TWI838016B - Electronic apparatus and fan speed adjustment method thereof - Google Patents

Abstract

An electronic apparatus and a fan speed adjustment method thereof are provided. The method includes the following steps. Multiple system current values at different time are monitored. The monitored system current values are sequentially fed into a machine learning model, and multiple load types are sequentially inferred through the machine learning model. The inferenced load types are used to check whether the inferenced load types satisfy a certain condition. When the load types satisfy the specific condition, a fan speed of the fan is adjusted to become higher according to the load type

Description

Electronic device and fan speed adjustment method

The present invention relates to an electronic device, and in particular to an electronic device and a fan speed adjustment method.

With the rapid development of technology, computer systems have become one of the important hardware foundations of the modern information society. In addition, the computing speed of computer systems is also increasing day by day to cope with the huge amount of computing and shorten the time of data processing. However, as the computing speed of computer systems increases, the heat generated by each electronic component in the computer system (especially the central processing unit) is also higher. Therefore, in order to allow the computer system to operate stably and continuously, a fan must be installed in the computer system to control the temperature of the system within a safe range.

At present, most existing fan speed control methods control the fan speed based on the system temperature to ensure the long-term stability of the system. Taking a laptop platform as an example, an embedded controller (EC) usually collects the temperature sensed by the temperature sensor on the motherboard or the temperature of a specific electronic chip (such as the central processing unit), and then dynamically adjusts the fan speed according to the preset fan table based on these temperature information. However, the current fan control mechanism adjusts the fan speed only when the system temperature or system performance has been determined to reach a specific condition. Therefore, it is possible that the system computing performance has been adversely affected by the high temperature, resulting in the central processing unit being unable to exert its maximum performance.

In view of this, the present invention proposes an electronic device and a fan speed adjustment method thereof, which can solve the above technical problems.

The present invention provides a fan speed adjustment method applicable to an electronic device including a fan, and includes the following steps. Monitor multiple system current values corresponding to different time points. Input the multiple system current values into a machine learning model in sequence, and predict multiple load types in sequence through the machine learning model. And, determine whether the multiple load types meet specific conditions. When these load types meet the specific conditions, increase the fan speed of the fan according to these load types.

The present invention provides an electronic device, which includes a fan and a control module. The control module is coupled to the fan and configured to perform the following steps. Monitor multiple system current values corresponding to different time points. Input the multiple system current values into a machine learning model in sequence, and predict multiple load types in sequence through the machine learning model. And, determine whether the multiple load types meet specific conditions. When these load types meet the specific conditions, increase the fan speed of the fan according to these load types.

Based on the above, in an embodiment of the present invention, the system current value is continuously monitored and sequentially input into the machine learning model, so that the machine learning model continuously predicts multiple load types associated with these input system current values. Therefore, the fan speed can be increased in response to these load types meeting specific conditions. Based on this, before the system overload actually comes, the fan speed can be increased in advance according to the model prediction results to avoid the occurrence of over-temperature as much as possible and strive for better system performance.

100: Electronic devices

110: Fan

120: Control module

130: Charging circuit

140:Port

200: Power adapter

121:Embedded Controller

122: Memory

123:Processor

L1~L4: Curve

S310~S350,S510~S540,S541~S543: Steps

FIG1 is a block diagram of an electronic device according to an embodiment of the present invention.

Figure 2A is a schematic diagram of a control module according to an embodiment of the present invention.

Figure 2B is a schematic diagram of a control module according to an embodiment of the present invention.

Figure 3 is a flow chart of a fan speed adjustment method according to an embodiment of the present invention.

FIG4A is a schematic diagram of system current of a system with idle load according to an embodiment of the present invention.

FIG4B is a schematic diagram of system current at a continuous low load according to an embodiment of the present invention.

FIG4C is a schematic diagram of system current under intermittent high load according to an embodiment of the present invention.

FIG4D is a schematic diagram of system current under continuous high load according to an embodiment of the present invention.

Figure 5 is a flow chart of a fan speed adjustment method according to an embodiment of the present invention.

Some embodiments of the present invention will be described in detail with reference to the accompanying drawings. The component symbols cited in the following description will be regarded as the same or similar components when the same component symbols appear in different drawings. These embodiments are only part of the present invention and do not disclose all possible implementation methods of the present invention. More precisely, these embodiments are only examples of methods and devices within the scope of the patent application of the present invention.

FIG1 is a block diagram of an electronic device according to an embodiment of the present invention. Referring to FIG1 , the electronic device 100 includes a fan 110, a control module 120, a charging circuit 130, and a connection port 140. The electronic device 100 may be, for example, a notebook computer, a server, or other electronic product with a fan heat dissipation mechanism, and the present invention is not limited thereto.

The fan 110 is a physical fan installed on the electronic device 100 and is used to dissipate heat from the electronic device 100. For example, the fan 110 includes rotatable blades. When the blades of the fan 110 rotate, the heat inside the electronic device 100 can be taken out to the outside of the electronic device 100. The fan 110 is, for example, a water-cooled fan or an air-cooled fan, and the present invention is not limited thereto. In addition, the fan speed of the fan 110 is controllable and has different heat dissipation capabilities. The higher the fan speed of the fan 110, the higher the heat dissipation capability provided by the fan 110. Conversely, the lower the fan speed of the fan 110, the lower the heat dissipation capability provided by the fan 110. In some embodiments, the speed of the fan 110 can be controlled by a pulse-width modulation signal (PWM signal).

The charging circuit 130 is, for example, a power control circuit such as a charger integrated circuit (Charger IC), which can be connected to the power adapter 200. In some embodiments, the charging circuit 130 can be connected to the power adapter 200 via the connection port 140. The power adapter 200 is used to receive power and supply power to the electronic device 100. For example, the power adapter 200 can receive AC power by connecting to a socket via a power cord, and provide the AC power to the electronic device 100 after converting the AC power into DC power. In addition, the charging circuit 130 can also be coupled to a battery (not shown). The charging circuit 130 can be used to determine whether to supply the power provided by the power adapter 200 or/and the battery to the electronic device 100.

In some embodiments, the charging circuit 130 can provide a system current to the system load of the electronic device 100. The system load can be, for example, a system circuit in the electronic device 100. For example, the system load of the electronic device 100 can include the control module 120, the fan 110, and other electronic components in FIG. 1. In some embodiments, the charging circuit 130 can detect the current value of the system current (hereinafter referred to as the system current value). In some embodiments, the system current can be the current provided by the power adapter 200, the current provided by the battery, or a combination thereof. In addition, the system current value can be obtained by measuring the voltage value at both ends of the resistor.

The control module 120 electrically couples the fan 110 and the charging circuit 130. The control module 120 can control the speed of the fan 110 and continuously receive the system current value reported by the charging circuit 130. The control module 120 can perform various operations in the fan speed adjustment method of the embodiment of the present invention.

Please refer to FIG. 2A, which is a schematic diagram of a control module according to an embodiment of the present invention. In some embodiments, the control module 120 may include an embedded controller 121, and the embedded controller 121 executes each operation in the fan speed adjustment method of the embodiment of the present invention. In detail, the embedded controller 121 may couple the fan 110 and the charging circuit 130, and control the fan speed of the fan 110 according to the system current value reported by the charging circuit 130. The embedded controller 121 may be implemented as a system-on-chip (SoC) with computing capabilities.

Please refer to FIG. 2B, which is a schematic diagram of a control module according to an embodiment of the present invention. In some embodiments, the control module 120 may include an embedded controller 121, a memory 122, and a processor 123. The embedded controller 121 may couple the fan 110 and the charging circuit 130, and the processor 123 couples the embedded controller 121 and the memory 122. The embedded controller 121 provides the system current value reported by the charging circuit 130 to the processor 123, and the processor 123 may control the fan speed of the fan 110 through the embedded controller 121. That is, the embedded controller 121 may adjust the fan speed of the fan 110 according to the control signal sent by the processor 123.

The memory 122 can be, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk or other similar devices, integrated circuits and combinations thereof. The processor 123 can be, for example, a central processing unit (CPU), an application processor (AP), or other programmable general-purpose or special-purpose microprocessor (microprocessor), a digital signal processor (DSP) or other similar devices, integrated circuits and combinations thereof. The processor 123 can access and execute the program code recorded in the memory 122 to implement the fan speed adjustment method in the embodiment of the present invention.

It should be noted that, regardless of the embodiment shown in FIG. 2A or FIG. 2B , the embedded controller 121 can be used to control the speed of the fan 110. Generally speaking, the memory 122 can record a fan speed table (also called a temperature control speed table). The fan speed table records the corresponding relationship between the temperature and the preset speed of the fan 110. The processor 123 can refer to the fan speed table according to the sensed temperature to determine the fan speed of the fan 110. The sensed temperature can be, for example, a sensed value generated by a temperature sensor (not shown) in the electronic device 100. For example, Table 1 is an example of a fan speed table.

嵌入式控制器121可利用處理器123透過表1所決定之風扇轉速表來控制風扇110的風扇轉速。例如，當感測溫度為落入T2至T3的溫度範圍內的話，嵌入式控制器121將風扇110的轉速調整至第二階轉速。當感測溫度為落入T4至T5的溫度範圍內的話，嵌入式控制器121將風扇110的轉速調整至第四階轉速。

The embedded controller 121 can use the processor 123 to control the fan speed of the fan 110 through the fan speed table determined in Table 1. For example, when the sensed temperature falls within the temperature range of T2 to T3, the embedded controller 121 adjusts the speed of the fan 110 to the second speed. When the sensed temperature falls within the temperature range of T4 to T5, the embedded controller 121 adjusts the speed of the fan 110 to the fourth speed.

FIG3 is a flow chart of a fan speed adjustment method according to an embodiment of the present invention, and the method flow of FIG3 can be implemented by various components of the electronic device 100 of FIG1. Please refer to FIG1 and FIG3 at the same time. The following is a description of the steps of the fan speed adjustment method of this embodiment in conjunction with various components of the electronic device 100 in FIG1.

In step S310, the control module 120 monitors a plurality of system current values corresponding to different time points. In detail, the charging circuit 130 may periodically (for example, every 1 second) report the system current value to the control module 120. Correspondingly, the control module 120 may continuously monitor a plurality of system current values provided by the charging circuit 130 to the system load at different time points. More specifically, the charging circuit 130 may detect a system current value at a time point T1, and report the system current value corresponding to the time point T1 to the control module 120. Thereafter, the charging circuit 130 may detect a next system current value at a next time point T2, and report the next system current value corresponding to the next time point T2 to the control module 120. For example, the charging circuit 130 may be connected to the embedded controller 121 via an I ² C interface, and the embedded controller 121 may read the system current value via the I ² C interface.

In step S320, the control module 120 can sequentially input multiple system current values into the machine learning model, and sequentially predict multiple load types through the machine learning model. Specifically, the system voltage provided by the power adapter 200 or the battery is generally set to a fixed value, so the size of the system current can be used to measure the level of the system load. In the embodiment of the present invention, the control module 120 can predict the load type of the system load based on the system current value and the machine learning model.

In some embodiments, the machine learning model may be a classification model. In some embodiments, the machine learning model may include a recurrent neural network (RNN) model. For example, the machine learning model may be a long short-term memory (LSTM) model or a Transformer model, and the present invention is not limited thereto. The control module 120 may sequentially input these system current values corresponding to different time points into the trained RNN model. In different embodiments, the model parameters (such as weight information, etc.) of the trained RNN model may be recorded in the memory 122.

In some embodiments, each load type sequentially predicted by the trained machine learning model may include a low load type or a high load type. In more detail, in some embodiments, when the control module 120 inputs the system current value corresponding to a certain time point into the RNN model, the RNN model may output the classification result (i.e., load type) of the system load corresponding to the time point, i.e., a low load type or a high load type. Then, when the control module 120 inputs the system current value corresponding to the next time point into the RNN model, the RNN model may output the classification result of the system load corresponding to the next time point. It can be seen that the control module 120 can use the machine learning model to output multiple load types in sequence and in a timely manner.

In step S330, the control module 120 determines whether multiple load types within a period of time meet a specific condition. Specifically, the control module 120 can determine whether these load types predicted at different time points within a period of time change from a low load type to a high load type, so as to further predict whether the system load is about to increase.

When multiple load types meet specific conditions (step S330 is judged as yes), in step S340, the control module 120 increases the fan speed of the fan 110 according to the multiple load types. In other words, when it is determined that multiple load types meet specific conditions, it means that these load types in a period of time are changed from low load types to high load types. Therefore, the control module 120 can increase the fan speed of the fan 110 in advance before the temperature rises to exceed the upper limit of a temperature range in the fan speed table, so as to increase the fan speed in advance and reduce the system temperature. Based on this, the electronic device 100 can have more temperature margin to achieve better system performance.

On the other hand, when multiple load types do not meet the specific conditions (step S330 is judged as no), in step S350, the control module 120 can maintain the fan speed of the fan 110 based on the sensed temperature. In other words, when the multiple load types predicted by the control module 120 using the machine learning model do not change from a low load type to a high load type, the control module 120 can control the fan speed of the fan 110 according to the fan speed table and the current sensed temperature.

In some embodiments, the low load type may include system idle load and continuous low load. Please refer to FIG. 4A, which is a schematic diagram of the system current of the system idle load according to an embodiment of the present invention. Please refer to FIG. 4B, which is a schematic diagram of the system current of the continuous low load according to an embodiment of the present invention. It should be noted that the embodiment of the present invention may define a current threshold value THc as half of the system current value when the electronic device 100 has the maximum system performance, but it is not limited to this.

In FIG. 4A , when the control module 120 determines that the load type is a system idle load, the curve L1 shows that the system current values corresponding to different time points (from the 1st second to the 20th second) in the past period of time (20 seconds) are all less than the current critical value THc, and the system current values at most time points in this period of time are less than one quarter of the current critical value THc. In FIG. 4B , when the control module 120 determines that the load type is a continuous low load, the curve L2 shows that the system current values corresponding to different time points (from the 1st second to the 20th second) in the past period of time (20 seconds) are all less than the current critical value THc, and the system current values at most time points in a period of time are maintained at more than one quarter of the current critical value THc.

In some embodiments, the high load type may include intermittent high load and continuous high load. Please refer to FIG. 4C, which is a schematic diagram of the system current of intermittent high load according to an embodiment of the present invention. Please refer to FIG. 4D, which is a schematic diagram of the system current of continuous high load according to an embodiment of the present invention.

In FIG. 4C , when the control module 120 determines that the load type is intermittent high load, curve L3 shows that the system current value at some time points in the past period of time (20 seconds) has exceeded the current critical value THc, but the system current value at other time points remains below the current critical value THc. In FIG. 4D , when the control module 120 determines that the load type is continuous high load, curve L4 shows that the system current values corresponding to different time points in the past period of time are all greater than the current critical value THc.

Based on the description of FIG. 4A to FIG. 4D , in some embodiments, the machine learning model can classify the system load into one of system idle load, continuous low load, intermittent high load, and continuous high load according to the input system current value. That is, the machine learning model can infer which of the four types of system load is based on the input system current value. However, the time length and time interval shown in FIG. 4A to FIG. 4D are only exemplary descriptions and are not used to limit the present invention.

Based on the examples of FIG. 4A to FIG. 4D, FIG. 5 is a flow chart of a fan speed adjustment method according to an embodiment of the present invention, and the method flow of FIG. 5 can be implemented by various components of the electronic device 100 of FIG. 1. Please refer to FIG. 1 and FIG. 5 at the same time. The following is a description of the steps of the fan speed adjustment method of this embodiment in conjunction with various components of the electronic device 100 in FIG. 1.

In step S510, the control module 120 monitors multiple system current values corresponding to different time points. In step S520, the control module 120 sequentially inputs the multiple system current values into the machine learning model, and sequentially predicts multiple load types through the machine learning model. The implementation details of steps S510 to S520 can refer to the contents of the aforementioned embodiments, and will not be elaborated here.

In step S530, the control module 120 determines whether multiple load types meet a specific condition. In this embodiment, step S530 can be implemented as step S531 to step S532.

In step S531, the control module 120 determines whether the M first load types corresponding to the M first time points among the multiple load types are low load types. M is a positive integer. For example, M can be equal to 4, but is not limited thereto. The control module 120 sequentially inputs the system current values of the M consecutive first time points into the machine learning model, so that the machine learning model sequentially predicts the M first load types corresponding to the M consecutive first time points.

When the M first load types corresponding to the M first time points are all low load types (step S531 is judged as yes), in step S532, the control module 120 determines whether the N second load types corresponding to the N second time points among the multiple load types are high load types. N is a positive integer. For example, N can be equal to 2, but is not limited to this. The control module 120 sequentially inputs the system current values of the N consecutive second time points into the machine learning model so that the machine learning model sequentially predicts the N second load types corresponding to the N consecutive second time points. It should be noted that the M first time points are earlier than the N second time points, and the M first time points and the N second time points are multiple continuous time points that are different from each other.

When multiple load types meet specific conditions (both step S531 and step S532 are judged as yes), in step S540, the control module 120 increases the fan speed of the fan 110 according to the multiple load types. More specifically, when the M first load types corresponding to the M first time points are low load types and the N second load types corresponding to the N second time points are high load types, in step S540, the control module 120 may increase the fan speed of the fan 110 according to the N second load types.

That is, through the judgment of step S531 and step S532, the control module 120 can judge whether the load type changes from the low load type to the high load type over time. In addition, when multiple load types do not meet the specific conditions (step S531 or step S532 is judged as no), the control module 120 determines the fan speed of the fan 110 based on the fan speed table and the sensed temperature, and returns to step S510.

It should be noted that in this embodiment, step S540 can be implemented as steps S541 to S543. In step S541, the control module 120 determines that at least one of the N second load types is an intermittent high load or a continuous high load. In some embodiments, the control module 120 can determine that the N second load types are intermittent high loads or continuous high loads. Alternatively, the control module 120 can determine that a second load type predicted at a second time closest to the current time among the N second load types is an intermittent high load or a continuous high load.

When at least one of the N second load types is an intermittent high load, in step S542, the control module 120 increases the fan 110 from the current speed to the first speed. When at least one of the N second load types is a continuous high load, in step S543, the control module 120 increases the fan 110 from the current speed to the second speed. Here, the second speed is higher than the first speed. Specifically, continuous high loads generate more heat energy than intermittent high loads, so continuous high loads require the fan 110 to provide stronger heat dissipation capabilities to discharge more heat energy outside the electronic device 100 compared to intermittent high loads.

Alternatively, in some embodiments, the control module 120 may determine whether the number of intermittent high loads among the N second load types is greater than the number of continuous high loads. When the number of intermittent high loads among the N second load types is greater than the number of continuous high loads, the control module 120 increases the fan 110 from the current speed to the first speed. Conversely, when the number of intermittent high loads among the N second load types is less than the number of continuous high loads, the control module 120 increases the fan 110 from the current speed to the second speed.

It should also be noted that, in some embodiments, the above-mentioned current speed, first speed, and second speed are recorded in the fan speed table of the embedded controller 121. Specifically, based on the current sensed temperature, the control module 120 can determine that the fan speed of the fan 110 is the nth speed in the fan speed table (i.e., the current speed). Thereafter, the control module 120 continuously predicts the load type based on the real-time system current value. When these load types meet specific conditions and at least one of the N second load types is an intermittent high load, the control module 120 can increase the fan speed of the fan 110 from the nth speed to the (n+1)th speed in the fan speed table. Alternatively, when these load types meet certain conditions and at least one of the N second load types is a continuous high load, the control module 120 can increase the fan speed of the fan 110 from the nth speed to the (n+2)th speed in the fan speed table. The (n+2)th speed is greater than the (n+1)th speed, and the (n+1)th speed is greater than the nth speed.

In summary, in the embodiment of the present invention, the system current value is continuously monitored and sequentially input into the machine learning model, so that the machine learning model continuously predicts multiple load types related to these input system current values. Therefore, when it is determined that these load types change from low load types to high load types, the fan speed of the fan can be increased. Based on this, before the system overload actually comes or the temperature is too high and affects the system performance, the electronic device and fan speed adjustment method of the embodiment of the present invention can pre-adjust the fan speed in advance according to the model prediction results. In this way, the fan speed can be increased in advance to reduce the system temperature, so that the electronic device has more temperature margin and strives for better system performance.

Although the present invention has been disclosed as above by the embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the relevant technical field can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the attached patent application.

S310~S350: Steps

Claims (8)

A fan speed adjustment method is applicable to an electronic device including a fan, the method comprising: monitoring a plurality of system current values corresponding to different time points; sequentially inputting the plurality of system current values into a machine learning model, and sequentially predicting a plurality of load types through the machine learning model; determining whether the plurality of load types meet a specific condition; and when the plurality of load types meet the specific condition, increasing the fan speed of the fan according to the plurality of load types, wherein each of the plurality of load types includes a low load type or a high load type, wherein the plurality of load types are The step of determining whether the multiple load types meet the specific condition includes: determining whether the M first load types corresponding to the M first time points among the multiple load types are the low load types, where M is a positive integer; and determining whether the N second load types corresponding to the N second time points among the multiple load types are the high load types, where N is a positive integer, wherein the M first time points are earlier than the N second time points, and the M first time points and the N second time points are multiple continuous time points that are different from each other. The fan speed adjustment method as described in claim 1, wherein the method further comprises: When the multiple load types do not meet the specific conditions, maintaining the fan speed of the fan. The fan speed adjustment method as described in claim 1, wherein when the multiple load types meet the specific conditions, the step of increasing the fan speed of the fan according to the multiple load types includes: when the M first load types corresponding to the M first time points are the low load types and the N second load types corresponding to the N second time points are the high load types, increasing the fan speed of the fan according to the N second load types. The fan speed adjustment method as described in claim 3, wherein the high load type includes an intermittent high load and a continuous high load, and the step of increasing the fan speed of the fan according to the N second load types includes: determining that at least one of the N second load types is the intermittent high load or the continuous high load; when at least one of the N second load types is the intermittent high load, increasing the fan speed from a current speed to a first speed; and when at least one of the N second load types is the continuous high load, increasing the fan speed from the current speed to a second speed, wherein the second speed is higher than the first speed. The fan speed adjustment method as described in claim 4, wherein the current speed, the first speed and the second speed are recorded in a fan speed table. The fan speed adjustment method as described in claim 1, wherein the machine learning model includes a recurrent neural network (RNN) model. In the fan speed adjustment method as described in claim 1, the step of detecting the multiple system current values corresponding to different time points includes: continuously monitoring the multiple system current values provided by a charging circuit to a system load at different time points. An electronic device includes: a fan; and a control module coupled to the fan and configured to: monitor a plurality of system current values corresponding to different time points; sequentially input the plurality of system current values into a machine learning model, and sequentially predict a plurality of load types through the machine learning model; determine whether the plurality of load types meet a specific condition; and when the plurality of load types meet the specific condition, increase the fan speed of the fan according to the plurality of load types, wherein each of the plurality of load types includes a low load type or a high load type. High load type, wherein the control module is further configured to: determine whether the M first load types corresponding to the M first time points in the multiple load types are the low load type, wherein M is a positive integer; and determine whether the N second load types corresponding to the N second time points in the multiple load types are the high load type, wherein N is a positive integer, wherein the M first time points are earlier than the N second time points, and the M first time points and the N second time points are multiple continuous time points different from each other.

Images