Current sensor fault diagnosis method and system, and storage medium

Abstract

The invention relates to a current sensor fault diagnosis method and system, and a storage medium. The method comprises the steps of collecting a three-phase current of a current sensor at a current moment; summing the three-phase current of the current sensor and taking an absolute value to obtain the sum of the three-phase current; performing exponential moving average calculation on the three-phase current sum to obtain an exponential moving average value of the three-phase current sum; performing approximate variance operation on the three-phase current sum and the exponential moving average value of the three-phase current sum to obtain an approximate variance value; performing exponential moving average calculation on the approximate variance value to obtain an approximate moving variance value; and judging whether the current sensor has a fault or not according to the approximate moving variance value. According to the method, potential faults can be found in time, the calculation efficiency and the prediction precision are improved, and the model robustness is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electronic devices, in particular to a current sensor fault diagnosis method and system thereof, and a storage medium. BACKGROUND

[0002] The fault of the current sensor belongs to a sudden change in the internal control system, which makes the processing and transmission process of the required feedback physical quantity and / or variable of the control system invalid or inaccurate, and further brings instability or even loss of control to the real-time operation of the entire closed-loop control system, and finally causes serious harm. Most of the fault diagnosis methods of the current sensor are to perform corresponding operations on the current value collected by the current sensor to obtain the required characteristic value, and then select whether to need to perform smoothing filtering processing. If the smoothing filtering processing is needed, the median filtering or simple moving average is mostly used, and then the characteristic value is directly compared with the set threshold to measure whether it is out of limit. If it is out of limit, the alarm flag is raised. Such fault diagnosis method has limitations in the aspects of characteristic value preprocessing and decision mechanism.

[0003] From the aspect of characteristic value preprocessing, the characteristic value after the mathematical operation of the three-phase current is subjected to the median filtering or simple moving average, which is limited by the selection of the data window and has limitations in performance. Moreover, the characteristics of the current sensor fault model are not considered, the gradual change of the long-term trend is ignored, and only the instantaneous state of the data in the window is relied on. The capture of the new trend needs to wait for the filling of the new window data, which has significant lag and is prone to form sudden interference at the tail of the window, resulting in "hard cut-off" of the historical data. Moreover, the median filtering needs to sort the data in the window and the simple moving average needs to maintain the window data, which has high computational complexity.

[0004] From the aspect of decision mechanism, the threshold size comparison is simply based on the mathematical model driven method, which does not quantify the statistical influence of noise and is not robust to occasional disturbances or steady-state noise. Moreover, due to the lack of sufficient statistics, the diagnosis relies on one-sided features and strong assumptions, which cannot adapt to the statistical nature of complex systems. SUMMARY

[0005] The purpose of the present application is to provide a current sensor fault diagnosis method and system thereof, and a storage medium, which can timely discover potential faults, improve computational efficiency and prediction accuracy, and enhance model robustness.

[0006] In the first aspect, the current sensor fault diagnosis method comprises the following steps: collecting the three-phase current of the current sensor at the current time; summing the three-phase current of the current sensor and taking the absolute value to obtain the three-phase current sum; performing an exponential moving average calculation on the three-phase current sum to obtain an exponential moving average value of the three-phase current sum; performing an approximate variance operation on the three-phase current sum and the exponential moving average value of the three-phase current sum to obtain an approximate variance value; performing an exponential moving average calculation on the approximate variance value to obtain an approximate moving variance value; judging whether the current sensor is faulty according to the approximate moving variance value.

[0007] Optionally, a calculation formula of the exponential moving average value of the three-phase current sum is: EMA_I sum (t) = a1*EMA_I sum (t-1) + (1-a1)*I sum (t); wherein, I sum (t) represents the three-phase current sum at a current time t; EMA_I sum (t) represents the exponential moving average value of the three-phase current sum at the current time t; EMA_I sum (t-1) represents the exponential moving average value of the three-phase current sum at a previous time t−1; and a1 represents a first forgetting factor.

[0008] Optionally, a calculation formula of the approximate variance value is: Pre_Var(t) = (EMA_I sum (t) - I sum (t)) / K; 2 wherein, Pre_Var(t) represents the approximate variance value at the current time t; EMA_I sum (t) represents the exponential moving average value of the three-phase current sum at the current time t; I sum (t) represents the three-phase current sum at the current time t; and K is a scaling coefficient implemented as a positive real number to ensure numerical stability of the approximate variance value.

[0009] Optionally, a calculation formula of the approximate moving variance value is: EMA_Var(t) = a2*EMA_Var(t-1) + (1-a2)*Pre_Var(t); wherein, EMA_Var(t) represents the approximate moving variance value at the current time t; a2 represents a second forgetting factor; EMA_Var(t-1) represents the approximate moving variance value at the previous time t−1; and Pre_Var(t) represents the approximate variance value at the current time t.

[0010] ​Optionally, the step of determining whether the current sensor is faulty according to the approximate moving variance value comprises: comparing the approximate moving variance value with a preset moving variance threshold value; if the approximate moving variance value is greater than the preset moving variance threshold value, starting an incremental counting mode and incrementing a counting variable by a sampling period; only after the incremental counting mode has been started, if the approximate moving variance value is less than or equal to the preset moving variance threshold value, decrementing the counting variable by the sampling period; comparing a current value of the counting variable with a preset counting alarm threshold value; if the current value of the counting variable is greater than the preset counting alarm threshold value, generating a current sensor fault indication; after generating the current sensor fault indication, if the current value of the counting variable is less than or equal to the preset counting alarm threshold value, canceling the current sensor fault indication.

[0011] Optionally, the operation of incrementing the counting variable is only performed when the current value of the counting variable is less than an upper limit value of the counter, and the operation of decrementing the counting variable is only performed when the current value of the counting variable is greater than zero.

[0012] Optionally, the method further comprises a fault maintaining mode: if the fault maintaining mode is enabled, once the current sensor fault indication is generated, the fault indication will be continuously maintained until a system reset signal is received.

[0013] Optionally, the step of determining whether the current sensor is faulty according to the approximate moving variance value comprises: comparing the approximate moving variance value with a preset moving variance threshold value; if the approximate moving variance value is greater than the preset moving variance threshold value, generating a current sensor fault indication; if the approximate moving variance value is less than or equal to the preset moving variance threshold value, canceling the current sensor fault indication.

[0014] In a second aspect, the current sensor fault diagnosis system of the present application is used to execute the steps of the current sensor fault diagnosis method described above, comprising: a data acquisition module for acquiring three-phase currents of a current sensor; a logic operation module for calculating a sum of the three-phase currents, an exponential moving average value of the sum of the three-phase currents, an approximate variance value and an approximate moving variance value; a determination module for determining whether the current sensor is faulty according to the approximate moving variance value; The data acquisition module and the judgment module are respectively connected to the logic operation module.

[0015] Thirdly, the storage medium of the present invention stores a computer program, which, when executed by a processor, implements the steps of the above-described current sensor fault diagnosis method.

[0016] The beneficial effects of this invention are: (1) Improve computational efficiency: Exponential moving average filtering does not require storing all past data points, but only the moving average of the previous time point and the current data point.

[0017] (2) Optimize storage space usage: Traditional direct calculation of moving variance requires saving all historical data within the window, which has high storage complexity. However, this invention uses two exponential moving average filters, which only needs to maintain the state of the previous moment, resulting in low storage complexity and making it suitable for resource-constrained real-time systems (such as embedded systems).

[0018] (3) High diagnostic sensitivity: The selected feature signal of the decision-making mechanism adopts the variance value. The variance value can more sensitively reflect the degree of dispersion of the sensor measurement value and can diagnose the sudden change of the current sensor. When the sensor performance begins to decline or a fault occurs, even if the average value does not change significantly, the variance may have already captured this small fluctuation, so as to discover potential faults in time and take corresponding maintenance measures. Attached Figure Description

[0019] Figure 1 This is a flowchart of the current sensor fault diagnosis method described in this embodiment of the invention; Figure 2 This is a schematic diagram of the current sensor fault diagnosis system described in an embodiment of the present invention.

[0020] Figure labeling: 1-Data acquisition module; 2-Logic operation module; 3-Judgment module. Detailed Implementation

[0021] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0022] See Figure 1 As shown in the figure, an embodiment of the present invention discloses a current sensor fault diagnosis method, the steps of which include: Collect the three-phase current from the current sensor at the current moment.

[0023] The sum of the three-phase currents from the current sensor is obtained by summing the absolute values.

[0024] The exponential moving average of the sum of the three-phase currents is calculated to obtain the exponential moving average of the sum of the three-phase currents.

[0025] The approximate variance is calculated by performing an approximate variance calculation on the sum of the three-phase currents and the exponential moving average of the sum of the three-phase currents.

[0026] The approximate moving variance is obtained by performing an exponential moving average on the approximate variance value.

[0027] Determine whether the current sensor is faulty based on the approximate moving variance value.

[0028] The fault diagnosis method for current sensors in this application uses the sum of the three-phase currents of the current sensor as the judgment criterion. Under normal circumstances, the instantaneous sum of the three-phase currents should be close to 0. If the current sensor degrades, ages, experiences environmental interference, or is affected by other fault factors, this sum of the three-phase currents will deviate significantly. This diagnostic method can identify this abnormal state in advance. The diagnostic method of this application considers the characteristics of the current sensor fault model, has high sensitivity, can promptly detect potential faults and take corresponding maintenance measures, and effectively balances real-time performance, historical dependence, and computational cost, improving computational efficiency and prediction accuracy, and also enhancing model robustness.

[0029] This application employs exponential moving average calculation, which better reflects the changing trend of data. The weighting of historical data in the exponential moving average calculation decreases exponentially, meaning that the influence of earlier data points on the current average decreases rapidly, with the magnitude of this decrease determined by the size of the forgetting factor. This application's exponential moving average calculation does not require storing all past data points; it only needs to store the moving average of the previous time point and the current data point. Therefore, it is more computationally efficient, consumes less memory, reduces computational load, improves prediction accuracy, and enhances model robustness. The exponential moving average calculation assigns higher weight to recent data, allowing variance estimation to not only reflect the current volatility intensity but also implicitly attenuate the influence of historical volatility, meeting the monitoring needs of non-steady-state signals (such as sudden noise and trend drift). Traditional direct calculation of moving variance requires storing all historical data within the window, resulting in high storage complexity. This application, by employing two exponential moving average calculations, only needs to maintain the state of the previous time point, resulting in low storage complexity. It is suitable for resource-constrained real-time systems (such as embedded systems) and optimizes storage space utilization.

[0030] The decision-making mechanism in this application uses approximate variance values ​​as the selected characteristic signals. This is based on the characteristics of current sensor fault models, because the early stages of current sensor faults often manifest as stability degradation, such as increased noise or intermittent failures, rather than instantaneous shifts. This degradation directly leads to an increase in variance, while the average value may temporarily remain normal. As the fault progresses, factors such as component aging and accumulated environmental interference cause systematic deviations in the mean value. This application uses approximate variance values. By monitoring the dynamic changes in the approximate moving variance of the sum of three-phase currents, it can effectively distinguish between normal background fluctuations and abnormal fluctuations caused by sensor faults. It can more sensitively reflect changes in the dispersion of sensor measurements, thereby promptly detecting potential faults and taking corresponding maintenance measures. This diagnostic method has high sensitivity and can diagnose sudden changes in current sensors. When sensor performance begins to decline or a fault occurs, even if the average value does not change significantly, the variance can capture these minute fluctuations.

[0031] In this embodiment of the application, the formula for calculating the exponential moving average of the sum of the three-phase currents is: EMA_I sum (t)=α1*EMA_I sum (t-1)+(1-α1)*I sum (t); I sum (t)=ABS(I a (t)+I b (t)+I c (t)); Among them, I sum (t) represents the total three-phase current at the current time t; EMA_I sum (t) represents the exponential moving average of the sum of the three-phase currents at the current time t; EMA_I sum (t-1) represents the exponential moving average of the sum of the three-phase currents at the previous time t−1; α1 represents the first forgetting factor, with a value ranging from (0,1). The first forgetting factor α1 characterizes the degree of weight decay and determines the actual value I of the current sample. sum (t) and the previous exponential moving average EMA_I sum (t-1) Calculate the current exponential moving average EMA_I sumThe relative importance at time (t). Due to the existence of the first forgetting factor α1, the weight of earlier data points in the calculation process gradually decreases over time. The selection of the first forgetting factor α1 is based on actual needs and the best practice value obtained after testing, and is not limited here. Changing the setting value of the first forgetting factor α1 can change the rate of weight decay of historical data. In the aforementioned formula, the larger the first forgetting factor α1, the greater the influence of historical data on the current average, that is, the slower the decay. a (t) represents the phase a current at time t, I b (t) represents the phase b current at time t, I c (t) represents the c-phase current at the current time t.

[0032] In this embodiment of the application, the formula for calculating the approximate variance is as follows: Pre_Var(t)=(EMA_I sum (t)-I sum (t)) 2 / K; Where Pre_Var(t) represents the approximate variance value at the current time t; I sum (t) represents the total three-phase current at the current time t; EMA_I sum (t) represents the exponential moving average of the sum of the three-phase currents at the current time t; K is the scaling factor, which is implemented as a positive real number to ensure the numerical stability of the approximate variance value, including preventing calculation anomalies caused by zero denominator or extreme values, or to ensure the reasonableness of the scaling range of the approximate variance value.

[0033] Specific implementations of K include, but are not limited to, the following: In one embodiment, K is implemented as follows: K=Max(g1,EMA_I sum (t)), where g1 and EMA_I are about to be used. sum (t) After comparison, the larger value is taken. In another embodiment, K can be implemented as a predetermined, constant positive constant. For example, K can be set to 1, in which case the formula simplifies to calculating the squared deviation itself; K can also be set to other positive real numbers, which can be selected by those skilled in the art according to the system's sensitivity requirements. It is worth noting that the above approximate variance calculation formula differs from the traditional formula for calculating sample variance or population variance, avoiding the problem of needing to store all historical data and sum them when calculating the mean and variance in the traditional way. The exponential moving average calculation updates the mean and variance through a recursive formula, requiring only the statistics (mean and variance) from the previous time step and the current data point each time.

[0034] In this embodiment of the application, the formula for calculating the approximate moving variance is: EMA_Var(t)=α2*EMA_Var(t-1)+(1-α2)*Pre_Var(t); Where EMA_Var(t) represents the approximate moving variance at the current time t; EMA_Var(t-1) represents the approximate moving variance at the previous time t−1; and α2 represents the second forgetting factor, with a value ranging from (0,1). The second forgetting factor α2 characterizes the degree of weight decay, determining the relative importance of the approximate variance Pre_Var(t) at the current time t and the approximate moving variance EMA_Var(t-1) at the previous time t−1 in calculating the approximate moving variance EMA_Var(t) at the current time t. Due to the existence of the second forgetting factor α2, the weight of earlier data points gradually decreases over time during the calculation process. The selection of the second forgetting factor α2 is based on actual needs and best practice values ​​obtained after testing, and is not limited here. Changing the setting value of the second forgetting factor α2 can change the rate of weight decay of historical data. In the aforementioned formula, the larger the second forgetting factor α2, the greater the influence of historical data on the current average, that is, the slower the decay. This approximate calculation of moving variance can, in some cases, be better adapted to the analysis or model under specific environments and objects.

[0035] Overall, the statistical processing methods in this approach can qualitatively characterize the degree of deviation between the sample and the mean, and are not limited by a fixed sample size.

[0036] In this embodiment of the application, determining whether a current sensor is faulty based on the approximate moving variance value includes: Compare the approximate moving variance value with the preset moving variance threshold; If the approximate moving variance value is greater than the preset moving variance threshold, the incrementing counting mode is activated, and a counting variable is incremented according to the sampling period. Only after the incremental counting mode has been activated, if the approximate moving variance value is less than or equal to the preset moving variance threshold, the counting variable will be decremented according to the sampling period. Compare the current value of the count variable with the preset count alarm threshold; If the current value of the counting variable is greater than the preset counting alarm threshold, a current sensor fault indication will be generated. After generating a current sensor fault indication, if the current value of the counting variable is less than or equal to the preset counting alarm threshold, the current sensor fault indication is cancelled.

[0037] The preset moving variance threshold is denoted as Var_LMT, which is the maximum upper limit of the pre-set approximate moving variance, reflecting the acceptable fluctuation range. It is compared with the approximate moving variance value EMA_Var(t). The count variable is denoted as CNT_SEN, used to count the sampling period. The preset count alarm threshold is denoted as CNT_ALM_LMT. For example, if CNT_ALM_LMT = 20, assuming a sampling period of 0.0001s (PWM switching frequency of 10kHz), when EMA_Var(t) > Var_LMT, and the count variable CNT_SENS starts to increase beyond 20 sampling periods (i.e., 0.002s), a fault indication will be generated. Unless the count is decremented to make the count variable CNT_SEN less than or equal to 20, the fault indication will persist.

[0038] In this application example, the increment operation of the count variable is performed only when the current value of the count variable is less than the upper limit value of the counter; the decrement operation of the count variable is performed only when the current value of the count variable is greater than zero. The upper limit value of the counter is denoted as CNT_MAX_LMT. Different trigger conditions need to be defined for the increment and decrement modes to enable their operation. The enabling trigger condition for the increment mode is that the count value of the count variable CNT_SENS in the current sampling period is less than the upper limit value CNT_MAX_LMT, which is typically set to be greater than the counting alarm threshold CNT_ALM_LMT. The enabling trigger condition for the decrement mode is that the count value of the count variable CNT_SENS in the sampling period is greater than 0.

[0039] In this application example, a fault holding mode is also included: if the fault holding mode is enabled, once a current sensor fault indication is generated, the fault indication will be maintained until a system reset signal is received. This mode can be implemented by adding a fault holding module (Hold function) at the back end of the sensor fault indication generation module. If the fault holding mode is enabled, once a current sensor fault indication is generated, the fault indication will be maintained until a system reset signal is received. The fault holding function is an optional extension function based on the basic counting mechanism, used to "lock" the fault state.

[0040] The decision rules for statistical results in this application require that a fault warning be triggered only when the approximate moving variance value EMA_Var(t) exceeds the preset moving variance threshold Var_LMT multiple times consecutively, rather than a single accidental exceedance. It must simultaneously meet the dual conditions of consecutive event occurrence and exceeding the event occurrence limit, thus reducing the false alarm rate. For example, a sudden change in variance might be noise; multiple consecutive exceedances are more likely to reflect a true anomaly and enhance sensitivity to systematic drift or gradual changes.

[0041] In this embodiment of the application, determining whether a current sensor is faulty based on the approximate moving variance value includes: Compare the approximate moving variance value with the preset moving variance threshold; If the approximate moving variance value is greater than the preset moving variance threshold, a current sensor fault indication will be generated. If the approximate moving variance value is less than or equal to the preset moving variance threshold, the current sensor fault indication will be canceled.

[0042] The preset moving variance threshold is denoted as Var_LMT. The approximate moving variance value EMA_Var(t) is compared with Var_LMT. If EMA_Var(t) > Var_LMT, a fault indication is generated. This decision rule is simpler and more direct, without using complex counting mechanisms, making it suitable for diagnosing and handling immediate faults. If the preset moving variance threshold is exceeded, an early warning is issued immediately. This rule is optional; the choice between this simplified rule and a rule incorporating a counting mechanism can be made based on the specific circumstances.

[0043] In this embodiment, the current sensor fault indication can be generated by displaying a current sensor fault warning sign, thereby enabling timely maintenance. In some embodiments, the current sensor fault indication also includes voice prompts and text prompts. At least one of these settings can be selected according to the actual situation, and no limitation is made here.

[0044] See Figure 2 As shown in the figure, this invention also discloses a current sensor fault diagnosis system for performing the steps of the above-described current sensor fault diagnosis method. The system includes a data acquisition module 1, a logic operation module 2, and a judgment module 3, with the data acquisition module 1 and judgment module 3 respectively connected to the logic operation module 2. The data acquisition module 1 is used to acquire the three-phase currents of the current sensor. The logic operation module 2 is used to calculate the sum of the three-phase currents, the exponential moving average of the sum of the three-phase currents, the approximate variance, and the approximate moving variance. The judgment module 3 is used to determine whether the current sensor has malfunctioned based on the approximate moving variance.

[0045] This invention also discloses a storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described current sensor fault diagnosis method.

[0046] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for diagnosing faults in a current sensor, characterized in that, Includes the following steps: Collect the three-phase current from the current sensor at the current moment; The sum of the three-phase currents of the current sensor is obtained by summing the three-phase currents and taking their absolute values. The exponential moving average of the sum of the three-phase currents is calculated to obtain the exponential moving average of the sum of the three-phase currents; An approximate variance value is obtained by performing an approximate variance calculation on the sum of the three-phase currents and the exponential moving average of the sum of the three-phase currents; The approximate moving variance value is obtained by performing an exponential moving average on the approximate variance value. The current sensor is determined to be faulty based on the approximate moving variance value.

2. The current sensor fault diagnosis method according to claim 1, characterized in that, The formula for calculating the exponential moving average of the sum of the three-phase currents is: EMA_I sum (t)=α1*EMA_I sum (t-1)+(1-α1)*I sum (t); Among them, I sum (t) represents the total three-phase current at the current time t; EMA_I sum (t) represents the exponential moving average of the sum of the three-phase currents at the current time t; EMA_I sum (t-1) represents the exponential moving average of the sum of the three-phase currents at the previous time t−1; α1 represents the first forgetting factor.

3. The current sensor fault diagnosis method according to claim 1, characterized in that, The formula for calculating the approximate variance is as follows: Pre_Var(t)=(EMA_I sum (t)-I sum (t)) 2 / K; Where Pre_Var(t) represents the approximate variance value at the current time t; EMA_I sum (t) represents the exponential moving average of the sum of the three-phase currents at the current time t; I sum (t) represents the sum of the three-phase currents at the current time t; K is the scaling factor, which is implemented as a positive real number to ensure the numerical stability of the approximate variance value.

4. The current sensor fault diagnosis method according to claim 1, characterized in that, The formula for calculating the approximate moving variance is as follows: EMA_Var(t)=α2*EMA_Var(t-1)+(1-α2)*Pre_Var(t); Where EMA_Var(t) represents the approximate moving variance at the current time t; α2 represents the second forgetting factor; EMA_Var(t-1) represents the approximate moving variance at the previous time t−1; and Pre_Var(t) represents the approximate variance at the current time t.

5. The current sensor fault diagnosis method according to claim 1, characterized in that, Determining whether the current sensor is faulty based on the approximate moving variance value includes: The approximate moving variance value is compared with a preset moving variance threshold. If the approximate moving variance value is greater than the preset moving variance threshold, then the incrementing counting mode is started, and a counting variable is incremented according to the sampling period; Only after the incremental counting mode has been activated, if the approximate moving variance value is less than or equal to the preset moving variance threshold, the counting variable is decremented according to the sampling period. The current value of the count variable is compared with a preset count alarm threshold; If the current value of the counting variable is greater than the preset counting alarm threshold, a current sensor fault indication is generated; After generating a current sensor fault indication, if the current value of the counting variable is less than or equal to the preset counting alarm threshold, the current sensor fault indication is revoked.

6. The current sensor fault diagnosis method according to claim 5, characterized in that, include: The increment operation on the count variable is performed only when the current value of the count variable is less than the upper limit value of the counter; The operation of decrementing the count variable is performed only when the current value of the count variable is greater than zero.

7. The current sensor fault diagnosis method according to claim 5, characterized in that, Also includes fault-tolerant mode: If the fault holding mode is enabled, once the current sensor fault indication is generated, the fault indication will be maintained until a system reset signal is received.

8. The current sensor fault diagnosis method according to claim 1, characterized in that, The step of determining whether the current sensor is faulty based on the approximate moving variance value includes: The approximate moving variance value is compared with a preset moving variance threshold. If the approximate moving variance value is greater than the preset moving variance threshold, a current sensor fault indication is generated; if the approximate moving variance value is less than or equal to the preset moving variance threshold, the current sensor fault indication is canceled.

9. A current sensor fault diagnosis system, used to perform the steps of the current sensor fault diagnosis method as described in any one of claims 1 to 8, characterized in that, include: The data acquisition module (1) is used to acquire the three-phase current of the current sensor; The logic operation module (2) is used to calculate the sum of the three-phase currents, the exponential moving average of the sum of the three-phase currents, the approximate variance value, and the approximate moving variance value; The judgment module (3) is used to determine whether the current sensor is faulty based on the approximate moving variance value. The data acquisition module (1) and the judgment module (3) are respectively connected to the logic operation module (2).

10. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the current sensor fault diagnosis method as described in any one of claims 1 to 8.

Images