide it to the battery meter 110. The battery meter 110 generates the current charge change △SOC_T according to the voltage difference △V and the adjusted gain K. The accumulator 130 is used to generate the next state of charge SOC_T+1 according to the current state of charge SOC_T and the current charge change △SOC_T.

According to the present invention, the open circuit voltage OCV, the voltage difference △V, the battery voltage VBAT, and the current state of charge SOC_T are calculated by iterative operation, and the gain K is compensated by the gain control engine 120 to generate the current state of charge change △SOC_T.

Unlike the traditional method based solely on voltage, the previous technology directly uses the battery voltage VBAT and the open circuit voltage OCV to generate the next moment of charge state SOC_T+1, which easily leads to sudden jumps in the charge state. The voltage and current compensation gain combined sensing architecture of the present invention can effectively prevent such sudden jumps and maintain the stability of the charge state through a more stable architecture.

One of the advantages of the present invention is that it avoids the sudden changes in the state of charge caused by the traditional pure voltage algorithm, achieves a smoother and more stable state of charge estimation, and accurately compensates through the gain control engine to adapt to different operating conditions.

In one embodiment, the gain control engine 120 adaptively adjusts the gain based on, in addition to the battery current IBAT and the fully charged capacity FCC, the gain K and the voltage difference △V before adjustment, such as but not limited to the comparison of the product.

In one embodiment, the gain control engine 120 adjusts the gain K by increasing or decreasing a fixed gain difference in an iterative manner. For example, the procedure of estimating the state of charge of the battery is executed once in each cycle, that is, there is a cycle between the last moment, the current moment, and the next moment. In each cycle, the gain control engine 120 increases or decreases the gain K before adjustment by a fixed gain difference according to the battery current IBAT and the fully charged capacity FCC, or further according to the gain K before adjustment (the gain K before adjustment is defined as the gain K of the previous moment, and the gain K after adjustment is the current gain K) and the voltage difference △V to generate the current gain K, and repeats multiple cycles to continuously adjust the gain K to keep the adjusted gain K in a dynamic manner.

In one embodiment, the gain control engine 120 further adjusts the gain according to the current difference between the last moment battery current IBAT_T-1 and the current battery current IBAT. For example, FIG2A shows a schematic diagram of the relationship between the battery current IBAT and time. As shown in FIG2A, when the last moment battery current IBAT_T-1 is a non-light load current and relatively high, and the current battery current IBAT is a light load current and relatively low, the current difference △I exceeds the preset current difference threshold, and the system 100 for judging and estimating the battery state of charge operates in a light load state or a non-light load state to adjust the gain K. It should be noted that the so-called current and current difference refer to the absolute value of the current or the current difference. Of course, in another embodiment, the system 100 for estimating the battery state of charge can also be directly judged based on the absolute value of the battery current IBAT to operate in a light load state or a non-light load state.

In another embodiment, the gain control engine 120 may also define a light load condition when the voltage difference ΔV does not exceed the preset voltage difference or the battery current IBAT does not exceed the preset current IBATth, and the gain control engine 120 adjusts and maintains the gain K as a fixed light load gain. From another perspective, when the light load condition occurs, after the gain K is maintained at a fixed light load gain, the gain control engine 120 no longer makes adaptive adjustments to the gain K, or periodically calibrates the gain K with a relatively long period. FIG2B shows a schematic diagram of the relationship between the battery voltage VBAT, the open circuit voltage OCV, and time. As shown in FIG2B , in one embodiment, the voltage difference △V1 exceeds the preset voltage difference or the battery current IBAT does not exceed the preset current IBATth, which is defined as a non-light load state, and the voltage difference △V2 does not exceed the preset voltage difference, which is defined as a light load state. It should be noted that in the embodiment shown in FIG2A , the battery current IBAT is a negative value.

According to one embodiment of the present invention, under non-light load conditions, when the voltage difference △V is large enough, the battery current IBAT is used to dominate the calculation of the charge state change △SOC. The battery voltage VBAT and the battery current IBAT are combined with the sensing architecture to track and adaptively adjust the gain K by controlling the gain K, thereby optimizing the calculation of the charge state change △SOC and providing good transient response and short-term accuracy. Its performance is comparable to that of a Coulomb Counter, but it does not rely on charge count to calculate the charge state SOC, avoiding cumulative errors and saving power, without the need for full-time analog-to-digital converter (CADC) conversion.

Under light load conditions, since the change in the battery current IBAT measurement is small, the CADC error percentage increases, so the weight of the gain K is reduced, and the battery voltage dominates the calculation of the state of charge change △SOC, which is implemented in an adaptive manner using the voltage difference △V. The gain control engine 120 can also periodically calibrate the state of charge SOC in a relatively long cycle to provide a long-term stable state of charge SOC estimate.

It should be noted that the CADC error percentage indicates the percentage of the error generated in the process of converting analog signals to digital signals due to the limitation of conversion accuracy or other influencing factors. Specifically, when CADC converts analog signals (such as battery voltage or current) into digital signals, the conversion process may be affected by resolution, noise or quantization errors, resulting in errors between the digital signal and the actual analog signal. This error is usually expressed as a percentage to quantify and analyze the accuracy of CADC. In the battery management system, the error percentage of CADC will affect the measurement accuracy of voltage, current or other parameters, and thus affect the estimation accuracy of the battery state of charge SOC.

In summary, according to the present invention, under non-light load conditions, the battery current IBAT is used to dominate the calculation of the charge state change △SOC, achieving an accuracy equivalent to that of a coulomb counter and avoiding the problems of cumulative error and power consumption. Under light load conditions, the battery voltage VBAT is used to dominate the calculation of the charge state change △SOC, and the natural driving force is used to maintain a long-term stable charge state SOC, thereby further improving the reliability and stability of the system.

FIG3 shows a block diagram of a hardware system for estimating the state of charge of a battery according to another embodiment of the present invention. As shown in FIG3, the present embodiment provides a system 100 for estimating the state of charge of a battery. The system 100 for estimating the state of charge of a battery includes a battery gauge 110, a gain control engine 120, and an accumulator 130.

In this embodiment, the gain control engine 120 compares the charge change △SOC_T-1 of the previous moment with the change target value △SOCI, and adjusts the gain K accordingly, wherein the change target value △SOCI is related to the battery current IBAT and the fully charged capacity FCC. In one embodiment, the change target value △SOCI is positively related to the battery current IBAT. The voltage difference △V is positively related to the difference between the battery voltage VBAT and the open circuit voltage OCV. In one embodiment, the change target value is more related to the fully charged capacity FCC. For example, the target value of variation △SOCI is negatively related to the fully charged capacity FCC, and the target value of variation △SOCI is negatively related to the quotient of the fully charged capacity FCC divided by the battery current IBAT. For example, when the charge variation △SOC_T-1 of the previous moment is lower than the target value of variation △SOCI, the gain K is increased; and when the charge variation △SOC_T-1 of the previous moment is higher than the target value of variation △SOCI, the gain K is decreased.

It should be noted that Full Charged Capacity (FCC) refers to the maximum amount of electricity that a battery can store when fully charged. Full Charged Capacity FCC is related to the load condition, battery temperature TBAT and/or battery aging degree AGS. Full Charged Capacity FCC will gradually decrease with the use and aging of the battery, and is one of the important indicators for evaluating the health and performance of the battery. This is well known to those with general knowledge in this field and will not be elaborated here.

FIG4 shows a hardware block diagram of a system for estimating the state of charge of a battery according to an embodiment of the present invention. As shown in FIG4, the present embodiment provides a more specific embodiment of a system 100 for estimating the state of charge of a battery. In the present embodiment, the system 100 for estimating the state of charge of a battery includes a battery meter 110, a gain control engine 120, and an accumulator 130. The battery meter 110 includes a weighted fuzzifier 111, a pressure difference fuzzifier 112, a multiplier 113, a compensator 114, an optimizer 115, a table lookup 116, and a subtractor 117.

The weighted fuzzifier 111 estimates the weight W using the voltage weight model according to the battery voltage VBAT, wherein the voltage weight model is a first preset relationship between the battery voltage VBAT and the weight W in the battery information collected during the charging, discharging and relaxing of the battery. The voltage difference fuzzifier 112 estimates the fuzzy voltage difference f△V according to the voltage difference △V using the voltage difference model, wherein the voltage difference model is a second preset relationship between the voltage difference △V and the fuzzy voltage difference f△V. The multiplier 113 performs a multiplication operation on the weight W and the fuzzy voltage difference f△V to obtain the product of the weight and the fuzzy voltage difference Wf△V. The compensator 114 compensates or calibrates the product Wf△V of the weight W and the fuzzy pressure difference f△V according to the gain K to generate the charge change △SOC. In one embodiment, the optimizer 115 can apply an additional gain to the weighted charge change △SOC for optimization.

In this embodiment, the accumulator 130 accumulates the weighted charge change ΔSOC by, for example but not limited to, inverse Z transformation to determine an estimated state of charge SOC. Then, the estimated state of charge SOC is fed back to the lookup table 116 to generate an estimated open circuit voltage OCV. The lookup table 116 is used to query a preset state of charge and open circuit voltage relationship table according to the state of charge SOC to generate the open circuit voltage OCV. The system 100 for estimating the state of charge of a battery repeatedly performs the above steps to estimate the state of charge SOC.

Figures 5A and 5B show the correlation results of the weighted fuzzifier 111 and the pressure difference fuzzifier 112 in the algorithm for estimating the state of charge of the battery. Figures 5A and 5B show the correlation results before estimating the current state of charge SOC in estimating the state of charge of the battery.

Figure 5A shows the relationship between the battery voltage VBAT and the state of charge SOC measurement results under different charging conditions. The charging condition OCV means that the battery can be charged 2% per hour; the charging condition 0.5C means that the battery can be charged 50% per hour; the charging condition 0.25C means that the battery can be charged 25% per hour. Figure 5A shows that under the same state of charge SOC, the higher the charging rate, the higher the battery voltage VBAT.

FIG5B shows the relationship between the measurement results of the battery voltage VBAT and the state of charge SOC under different discharge conditions. The discharge condition OCV represents that the battery can be discharged by 2% per hour; the discharge condition 0.5C represents that the battery can be discharged by 50% per hour; the discharge condition 0.25C represents that the battery can be discharged by 25% per hour; the discharge condition 0.15C represents that the battery can be discharged by 15% per hour; and the discharge condition 0.1C represents that the battery can be discharged by 10% per hour. FIG5B shows that under the same state of charge SOC value, when the discharge rate is higher, the battery voltage VBAT is lower. The following is a description of how to establish the voltage difference fuzzifier 112 shown in FIG4.

FIG6 shows an embodiment of establishing a portion of the differential pressure model of the differential pressure fuzzifier 112. FIG6 includes table 310, table 320, graph 330, and graph 340. Table 310 includes data captured from FIG5A. For example, when the state of charge SOC values are all 80%, for a charging condition where the battery can be charged 2% per hour, the battery voltage VBAT is 4000 mV. For a charging condition where the battery can be charged 25% per hour, the battery voltage VBAT is 4179 mV. In addition, when the state of charge SOC values are all 60%, for a charging condition where the battery can be charged 2% per hour, the battery voltage VBAT is 3850 mV. For a battery that can be charged 25% per hour, the battery voltage VBAT is 4023mV.

Table 320 is generated based on the information in Table 310. For example, when the state of charge SOC value is 80%, the open circuit voltage OCV of the battery (which means that the battery can be charged 2% per hour) is used as a benchmark. The voltage difference between the open circuit voltage OCV of the battery and the battery voltage VBAT at a charging condition of 0.25C (which means that the battery can be charged 25% per hour) is 179mV (which is the result of subtracting 4000mV from 4179mV). Similarly, when the state of charge SOC value is 60%, the voltage difference between the battery open circuit voltage OCV and the battery voltage VBAT at the charging condition of 0.25C (which means that the battery can be charged 25% every hour) is 173mV (that is, the result of subtracting 3850mV from 4023mV). By repeating the above steps, table 320 can be obtained. In addition, according to table 320, the relationship 330 between the voltage difference and the charging rate under different state of charge SOC can be obtained. After the relationship 330 is standardized, the curve 340 can be obtained.

FIG. 7 shows an embodiment of another partial model establishment of the differential pressure model of the differential pressure fuzzifier 112 shown in FIG. 4. FIG. 7 includes table 410, table 420, graph 430, and graph 440. Table 410 includes data extracted from FIG. 5B. For example, when the state of charge SOC value is 80%, for a discharge condition where the battery can be discharged by 2% per hour, the battery voltage VBAT is 4000mV. For a discharge condition where the battery can be discharged by 10% per hour, the battery voltage VBAT is 3964mV. In addition, when the state of charge SOC value is 60%, for a discharge condition where the battery can be discharged by 2% per hour, the battery voltage VBAT is 3850mV. For a discharge condition that can discharge the battery by 10% per hour, the battery voltage VBAT is 3795mV.

Table 420 is generated based on the information in Table 410. For example, when the state of charge SOC value is 80%, the battery open circuit voltage OCV (which means that the battery can be discharged 2% per hour) is used as a benchmark. The voltage difference between the battery open circuit voltage OCV and the battery voltage VBAT at a discharge condition of 0.1C (which means that the battery can be discharged 10% per hour) is 36mV (which is the result of subtracting 3964mV from 4000mV). Similarly, when the state of charge SOC value is 60%, the voltage difference between the open circuit voltage OCV of the battery and the battery voltage VBAT at the discharge condition of 0.25C (which means that the battery can be discharged 25% every hour) is 55mV (that is, the result of subtracting 3795mV from 3850mV). By repeating the above steps, Table 420 can be obtained. In addition, according to Table 420, the relationship 430 between the voltage difference and the discharge rate under different state of charge SOC can be obtained. After the relationship 430 is standardized, the curve 440 can be obtained.

FIG8 shows an embodiment of establishing a differential pressure model for the differential pressure fuzzifier 112 shown in FIG4. By combining curves 340 and 440, the present embodiment establishes a differential pressure model 510 for the differential pressure fuzzifier 112 shown in FIG4. The differential pressure model 510 shows that when the absolute value of the voltage difference △V between the battery open circuit voltage OCV and the battery voltage VBAT during charging/discharging is higher, the charging/discharging current is higher (corresponding to the charge change △SOC shown in FIG4, therefore, there is a V-shaped relationship between the two.

FIG9 shows an embodiment of establishing a voltage weight model of the weighted fuzzifier 111 shown in FIG4. FIG9 includes Table 610, Table 620, Graph 630 (showing the relationship between the battery voltage VBAT and the weight W) and Graph 640 (showing the voltage weight model). Table 610 includes data extracted from FIG5B. For example, when the state of charge SOC value is 90%, for a discharge condition in which the battery can be discharged by 2% per hour, the battery voltage VBAT is 4100mV. For a discharge condition in which the battery can be discharged by 10% per hour, the battery voltage VBAT is 4065mV. In addition, when the state of charge SOC value is 80%, for the discharge condition that the battery can be discharged by 2% per hour, the battery voltage VBAT is 4000mV. For the discharge condition that the battery can be discharged by 15% per hour, the battery voltage VBAT is 3952mV. In addition, when the state of charge SOC value is 70%, for the discharge condition that the battery can be discharged by 2% per hour, the battery voltage VBAT is 3900mV. For the discharge condition that the battery can be discharged by 25% per hour, the battery voltage VBAT is 3811mV.

Table 620 is generated based on the information in Table 610. For example, when the state of charge SOC value is 90%, the discharge condition OCV (which means that the battery can be discharged 2% every hour) is used as a benchmark. The weight of the battery voltage being 4.1V and being discharged 10% every hour is equal to 0.29 (this is calculated by 10/(4100-4065)). The weight of the battery voltage being 4.0V and being discharged 15% every hour is equal to 0.31 (this is calculated by 15/(4000-3952)). The weight of the battery voltage being 3.9V and being discharged 25% every hour is equal to 0.28 (this is calculated by 25/(3900-3811)). By repeating the above steps, table 620 can be obtained. In addition, according to table 620, the relationship 630 between the battery voltage VBAT and the weight W (as shown in FIG. 4 ) under the discharge current can be obtained. After the relationship 630 is standardized, the voltage weight model 640 required by the weighted fuzzifier 111 shown in FIG. 4 can be obtained.

FIG10 shows a hardware block diagram and a data table required for a system for estimating the state of charge of a battery according to an embodiment of the present invention. As shown in FIG10 , the voltage difference model 510 shown in FIG8 and the voltage weight model 640 shown in FIG9 are embedded in the system 100 for estimating the state of charge of a battery. Since the battery is in a discharged state, a graph 440 that is a part of the voltage difference model 510 is drawn. In addition, the present embodiment provides a data table 710 corresponding to each node in the system 100 for estimating the state of charge of a battery.

According to the voltage weight model 640, the weight W may be between 0.8 and 1.8 depending on the value of the battery voltage VBAT. In this embodiment, when the value of the battery voltage VBAT is 3.894Volts, the weight W applied to the output of the voltage difference fuzzifier 112 is 0.9.

The voltage difference △V is used as the input of the voltage difference fuzzifier 112. The voltage difference △V is obtained by subtracting the open circuit voltage OCV of the battery from the battery voltage VBAT. The state of charge SOC calculated by the estimated battery state of charge system 100 is input to the open circuit voltage (OCV) lookup table 116. As shown in the voltage difference fuzzifier 112, when the absolute value of the voltage difference △V is larger, the absolute value of the charge change △SOC output by the battery gauge 110 is larger. Figure 440 shows that when the voltage difference △V is -100mV, for example, the charge change △SOC is -0.25.

As described above, the charge variation △SOC calculated by the pressure difference fuzzifier 112 is weighted by the output weight W of the weighted fuzzifier 111 and optimized by the optimizer 115. In one embodiment, after the optimizer 115 performs weighting, a gain K is generated according to the least mean square optimization method and the actual data of battery charging/discharging and the gain control engine 120 according to the adaptive adjustment of the battery current IBAT. This gain K is used to calculate the weighted charge variation △SOC.

Next, the battery state of charge estimation system 100 adds the weighted charge change ΔSOC to the accumulator 130 (for example, but not limited to, using the inverse Z transformation of the state of charge SOC) to determine a new state of charge SOC. This new state of charge SOC is fed back to the open circuit voltage OCV lookup table 116, where this step is repeated. Table 710 shows 3 battery samples, each separated by 36 seconds. According to the above content about the battery state of charge estimation system 100, the operation mode of the battery state of charge estimation system 100 of this embodiment is: first determine the voltage difference, and apply multiple fuzzy calculation methods to the voltage difference, and then adaptively adjust the gain according to the gain control engine 120.

FIG11 shows the experimental results obtained after applying the algorithm described in an embodiment of the present invention to the least mean square optimization. As shown in FIG11 , the algorithm 810 is similar to the system 100 for estimating the state of charge of a battery, and the difference between the two is that the algorithm 810 has an additional least mean square optimization function block 812. The battery voltage VBAT and the state of charge SOC corresponding to the algorithm 810 are shown in FIG820 and FIG830 respectively. The least mean square optimization function block 812 receives the ideal value of the state of charge SOC measured by an external measuring instrument in FIG830 and the estimated state of charge SOC provided by the system 100 for estimating the state of charge of a battery in FIG830. The least mean square optimization function block 812 gradually fine-tunes the optimizer 816. According to the fine-tuning made by the least mean square optimization function block 812, different gains (gain #1 to gain #3) are applied to the optimizer 816. As a result, for example, gain #1 is the best result among the three, so gain #1 is selected as an optimal gain K. It should be noted that in this embodiment, for ease of understanding, the adaptive adjustment of the gain K by the gain control engine 120 according to the battery current IBAT is omitted, in order to illustrate the least mean square optimization operation.

FIG12 shows the experimental results obtained by estimating the state of charge of a battery using the system 100 for estimating the state of charge of a battery in the embodiment shown in FIG4 . FIG12 includes three graphs 910 , 920 , and 930 , which show the error of the estimated state of charge SOC under different charge/discharge conditions. FIG910 shows that at a standard charge/discharge rate of 0.5C, the error of the estimated state of charge SOC is between -3% and +3%. FIG920 shows that at a standard charge/discharge rate of 0.25C, the error of the estimated state of charge SOC is also between -3% and +3%. FIG930 shows that at a partial charge/discharge rate of 0.5C, the error of the estimated state of charge SOC is between -4% and +4%. Therefore, graphs 910, 920, and 930 may illustrate the accuracy of the system 100 for estimating the battery state of charge.

FIG13 shows a flow chart of a method for estimating the state of charge of a battery according to an embodiment of the present invention. The present embodiment provides a method 1000 for estimating the state of charge SOC of a battery, the steps of which include: first, using a battery gauge (Voltaic Gauge) to calculate a voltage difference (△V) according to a battery voltage (VBAT) and an open circuit voltage (OCV) (step 1002). Then, using a gain control engine (Gain Control Engine) to adaptively adjust a gain according to a battery current (IBAT) and a fully charged capacity (FCC) (step 1004). Then, the battery gauge generates a charge change (△SOC) according to the voltage difference (△V) and the gain (step 1006). Next, based on the current state of charge (SOC_T) and the charge change (△SOC), a state of charge (SOC_T+1) at the next moment is generated (step 1008). Then, the state of charge (SOC_T+1) at the next moment is used as the current state of charge (SOC_T), and the process returns to step 1002 to repeat the process.

It should be noted that the embodiment shown in FIG. 13 is an iterative method. When the state of charge needs to be estimated, the gain is dynamically adjusted and corrected in an iterative manner. Therefore, after step 1008 is completed, the state of charge (SOC_T+1) at the next moment is used as the current state of charge (SOC_T), and the operation is repeated in step 1002; when the state of charge does not need to be estimated or the gain does not need to be adjusted (for example, in one embodiment, the gain is fixed without dynamically adjusting the gain when the load is light), the process of method 1000 is stopped or paused.

In one embodiment, the step (step 1004) of adaptively adjusting a gain using a gain control engine according to a battery current (IBAT) further includes: adaptively adjusting the gain using the gain control engine according to the gain (K) before adjustment and the voltage difference (△V).

In one embodiment, in the step (step 1004) of adaptively adjusting a gain using a gain control engine (Gain Control Engine) according to a battery current (IBAT), the gain control engine adjusts the gain by increasing or decreasing a fixed gain difference.

In one embodiment, in the step (step 1004) of adaptively adjusting a gain using a gain control engine according to a battery current (IBAT), the gain control engine further adjusts the gain according to a current difference between a battery current (IBAT_T-1) at a previous moment and the battery current (IBAT).

In one embodiment, in the step (step 1004) of adaptively adjusting a gain using a gain control engine according to a battery current (IBAT), when the voltage difference does not exceed a preset voltage difference, it is defined as a light load condition, and the gain control engine adjusts and maintains the gain as a fixed light load gain.

In one embodiment, the step of adaptively adjusting a gain according to a battery current (IBAT) using a gain control engine (Gain Control Engine) (step 1004) includes: using the gain control engine to compare a charge change (△SOC_T-1) at a previous moment with a change target value (△SOCI), and adaptively adjusting the gain, wherein the change target value is related to the battery current and the fully charged capacity (FCC).

In one embodiment, the fully charged capacity (FCC) is related to a load condition, a battery temperature (TBAT) and/or a battery aging level (AGS).

In one embodiment, the step of generating a charge variation according to the voltage difference and the gain includes: estimating a weight according to the battery voltage using a voltage weight model, wherein the voltage weight model is a first preset relationship between the battery voltage and the weight in a battery information collected during charging, discharging and relaxing of the battery; estimating a fuzzy voltage difference according to the voltage difference using a voltage difference model, wherein the voltage difference model is a second preset relationship between the voltage difference and the fuzzy voltage difference; and generating the charge variation according to the weight, the fuzzy voltage difference and the gain.

In one embodiment, the method 1000 for estimating the state of charge of a battery further includes: before generating the state of charge (SOC_T+1) at the next moment, collecting the battery information at different charging/discharging currents.

In one embodiment, the method 1000 for estimating the state of charge of a battery further includes: establishing the voltage weight model and the voltage difference model by estimating the state of charge and the battery voltage at different charge/discharge currents.

In one embodiment, the method 1000 for estimating the state of charge of a battery further includes: calculating the weight according to the difference between the battery voltage at the charge/discharge current and at different charge/discharge currents to establish the voltage weight model.

In one embodiment, the method 1000 for estimating the state of charge of a battery further includes: calculating the fuzzy voltage difference based on the charge/discharge current to establish the voltage difference model.

In one embodiment, the method 1000 for estimating the state of charge of a battery further includes: querying a preset state of charge and open circuit voltage relationship table according to the state of charge to generate the open circuit voltage.

In order to establish the model used in the method of the present invention, the present invention uses standard charging and discharging processes to collect battery information. For example, the present invention observes the state of charge SOC and the battery voltage VBAT by different charging and discharging currents. Accordingly, based on these observations, the present invention is able to establish a partner function (or relationship) between the following two: (1) a voltage difference between the battery voltage VBAT and an estimated open circuit voltage (OCV) of the battery; and (2) a charge change △SOC used to adjust the estimated state of charge SOC. In addition, based on these observations, the present invention is able to establish another partner function (or relationship) between a weight applied between the charge change △SOC and the battery voltage VBAT. These two partner functions form a set of standard models that can be optimized based on specific battery charging and discharging information. Specific battery data is the most commonly used user experience. Furthermore, an optimal gain can be obtained by the minimum mean square error algorithm, and the gain is adaptively adjusted according to the sensed battery current IBAT, thereby further fine-tuning the charge change △SOC. In addition, the gain control engine adaptively adjusts the gain according to the battery current IBAT, so that the present invention can still provide the correct state of charge under charging and discharging conditions, even when the charging/discharging current changes dramatically; and can further provide a more accurate state of charge SOC according to different battery loads, different battery temperatures, different battery capacities and/or different battery aging levels.

The present invention has been described above with reference to the preferred embodiments. However, the above description is only for the purpose of making it easier for those familiar with the art to understand the content of the present invention, and is not intended to limit the scope of the invention. The embodiments described are not limited to single application, but can also be applied in combination. For example, two or more embodiments can be used in combination, and a part of the components in one embodiment can also be used to replace the corresponding components in another embodiment. In addition, under the same spirit of the present invention, those familiar with the present technology can think of various equivalent changes and various combinations. For example, the present invention refers to "processing or calculating or generating an output result according to a certain signal", which is not limited to the signal itself, but also includes, when necessary, converting the signal into voltage-current, current-voltage, and/or ratio, and then processing or calculating the converted signal to generate an output result. It can be seen that under the same spirit of the present invention, those familiar with the present technology can think of various equivalent changes and various combinations, and there are many combinations, which are not listed here one by one. Therefore, the scope of the present invention should cover the above and all other equivalent changes.

100: System for estimating the battery state of charge 110: Battery gauge 120: Gain control engine 130: Accumulator IBAT: Battery current FCC: Fully charged capacity OCV: Open circuit voltage SOC_T: Current state of charge SOC_T+1: Next state of charge VBAT: Battery voltage ΔSOC_T: Current charge change

Claims (27)

A method for estimating the state of charge (SOC) of a battery comprises: (a) using a battery gauge to calculate a voltage difference (ΔV) according to a battery voltage (VBAT) and an open circuit voltage (OCV); (b) using a gain control engine to adaptively adjust a gain (K) according to a battery current (IBAT) and a fully charged capacity (FCC); (c) using the battery gauge to generate a current charge change (ΔSOC_T) according to the voltage difference (ΔV) and the adjusted gain (K); and (d) using an accumulator to generate a state of charge (SOC_T+1) at the next moment according to a current state of charge (SOC_T) and the current charge change (ΔSOC_T). The method for estimating the state of charge of a battery as described in claim 1, wherein the step (b) further includes: using the gain control engine to adaptively adjust the gain based on the gain (K) and the voltage difference (ΔV) before adjustment. A method for estimating the state of charge of a battery as described in claim 1, wherein the step (a), the step (b), the step (c) and the step (d) are performed in sequence, and after the step (d), the state of charge at the next moment is used as the current state of charge, and the method returns to the step (a). A method for estimating the state of charge of a battery as described in claim 3, wherein in the step (b), the gain control engine adjusts the gain by increasing or decreasing a fixed gain difference. A method for estimating the state of charge of a battery as described in claim 4, wherein in the step (b), the gain control engine further adjusts the gain based on a current difference between a battery current (IBAT_T-1) at a previous moment and the battery current (IBAT). A method for estimating the state of charge of a battery as described in claim 1, wherein in the step (b), when the voltage difference does not exceed a preset voltage difference or the battery current does not exceed a preset current, it is defined as a light load condition, and the gain control engine adjusts and maintains the gain as a fixed light load gain. A method for estimating the state of charge of a battery as described in claim 1, wherein the step (b) includes: using the gain control engine to compare a charge change at a previous moment (ΔSOC_T-1) with a change target value (ΔSOCI), and adaptively adjusting the gain, wherein the change target value is related to the battery current and the fully charged capacity (FCC). A method for estimating the state of charge of a battery as described in claim 1, wherein the full charge capacity (FCC) is related to a load condition, a battery temperature (TBAT) and/or a battery aging level (AGS). A method for estimating the state of charge of a battery as described in claim 1, wherein the step (c) includes: estimating a weight based on the battery voltage using a voltage weight model, wherein the voltage weight model is a first preset relationship between the battery voltage and the weight in battery information collected during charging, discharging and relaxing of the battery; estimating a fuzzy voltage difference based on the voltage difference using a pressure difference model, wherein the pressure difference model is a second preset relationship between the voltage difference and the fuzzy pressure difference; and generating the charge change based on the weight, the fuzzy pressure difference and the gain. The method for estimating the state of charge of a battery as described in claim 9 further includes: collecting the battery information at different charging/discharging currents before generating the state of charge (SOC_T+1) at the next moment. The method for estimating the state of charge of a battery as described in claim 9 further includes: establishing the voltage weight model and the voltage difference model by estimating the current state of charge and the battery voltage under different charging/discharging currents. The method for estimating the state of charge of a battery as described in claim 11 further includes: calculating the weight based on the difference between the battery voltage at the charge/discharge current and at different charge/discharge currents to establish the voltage weight model. The method for estimating the state of charge of a battery as described in claim 11 further includes: calculating the fuzzy voltage difference based on the charge/discharge current to establish the voltage difference model. The method for estimating the state of charge of a battery as described in claim 1 further includes: querying a preset state of charge and open circuit voltage relationship table based on the current state of charge to generate the open circuit voltage. A system for estimating the state of charge (SOC) of a battery includes: a battery gauge for calculating a voltage difference (ΔV) according to a battery voltage (VBAT) and an open circuit voltage (OCV); a gain control engine for adaptively adjusting a gain (K) according to a battery current (IBAT) and a fully charged capacity (FCC); and an accumulator for generating a state of charge (SOC_T+1) at the next moment according to a current state of charge (SOC_T) and a current charge change (ΔSOC_T); wherein the battery gauge generates the current charge change (ΔSOC_T) according to the voltage difference (ΔV) and the adjusted gain (K). A system for estimating the state of charge of a battery as described in claim 15, wherein the gain control engine further adaptively adjusts the gain based on the gain (K) and the voltage difference (ΔV) before adjustment. A system for estimating the state of charge of a battery as described in claim 15, wherein the gain control engine adjusts the gain by increasing or decreasing a fixed gain difference in an iterative manner. A system for estimating the state of charge of a battery as described in claim 17, wherein the gain control engine further adjusts the gain based on a current difference between a battery current (IBAT_T-1) at a previous moment and the battery current (IBAT). A system for estimating the state of charge of a battery as described in claim 15, wherein the gain control engine defines a light load condition when the voltage difference does not exceed a preset voltage difference or the battery current does not exceed a preset current, and the gain control engine adjusts and maintains the gain as a fixed light load gain. A system for estimating the state of charge of a battery as described in claim 15, wherein the gain control engine compares a charge change at a previous moment (ΔSOC_T-1) with a change target value (ΔSOCI) and adaptively adjusts the gain, wherein the change target value is related to the battery current and the full charge capacity (FCC). A system for estimating the state of charge of a battery as described in claim 15, wherein the full charge capacity (FCC) is related to a load condition, a battery temperature (TBAT) and/or a battery aging level (AGS). A system for estimating the state of charge of a battery as described in claim 15, wherein the battery gauge includes: a weighted fuzzifier for estimating a weight based on the battery voltage using a voltage weight model, wherein the voltage weight model is a weighted fuzzifier for estimating the state of charge of a battery using a voltage weight model based on the battery voltage during charging, discharging and relaxation. A battery information collected during relaxation includes a first preset relationship between the battery voltage and the weight; a voltage difference fuzzifier for estimating a fuzzy voltage difference based on the voltage difference using a voltage difference model, wherein the voltage difference model is a second preset relationship between the voltage difference and the fuzzy voltage difference; a multiplier for performing a multiplication operation on the weight and the fuzzy voltage difference to obtain the product of the weight and the fuzzy voltage difference; and a compensator for compensating or calibrating the product of the weight and the fuzzy voltage difference according to the gain to generate the charge change. A system for estimating the state of charge of a battery as described in claim 22, wherein the weighted fuzzifier and the voltage difference fuzzifier establish the voltage weight model and the voltage difference model according to the corresponding different current states of charge and different battery voltages under different charging/discharging currents. A system for estimating the state of charge of a battery as described in claim 23, wherein the weighted fuzzifier calculates the weight based on the difference between the battery voltages at different charging/discharging currents to establish the voltage weight model. A system for estimating the state of charge of a battery as described in claim 23, wherein the voltage difference fuzzifier calculates the fuzzy voltage difference based on the difference between the voltage differences at different charging/discharging currents to establish the voltage difference model. A system for estimating the state of charge of a battery as described in claim 15, wherein the battery gauge further includes a lookup table for looking up a preset state of charge and open circuit voltage relationship table based on the current state of charge to generate the open circuit voltage. A system for estimating the state of charge of a battery as described in claim 15, wherein the accumulator utilizes an inverse Z transformation of the current state of charge to accumulate at least one previous charge change to generate the state of charge at the next moment.