JP2009053211A - Creep error compensator for weight signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To more accurately compensate a creep error included in a weight signal.
A weight signal Wx (t) includes a creep error wx (t). The creep error wx (t) is a creep ratio K (θ) that changes with a temperature θ of the weight sensor and a time constant T. Each element of (θ) is included. Furthermore, the creep ratio K varies depending on the load W. Under this assumption, the creep error wx (t) is estimated based on the weight signal Wx (t). Based on the temperature θ of the weight sensor, the change in the creep ratio K (θ) and time constant T (θ) included in the creep error wx (t) due to the temperature θ is compensated. Further, the fluctuation rate R {Wx (t)} of the creep ratio K (θ) is reflected in the creep error wx (t). The creep error wx (t) is subtracted from the weight signal Wx (t), so that the creep error wx (t) is compensated.
[Selection] Figure 7

Description

  The present invention relates to a creep error compensation device, and more particularly to a creep error compensation device that is applied to a weighing device and compensates for an error due to a creep phenomenon of a weight signal corresponding to a load applied to a weight detection means.

  An example of this type of creep error compensator is disclosed in Patent Document 1. In Patent Document 1, when a load is applied to a strain generating body that constitutes a weight detection means (detector), the strain generating body is distorted in accordance with the load, and the strain amount is determined by a creep phenomenon. It is disclosed that it changes gradually with the passage of time, and specifically changes in a manner that approximates the step response of the first-order lag transfer function as shown in Equation 3a of Patent Document 1. In order to compensate for a change in the amount of distortion due to such a creep phenomenon, a so-called creep error, a compensation arithmetic circuit (20) is provided in the prior art disclosed in Patent Document 1. In this compensation calculation circuit, a transfer function showing an aspect that is essentially the reverse of the temporal change of the weight signal (measurement signal) generated according to the strain amount of the strain generating body is based on the equation 3a. Is set. By passing the weight signal through the compensation arithmetic circuit, the creep error is removed from the weight signal, that is, the creep error is compensated.

  The creep error eventually converges to a certain fixed value corresponding to the load, that is, the final creep value (final creep amount εcc). However, this final creep value varies depending on the temperature (θ) of the strain generating body. That is, when the temperature of the strain generating body changes, the final creep value also changes. The final creep value is expressed as a creep coefficient (β = εcc / ε (0); ε (0) is a strain amount of the strain generating body at the time when a load is applied). include. That is, when the temperature of the strain generating body changes, the aspect of the creep error change with time shown by the equation 3a also changes. For this reason, in the prior art, in order to compensate for the change in creep error due to the temperature change of the strain-generating body, the transfer function of the above-mentioned compensation arithmetic circuit, specifically this transfer, according to the temperature of the strain-generating body. A creep coefficient calculation circuit (11) for compensating for an element related to the creep coefficient in the function is also provided.

JP-A-4-12221

  By the way, in the prior art, the creep coefficient is constant regardless of the magnitude of the load. However, depending on the configuration of the weight detection means, the creep coefficient may change depending on the magnitude of the load. In other words, even if the temperature of the strain generating body is constant, the aspect of the creep error with time may change depending on the magnitude of the load. Therefore, the conventional technique cannot accurately compensate for the change in creep error due to the magnitude of the load. In other words, there is a problem that the creep error that changes depending on the magnitude of the load cannot be accurately compensated.

  Accordingly, an object of the present invention is to provide a creep error compensator that can realize even more accurate creep error compensation than before by accurately compensating for a change in creep error due to the magnitude of the load. To do.

  In order to achieve this object, the present invention relates to a creep error compensator for compensating a creep error of a weight signal corresponding to a load applied to a weight detecting means, and correlates with a time-dependent change state of the creep error with respect to the weight signal. Compensating means for performing a predetermined process including a process based on the transfer function to compensate for the creep error. Here, the transfer function includes a first coefficient that correlates with the final value of the creep error, that is, the final creep value. And this 1st coefficient changes with the magnitude of a load. The present invention further comprises weight correction means for correcting the transfer function in accordance with the weight signal.

  That is, in the present invention, when a load is applied to the weight detection means, a weight signal corresponding to the load is generated. This weight signal includes a creep error, that is, gradually changes with time. The compensation means compensates the creep error by performing a predetermined process on the weight signal. Note that the predetermined process includes a process based on a transfer function that correlates with an aspect of the creep error over time. However, depending on the configuration of the weight detection means, the aspect of the creep error with time changes depending on the magnitude of the load. In the present invention, the compensation means for weight corrects the transfer function of the compensation means according to the weight signal in order to compensate for the change in creep error due to the magnitude of the load. Specifically, the transfer function of the correction means includes a first coefficient that correlates with the final creep value, and this first coefficient varies depending on the magnitude of the load. The weight correcting means corrects the transfer function of the correcting means including the first coefficient in accordance with the weight signal.

  More specifically, the transfer function of the correction means includes a second coefficient that represents a variation rate of the first coefficient with respect to a change in load. The second coefficient is a function of the load. The weight correction means corrects the second coefficient according to the weight signal.

  As described above, according to the present invention, the change in the creep error due to the magnitude of the load is compensated according to the weight signal, that is, according to the load. Therefore, more accurate creep error compensation can be realized as compared with the above-described conventional technique that cannot compensate for the change in creep error due to the magnitude of the load.

  First, a reference example of the present invention will be described with reference to FIGS.

  As shown in FIG. 1, the weighing device 10 according to this reference example includes a weight sensor 12 as weight detection means. The weight sensor 12 is, for example, a strain gauge type load cell, which is not shown in the figure, but a metal strain generating body and a plurality of (for example, four) strainers attached to the strain generating body and constituting a bridge circuit. A strain gauge.

  When an object to be weighed (not shown) is placed on the weight sensor 12, and a load Wx is applied to the weight sensor 12, the weight sensor 12 (bridge circuit) outputs an analog signal of a voltage corresponding to the load Wx, In other words, a weight signal Wx (t) (t; time index) is output. The weight signal Wx (t) is amplified by the amplification circuit 14 at a predetermined amplification factor, and further, by a filtering process by the filter circuit (LPF) 16, electrical noise having a relatively high frequency, for example, several tens [Hz] or more is generated. After being removed, it is input to the A / D conversion circuit 18. The A / D conversion circuit 18 is, for example, of the Δ-Σ type, and the input weight signal Wx (t) is converted into a 20 [bit] digital signal (hereinafter, this digital signal also includes the weight signal Wx (t)). It is converted to The converted weight signal Wx (t) is input to a CPU (Central Processing Unit) 22 via an input / output circuit (I / O) 20.

The CPU 22 further performs digital filtering processing on the input weight signal Wx (t), thereby removing noise having a frequency lower than that described above, for example, noise caused by mechanical vibration. Then, the load Wx is determined based on the weight signal Wx (t) after the digital filtering process. Specifically, a data memory 24 is connected to the CPU 22, and the data memory 24 is used to convert a voltage measurement value of the weight signal Wx (t) into a load Wx value, that is, a measured value Wx ′. And a display formula (display program) for displaying the converted weight value Wx ′ on the liquid crystal display 26 are stored. The data memory 24 is, for example, a PROM (Programmable Read Only Memory) or an EEPROM (Electrically Erasable Programmable).
It is configured by a read-only memory (RAM) or a random access memory (RAM). The CPU 22 converts the weight signal Wx (t) into the measured value Wx ′ based on the conversion formula stored in the data memory 24, and further displays the measured value Wx ′ based on the display formula. Generate a signal. The generated measurement value display signal is input to the liquid crystal display (LCD) 26 via the input / output circuit 20, and thereby the measurement value Wx ′ of the load Wx is displayed on the liquid crystal display 26.

  The weighing device 10 includes a program memory 28 having a PROM configuration, for example, in addition to the data memory 24 described above, and a control program for controlling the operation of the CPU 22 is stored in the program memory 28. . The weighing device 10 also includes an operation key 30. When the operation mode 30 is selected by the operation key 30, the CPU 22 generates a measurement signal as described above, that is, executes measurement. Furthermore, the weighing device 10 includes a temperature sensor 32 for detecting the temperature θ of the weight sensor 12. The detection result (output signal) of the temperature θ by the temperature sensor 32 is input to the CPU 22 via the input / output circuit 20. The handling of this temperature θ will be described in detail later.

  By the way, the weight signal Wx (t) (strictly, the voltage of the weight signal Wx (t)) corresponds to the load Wx as described above, but when the load Wx is continuously applied, the weight signal Wx (T) gradually changes over time due to the creep phenomenon. Specifically, assuming that the weighing device 10 has already completed the zero point adjustment and the span adjustment, the weight signal Wx (t) is obtained when the load Wx is applied to the weight sensor 12 (t = 0). , Wx. That is, Wx (0) = Wx. When the load Wx is continuously applied, the weight signal Wx (t) gradually increases as time t elapses, for example, as shown in FIG. Such a property that the weight signal Wx (t) gradually increases is called a plus creep characteristic. The weight signal Wx (t) when the load Wx is applied is expressed by the following equation (1).

[Equation 1]
Wx (t) = Wx + wx (t) = Wx (0) + wx (t)

  Here, wx (t) represents a change in the weight signal Wx (t) with the passage of time t, that is, a creep error. The creep error wx (t) eventually converges to a certain value corresponding to the load Wx, that is, the final creep value wx (∞) when a sufficiently long time has passed. The time required for the creep error wx (t) to converge to the final creep value wx (∞) after the load Wx is applied to the weight sensor 12 varies depending on the structure of the weight sensor 12 and the like, but is approximately several tens. It is about minutes to several hours. This final creep value wx (∞) is expressed by the following formula 2.

[Equation 2]
wx (∞) = Wx (∞) −Wx (0)

  Here, Wx (∞) is the value (voltage value) of the weight signal Wx (t) when the creep error wx (t) converges to the final creep value wx (∞). This Wx (∞) is the time when the creep error wx (t) can be regarded as having converged to the final creep value wx (∞), in other words, several tens of minutes to several hours after the load Wx is applied. It can be substituted by the value of the weight signal Wx (t) at the time when it has passed (t = tens of minutes to several hours).

  When the load Wx is removed from the weight sensor 12 at a certain time t1, the weight signal Wx (t) decreases by the load Wx, indicating the creep error wx (t) at the time t1. That is, Wx (t1) = wx (t1). The weight signal Wx (t) gradually decreases with the elapse of time t and finally converges to zero.

  The weighing device 10 according to this reference example has a function of eliminating the influence of the creep error wx (t), that is, a creep error compensation function. In order to realize this, the following measures are taken in this reference example.

  That is, as described above, the final creep value wx (∞) corresponds to the load Wx, but from experience so far, for most strain gauge load cells, the ratio of the final creep value wx (∞) to the load Wx, In other words, it has been found that the creep ratio K (= wx (∞) / Wx; corresponding to the creep coefficient β in the prior art) as a first coefficient is constant regardless of the magnitude of the load Wx. That is, the following equation 3 holds.

[Equation 3]
K = wx (∞) / Wx = wx (∞) / Wx (0) = constant

  On the other hand, the mode (shape) of the change of the creep error wx (t) with the lapse of time t is that the step signal equivalent to the final creep value wx (∞) is a first-order lag transfer function (1 / (1 + T · s)). It is approximated to the output response waveform (step response waveform) of the control system when it is assumed that it is input to the represented control system. That is, the creep error wx (t) can be expressed by the following equation 4.

[Equation 4]
wx (t) = wx (∞) · {1-exp (−t / T)}

  Here, T is an inherent time constant of the weight sensor 12 (strain gauge load cell). Specifically, the creep error wx (t) after the load Wx is applied to the weight sensor 12 is the final creep value wx. This is the time taken to reach a value of 0.632 times (∞). Then, by substituting the modified equation of Equation 3 (the equation obtained for wx (∞)) into the final creep value wx (∞) in Equation 4, Equation 4 can be expressed as the following Equation 5. .

[Equation 5]
wx (t) = Wx (0) · K · {1-exp (−t / T)}

  Note that the creep error wx (t) is very small compared to the load Wx. For example, the final creep value wx (∞), which is the maximum value, is generally less than the number [%] of the load Wx and generally 1 [%]. Degree. That is, the difference between Wx (0) and the weight signal Wx (t) in Equation 5 is 1 [%] or less. Accordingly, by using Wx (t) instead of Wx (0) in Equation 5, Equation 5 can be expressed by the following approximate expression.

[Equation 6]
wx (t) ≈Wx (t) · K · {1-exp (−t / T)}

  According to Equation 6, if the weight signal Wx (t), the creep ratio K, and the time constant T are known at an arbitrary time point t when the load Wx is applied, the creep error wx at the arbitrary time point t is known. (T) can be estimated. Then, an accurate load Wx can be obtained by subtracting the estimated value (hereinafter referred to as a symbol wc (t)) from the weight signal Wx (t) at an arbitrary time point t. That is, the creep error wx (t) can be compensated.

  However, as described above, the final creep value wx (∞) changes depending on the temperature θ of the weight sensor 12, and the creep ratio K as a predetermined coefficient also changes. Specifically, the creep ratio K (strictly, its absolute value) increases as the temperature θ increases. Therefore, in this reference example, the creep ratio K is a function of the temperature θ of the weight sensor 12 and is specifically expressed by a quadratic expression shown in the following equation (7).

[Equation 7]
K (θ) = c1 · θ ² + c2 · θ + c3

  Here, c1, c2, and c3 are all constants, and are obtained as follows, for example. That is, the temperature θ of the weight sensor 12 is set stepwise to three different temperatures of θ1, θ2, and θ3. This setting can be realized, for example, by installing the weight sensor 12 in a constant temperature bath. The temperatures θ1, θ2, and θ3 are, for example, 0 [° C.], 20 [° C.], and 40 [° C.]. Under these temperature environments, an arbitrary load Wx, for example, a sample load Wm equivalent to the rated capacity of the weight sensor 12, is applied to the weight sensor 12, and the weight signal Wx (t) at that time is applied for a certain time ( For example, it is sequentially stored in the data memory 24 at intervals of a sampling period (several [ms]). Then, W (0) and Wx (∞) at each temperature θ1, θ2, and θ3 are extracted from the weight signal Wx (t) stored in the data memory 24, and these are substituted into the above-described formula 2. A final creep value wx (∞) at each of the temperatures θ1, θ2, and θ3 is obtained. Further, by substituting the final creep value wx (∞) and the load Wx into the above equation 3, the creep ratio K at each temperature θ1, θ2 and θ3, that is, K (θ1), K (θ2) and K ( θ3) is obtained. For these creep ratios (θ1), K (θ2), and K (θ3), the following quadratic expression similar to Equation 7 is constructed as shown in Equation 8.

[Equation 8]
K (θ1) = c1 · (θ1) ² + c2 · θ1 + c3
K (θ2) = c1 · (θ2) ² + c2 · θ2 + c3
K (θ3) = c1 · (θ3) ² + c2 · θ3 + c3

  Constants c1, c2, and c3 can be obtained by solving a ternary quadratic equation consisting of three quadratic expressions shown in Equation 8. The obtained constants c1, c2 and c3 are stored in the data memory 24.

  That is, at the time of actual weighing, that is, in the operation mode described above, by substituting these constants c1, c2 and c3 and the temperature θ of the weight sensor 12 at that time into the equation 7, The creep ratio K (θ) is obtained. In addition, you may obtain | require the said creep ratio K ((theta)) by other methods, such as the least square method.

  Further, if the aspect of the time-dependent change of the creep error wx (t) expressed by the above equation 6 is observed again, the change speed of the creep error wx (t) changes depending on the temperature θ of the weight sensor 12, that is, the creep error wx. It was found that the time constant T included in (t) changes. Specifically, similarly to the above, the temperature θ of the weight sensor 12 is set stepwise to three different temperatures of θ1, θ2 and θ3 (θ1 <θ2 <θ3), and an arbitrary load Wx is applied to the weight sensor 12. As shown in an exaggerated manner in FIG. 3, the time constant T increases as the temperature θ increases (a1 <a2 <a3). It was. As described above, the final creep value wx (∞) (creep ratio K) also increases as the temperature θ increases (b1 <b2 <b3). Therefore, in this reference example, the time constant T in the equation 6 is also expressed as a function of the temperature θ of the weight sensor 12 and specifically expressed by a quadratic equation as shown in the following equation 9.

[Equation 9]
T (θ) = d1 · θ ² + d2 · θ + d3

  Here, d1, d2 and d3 are all constants, and are obtained as follows in the same manner as the constants c1, c2 and c3 in Equation 7 above. That is, from the weight signal Wx (t) stored in the data memory 24 when obtaining the constants c1, c2, and c3, the time constants T, that is, T (θ1), T (θ2) at the respective temperatures θ1, θ2, and θ3. And T (θ3) are extracted. Then, for these time constants T (θ1), T (θ2), and T (θ3), as shown in Equation 10, a quadratic expression similar to Equation 9 is constructed.

[Equation 10]
T (θ1) = d1 · (θ1) ² + d2 · θ1 + d3
T (θ2) = d1 · (θ2) ² + d2 · θ2 + d3
T (θ3) = d1 · (θ3) ² + d2 · θ3 + d3

  Constants d1, d2 and d3 can be obtained by solving a ternary quadratic equation consisting of three quadratic expressions shown in Equation 10. The obtained constants d1, d2 and d3 are stored in the data memory 24.

  In the operation mode, if these constants d1, d2, and d3 and the temperature θ of the weight sensor 12 at that time are substituted into Equation 9, a time constant T (θ) corresponding to the temperature θ can be obtained. . The time constant T (θ) can also be obtained by other methods such as a least square method.

  That is, by substituting Equations 7 and 9 into Equation 6 described above, the creep error wx (t) at an arbitrary time t when the load Wx is applied can be estimated. That is, the estimated creep error wc (t) is expressed by the following formula 11.

[Equation 11]
wc (t) = Wx (t) · K (θ) · [1-exp {−t / T (θ)}]
≒ wx (t)

  Then, an accurate load can be obtained by subtracting the estimated creep error wc (t) obtained by Equation 11 from the weight signal Wx (t). That is, the measured value Wx (t) ′ of the load Wx is expressed by the following formula 12.

[Equation 12]
Wx (t) ′ = Wx (t) −wc (t) ≈Wx

  A control system for obtaining the measured value Wx (t) 'of the load Wx from the weight signal Wx (t) in this way is shown in FIG. That is, the control system as the compensation means includes a control element 50 as estimation means for estimating the creep error wx (t) from the weight signal Wx (t), and an estimated creep error wc estimated by the control element 50. And a control element 52 as subtraction means for subtracting (t) from the weight signal Wx (t). Further, the control element 50 is a control element 54 as a coefficient correction unit having a transfer function K (θ) and a time constant correction unit having a first-order lag transfer function 1 / (1 + T (θ) · s). The weight signal Wx (t) is continuously processed by these control elements 54 and 56 to be converted into an estimated creep error wc (t). Each control element 54 and 56 is provided with information indicating the temperature θ of the weight sensor 12 from the temperature sensor 32, and the transfer function of each control element 54 and 56 is corrected in accordance with this temperature θ. That is, temperature compensation is performed.

  Such a control system can be configured by, for example, hardware, but in the present reference example, a function similar to this is realized by the CPU 22. For this reason, the CPU 22 executes each process shown in the flowchart of FIG. 5 in the operation mode described above. Before executing the processing shown in this flowchart, the weighing device 12 is adjusted to zero and span in the adjustment mode. At the same time, in the adjustment mode, the constants c1, c2, and c3 in Equation 7 and the constants d1, d2, and c3 in Equation 8 are calculated and stored in the data memory 24. The selection of the adjustment mode and the operation mode is performed by the operation key 30. Further, the CPU 22 includes a flag F described later for temporarily recording whether or not the load Wx is applied in the operation mode. This flag F is reset once immediately before the operation mode is selected, that is, F = 0.

  Referring to FIG. 5, when the operation mode is selected, CPU 22 first determines in step S1 whether it is time to sample weight signal Wx (t). At the sampling timing, the process proceeds to step S3, and the weight signal Wx (t) is sampled. In step S5, it is determined whether or not the load Wx is applied to the weight sensor 12 from the weight signal Wx (t). This determination is made based on whether or not the weight signal Wx (t) exceeds a predetermined threshold value.

  When determining in step S5 that the load Wx has been applied, the CPU 22 proceeds to step S7 and determines whether or not “0” is set in the flag F described above. Here, when the flag F is set to “0”, that is, when the load Fx is not applied according to the flag F, the process proceeds to step S9, and the flag F is set to “1”. Set. In step S11, the count value of the counter for measuring the time t is once reset, and then the count operation by the counter is started. Further, in step S13, information regarding the temperature θ of the weight sensor 12 detected by the temperature sensor 32 is captured. If “0” is not set in the flag F in step S7 described above, that is, if the load wx is applied according to the flag F, step S9 and step S11 are skipped. Then, the process proceeds to step S13.

  After capturing the information about the temperature θ in step S13, the CPU 22 proceeds to step S15, and calculates the creep ratio K (θ) corresponding to the temperature θ based on the above-described equation 7. In step S17, the time constant T (θ) corresponding to the temperature θ is calculated based on the equation 9, and then the process proceeds to step S19, where the estimated creep error wc (t) is calculated from the above equation 11. Further, in step S21, the compensation weight signal Wx (t) 'is calculated from the above equation 12.

  Then, the CPU 22 proceeds to step S23, and determines how much load Wx the compensation weight signal Wx (t) 'indicates by using the conversion formula stored in the data memory 24 described above. In step S 25, the above-described measurement value display signal is generated based on the determination result Wx ′ in step S 23 and is input to the liquid crystal display 26 via the input / output circuit 20. As a result, the measured value Wx ′ of the load Wx is displayed on the liquid crystal display 26. After executing step S25, the CPU 22 returns to step S1.

  On the other hand, when determining in step S5 that the load Wx is not applied, the CPU 22 proceeds to step S27. In step S27, it is determined whether or not “1” is set in the flag F. Here, when the flag F is set to “1”, that is, when the load Wx is applied according to the flag F, the process proceeds to step S29, and the flag F is set to “0”. Set. In step S31, the display on the liquid crystal display 26 is cleared (erased), and then the process returns to step S1. If “1” is not set in the flag F in step S27, step S29 and step S31 are skipped and the process directly returns to step S1.

  Thus, according to the weighing device 10 according to the present reference example, the creep error wx (t) is compensated according to the temperature θ of the weight sensor 12, and this compensation target includes not only the creep ratio K but also the time constant. T is also included. Therefore, finer temperature compensation is performed as compared with the above-described conventional technique in which the time constant element is not compensated. Therefore, even more accurate creep error compensation than before can be realized. Such creep error compensation is particularly effective for a weighing device 10 that takes a relatively long time for measurement, such as a non-automatic balance.

  In this reference example, the control system of FIG. 5 is configured by software in terms of the CPU 22, but the control system may be configured by hardware as described above. A DSP (Digital Signal Processor) may be used.

  The temperature sensor 32 measures the temperature θ of the weight sensor 12 itself. Alternatively, the temperature of the environment (space) in which the weight sensor 12 is installed may be measured.

  Furthermore, instead of the weight signal Wx (t) in Equation 11 described above, the weight signal Wx (0) at the time (or immediately after) the load Wx is applied may be used. In order to realize this, for example, when a load Wx is applied, the weight signal Wx (0) at that time (immediately) is stored in the data memory 24. Then, the weight signal Wx (0) stored in the data memory 24 is substituted into Equation 11. In this way, the creep error wx (t) can be estimated more accurately, and thus more accurate creep error correction can be realized.

  Further, although the creep ratio K (θ) is calculated based on the above-described Equation 7, the present invention is not limited to this. For example, in the adjustment mode described above, the creep ratio K (θ) for each temperature θ is calculated in advance, and the calculation result is stored in the data memory 24 in the form of table data different from the above. In the operation mode, the creep ratio K (θ) corresponding to the temperature θ may be obtained by referring to the table data according to the temperature θ. The time constant T (θ) may be obtained in the same manner.

  With this reference example in mind, a specific embodiment of the present invention will now be described.

  That is, in the above-described reference example, it has been described that, for most strain gauge type load cells, the creep ratio K is constant regardless of the magnitude of the load Wx if the temperature θ is constant. This is particularly effective for the weighing device 10 that employs a strain gauge type load cell having different properties as the weight sensor 12. Referring to FIG. 6, in some strain gauge type load cells, even when temperature θ is constant, final creep value wx (∞) changes depending on the magnitude of load Wx, that is, creep ratio K (θ) changes. Sometimes. Some weight sensors 12 other than strain gauge type load cells have similar properties. Therefore, in the present embodiment, means for compensating for such a change in the creep ratio K (θ) due to the magnitude of the load Wx is taken.

  Specifically, first, it is grasped how much the creep ratio K (θ) changes depending on the magnitude of the load Wx. That is, the temperature θ of the weight sensor 12 is set to a certain constant temperature, for example, θ2 (= 20 [° C.]) described above. In this state, the three different loads Wx Wx1, Wx2 and Wx3 are individually applied to the weight sensor 12 with sufficient time intervals, in other words, with no creep error wx (t). Apply. Each of the loads Wx1, Wx2, and Wx3 has values of, for example, 1 (Wm), 1/2 (Wm / 2), and 1/4 (Wm / 4) of the sample load Wm described above. desirable. Then, the weight signal Wx (t) when these loads Wx1, Wx2, and Wx3 are applied is sequentially stored in the data memory at a constant time (for example, sampling period) interval. Then, Wx (0) and Wx (∞) for each of the loads Wx1, Wx2 and Wx3 are extracted from the stored weight signal Wx (t), and each of the loads Wx1, Wx2 and Wx3 is extracted from the extraction result. The final creep value wx (∞) for each of the Further, the creep ratios K (θ2, Wx1), K (θ2, Wx2) and K (θ2, Wx3) with respect to the loads Wx1, Wx2 and Wx3 are determined from the final creep values wx and the loads Wx1, Wx2 and Wx3. Ask for.

  Next, it is grasped from the creep ratios K (θ2, Wx1), K (θ2, Wx2) and K (θ2, Wx3) how much the creep ratio K (θ) varies as the load Wx changes. That is, one of the creep ratios K (θ2, Wx1), K (θ2, Wx2) and K (θ2, Wx3), for example, the creep ratio K (θ2, Wx1) with respect to the load Wx1 (= Wm), respectively Ratios of the creep ratios K (θ2, Wx1), K (θ2, Wx2), and K (θ2, Wx3), that is, fluctuation rates R (Wx1), R (Wx2), and R (Wx3) as second coefficients. And obtained based on the following equation (13).

[Equation 13]
R (Wx1) = K (θ2, Wx1) / K (θ2, Wx1) = 1
R (Wx2) = K (θ2, Wx2) / K (θ2, Wx1)
R (Wx3) = K (θ2, Wx3) / K (θ2, Wx1)

  Then, for each of these fluctuation rates R (Wx1), R (Wx2), and R (Wx3), a quadratic expression as shown in the following equation 14 is constructed using constants q1, q2, and q3.

[Formula 14]
R (Wx1) = q1 · (Wx1) ² + q2 · Wx1 + q3
R (Wx2) = q1 · (Wx2) ² + q2 · Wx2 + q3
R (Wx3) = q1 · (Wx3) ² + q2 · Wx3 + q3

  Further, a ternary quadratic equation composed of these three quadratic equations is solved to obtain the constants d1, d2 and d3. Then, using these constants d1, d2 and d3, a quadratic expression having an arbitrary load Wx as a variable is formed as shown in Equation 15.

[Equation 15]
R (Wx) = q1 · Wx ² + q2 · Wx + q3

  This equation 15 represents the fluctuation rate R (Wx) of the creep ratio K (θ) with respect to an arbitrary load Wx. Therefore, if the fluctuation rate R (Wx) is multiplied by the creep ratio K (θ), the change in the creep ratio K (θ) due to the difference in the load Wx can be compensated.

  As described above, the difference between the actual load Wx and the measured value represented by the weight signal Wx (t) is approximately 1 [%] or less. Accordingly, the weight signal Wx (t) can be used in place of the load Wx in Equation 15. That is, Expression 15 can be expressed by the following approximate expression of Expression 16.

[Equation 16]
R {Wx (t)} = q1 · {Wx (t)} ² + q2 · Wx (t) + q3

  That is, if the fluctuation rate R {Wx (t)} expressed by the equation 16 is multiplied by the above equation 11, the creep error wx (t) is accurately estimated without being affected by the magnitude of the load Wx. Can do. This can be expressed by the following equation (17).

[Equation 17]
wc (t) = Wx (t) · K (θ) · R {Wx (t)} · [1-exp {−t / T (θ)}]
≒ wx (t)

  As is apparent from Equation 17, if a control system as shown in FIG. 7 is configured, the load Wx can be accurately obtained regardless of the magnitude of the load Wx. In other words, the control system shown in FIG. 7 has a transfer function R {Wx (t)} between the output side of the control element 54 and the input side of the control element 56 in the control system of FIG. A control element 60 is provided as compensation means. The weight signal Wx (t) is input to the control element 60, and the transfer function R {Wx (t)} of the control element 60 is corrected according to the weight signal Wx (t), thereby the weight signal Wx. The change in the creep ratio K (θ) due to (t) is compensated.

  The control system shown in FIG. 7 can also be configured by hardware, for example, but in the present embodiment, the same function is realized by the CPU 22. For this reason, the CPU 22 executes each process shown in the flowchart of FIG. 8 in the operation mode described above. It should be noted that the constants q1, q2, and q3 described above are calculated in the adjustment mode and stored in the data memory 24 before executing the processing shown in this flowchart.

  As shown in FIG. 8, the flowchart in the present embodiment is the same as the flowchart in the reference example shown in FIG. 5, except that step S41 is newly provided between step S15 and step S17, and the estimated creep error wc ( Equation 17 is used instead of Equation 11 as a calculation formula for t). The other parts are the same as those in FIG. 5, and the description of these similar parts is omitted.

  That is, after calculating the creep ratio K (θ) in step S15, the CPU 22 proceeds to step S41 and calculates the fluctuation rate R {Wx (t)} based on the equation (16). In step S17, a time constant T (θ) is calculated, and in step S19, an estimated creep error wc (t) is calculated from the above-described equation 17. As a result, an accurate estimated creep error wc (t) that is not affected by the magnitude of the load Wx is calculated, and thus an accurate measured value is obtained.

  Thus, according to this embodiment, even if the creep ratio K (θ) changes depending on the magnitude of the load Wx, the change in the creep ratio K (θ) is compensated according to the load Wx. Therefore, even the weight sensor 12 affected by the magnitude of the load Wx can accurately compensate for the creep error wx (t).

  In the present embodiment, the fluctuation rate R {Wx (t)} is obtained based on the above-described equation 16, but instead of this, the weight signal at the time (or immediately after) the load Wx is applied. A variation rate R {Wx (0)} having Wx (0) as a variable may be used. Then, the estimated creep error wx (t) may be obtained by using the variation rate R {Wx (0)} instead of R {Wx (t)} in Equation 17. In this way, a more accurate estimated creep error wc (t) can be obtained, and thus more accurate creep error correction can be realized.

  Further, the variation rate R {Wx (t)} may be calculated based on preset table data instead of being calculated. That is, in the adjustment mode described above, the fluctuation rate R {Wx (t)} for each load Wx is calculated in advance, and the calculation result is stored in the data memory 24 in the form of table data. In the operation mode, the variation rate R {Wx (t)} suitable for the load Wx (weight signal Wx (t)) can be obtained by referring to the table data according to the weight signal Wx (t). Good.

  Next, another reference example of the present invention will be described.

  This another reference example eliminates the influence of creep errors caused by the mounting table when a mounting table (not shown) is attached to the weight sensor 12 in the configuration shown in FIG.

  That is, when there is a mounting table, if the weight of the mounting table is Wi, the weight signal Wx (t) is expressed by the following equation (18).

[Equation 18]
Wx (t) = Wx + wx (t) + Wi

  Here, the weight Wi of the mounting table includes a creep error caused by the mounting table, and it is considered that this creep error has already reached the final creep value. Therefore, in the adjustment mode described above, the weight Wi of the mounting table including the creep error is measured, and the measured value, that is, the initial load, is stored in the data memory 24, particularly PROM or EEPROM. Then, in the operation mode, by removing the initial load Wi from the weight signal Wx (t) as shown in Equation 19, only the signal component Wxa (t) of the load Wx and the creep error wx (t) due to the load Wx. Is calculated.

[Equation 19]
Wxa (t) = Wx (t) −Wi = Wx + wx (t)

  Then, the creep error wx (t) is estimated from the signal Wxa (t) calculated by Equation 19 in the manner described above. That is, the estimated creep error wc (t) is calculated by substituting Wxa (t) into Wx (t) in the equation 11 or the equation 17.

  This logic is represented in a diagram as shown in FIG. That is, FIG. 9 excludes a control element 70 having a control value corresponding to the initial load Wi and the control value Wi from the weight signal Wx (t) in the control system in the embodiment of the present invention shown in FIG. A control element 72 is added. Then, the signal Wxa (t) after the initial load Wi is eliminated by the control element 72 serving as exclusion means is input to the control elements 52, 54 and 60.

  In this other reference example, the control system shown in FIG. Specifically, the CPU 22 is caused to execute each process shown in the flowchart of FIG.

  As shown in FIG. 10, the flowchart in this other reference example is the same as the flowchart in the embodiment of the present invention shown in FIG. 8, except that step S51 is newly provided between step S3 and step S5, and in FIG. Step S53 is provided instead of step S41. The other parts are the same as those in FIG. 8, and thus the description of the same parts is omitted.

  That is, after sampling the weight signal Wx (t) in step S3, the CPU 22 proceeds to step S51 and eliminates the initial load Wi from the weight signal Wx (t) based on Equation 19, that is, the signal Wxa (t). Is calculated. In step S5, it is determined whether or not the load Wx is applied to the weight sensor 12 from the signal Wxa (t) after the initial load Wi is excluded. This determination is made based on whether or not the weight signal Wxa (t) exceeds a predetermined threshold value. Further, in step S53 after execution of step S15, a fluctuation rate R {Wxa (t)} with the signal Wxa (t) as a variable is calculated based on Equation 16.

  Thus, according to another reference example of the present invention, the initial load Wi is excluded from the weight signal Wx (t), and the creep error wx (t) is compensated based on the signal Wxa (t) after the exclusion. Therefore, the influence of the creep error due to the initial load Wi is eliminated. Therefore, accurate creep error compensation can be realized regardless of the presence of the initial load Wi.

  Further, since the initial load Wi is stored in the data memory 24 having a PROM or EEPROM configuration, that is, a non-volatile memory, the data of the initial load Wi is not erased even when the weighing device 10 is turned off. Therefore, when the power is turned on again, the creep error due to the initial load Wi can be eliminated without measuring the initial load Wi in the adjustment mode described above.

  In this reference example, as shown in FIG. 9, the control elements 70 and 72 are added to the control system of FIG. 7, but the present invention is not limited to this. For example, the control elements 70 and 72 may be added to the control system in FIG. 4 described above. In other words, step S51 may be provided between step S3 and step S5 of the flowchart shown in FIG. 5, and the CPU 22 may be operated according to this flowchart.

  In the above description, the case where the weight sensor 12 has a plus creep characteristic is illustrated, but the present invention is not limited to this. That is, some weight sensors 12 have a so-called negative creep characteristic in which the weight signal Wx (t) decreases with the passage of time t when a load Wx is applied, as shown in FIG. The present invention can also be applied to the weighing device 10 in which such a weight sensor 12 is employed.

It is a block diagram which shows schematic structure of the weighing | measuring apparatus which concerns on the reference example of this invention. It is a graph which shows a time-dependent change of the weight signal in the reference example. It is a graph which shows the difference of the creep error with respect to the difference of the temperature of the weight sensor in the reference example. It is a block diagram which shows notionally the structure of the control system in the reference example. It is a flowchart which shows operation | movement of CPU in the reference example. It is a graph for demonstrating the characteristic of one Embodiment of this invention. It is a block diagram which shows notionally the structure of the control system in the embodiment. It is a flowchart which shows operation | movement of CPU in the same embodiment. It is a block diagram which shows notionally the structure of the control system in another reference example of this invention. It is a flowchart which shows operation | movement of CPU in the said other reference example. It is a graph which shows a time-dependent change of the weight signal of the aspect different from FIG.

Explanation of symbols

10 Weighing device 22 CPU
32 Temperature sensor 50, 52, 54, 56 Control element

Claims (2)

In a creep error compensator for compensating for an error due to a creep phenomenon of a weight signal corresponding to a load applied to a weight detection means,
Compensating means for compensating the error by performing a predetermined process including a process based on a transfer function correlating with an aspect of the error over time with respect to the weight signal,
The transfer function includes a first coefficient that correlates with the final value of the error,
The first coefficient varies depending on the load,
Further comprising a weight correction means for correcting the transfer function according to the weight signal;
A creep error compensation device for a weight signal, characterized by: The transfer function includes a second coefficient representing a variation rate of the first coefficient with respect to a change in the load,
The second coefficient is a function of the load,
The weight correction means corrects the second coefficient according to the weight signal.
The weight signal creep error compensator according to claim 1.

Images