Sensor-based Multiplexing Circuit Overvoltage Fault Identification System

Abstract

The present invention relates to the technical field of data processing, and particularly to a sensor-based overvoltage fault identification system for a multiplexing circuit; obtaining a temperature influence coefficient of a preset temperature according to the difference characteristics between the monitored voltage and the actual voltage; obtaining a relative dispersion degree according to the distribution dispersion characteristics of the preset temperature and the temperature influence coefficient; adjusting a preset default gamma value according to the relative dispersion degree to obtain an initial gamma value and a corresponding SVM model; obtaining a model error degree according to the prediction error characteristics of the temperature influence coefficient; adjusting the initial gamma value to obtain an iterative gamma value. The present invention performs iterative adjustment according to the model error degrees of the iterative gamma value and the initial gamma value to obtain a final gamma value; obtaining a predicted temperature influence coefficient according to the SVM model of the final gamma value; obtaining a corrected monitored voltage according to the predicted temperature influence coefficient; improving the efficiency of overvoltage monitoring and predicting and fitting the temperature influence coefficient.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to an overvoltage fault identification system for a multiplexing circuit based on a sensor. Background Art

[0002] Overvoltage fault is a common problem affecting the operation of equipment. Monitoring overvoltage faults in a circuit is beneficial to protecting the stable operation of the equipment. In order to improve the monitoring efficiency and reliability, the multiplexing circuit technology based on sensors has been gradually applied to overvoltage fault detection. However, the resistance characteristics of electronic components in the monitoring circuit change with temperature, resulting in a deviation between the monitored voltage and the actual operating voltage of the equipment, and causing the monitored voltage values at different temperatures to show non-linear changes.

[0003] In order to improve the monitoring accuracy, it is necessary to determine the degree of voltage deviation caused by different temperatures. Since the temperature has a non-linear effect on the resistance characteristics, it is necessary to fit to obtain the temperature influence coefficient at different temperatures. Currently, SVM (Support Vector Machine) is usually used for data prediction and fitting. The hyperparameter gamma value in this algorithm is closely related to the prediction accuracy. However, due to the large selection range of the gamma value, the traversal process of the gamma value is relatively long, and it takes a long time to select a gamma value with higher prediction fitting accuracy, which affects the efficiency of calculating the temperature influence coefficient and overvoltage monitoring. Summary of the Invention

[0004] In order to solve the technical problem that the selection time of the hyperparameter gamma value in the above-mentioned SVM is relatively long, which affects the efficiency of calculating the temperature influence coefficient and overvoltage monitoring, the purpose of the present invention is to provide an overvoltage fault identification system for a multiplexing circuit based on a sensor, and the specific technical solution adopted is as follows:

[0005] A data acquisition module, configured to acquire the monitored voltage of the actual voltage at different preset temperatures in an experiment;

[0006] A temperature influence analysis module, configured to obtain the temperature influence coefficient of a preset temperature according to the preset temperature and the difference characteristics between the monitored voltage and the actual voltage at the preset temperature; obtain the temperature discrete characteristic value according to the distribution characteristics of the preset temperature; obtain the influence coefficient discrete characteristic value according to the distribution characteristics of the temperature influence coefficient; and obtain the relative discrete degree according to the difference characteristics between the temperature discrete characteristic value and the influence coefficient discrete characteristic value;

[0007] A model construction module, which is used to adjust the preset default gamma value of the SVM (Support Vector Machine) according to the relative discreteness degree to obtain an initial gamma value; construct an SVM model of a preset temperature and a temperature influence coefficient according to the initial gamma value; obtain the model error degree of the initial gamma value according to the prediction error characteristics of the temperature influence coefficient in the SVM model; perform an increase or decrease adjustment of a preset multiple on the initial gamma value to obtain an iterative gamma value; continue to perform an increase or decrease adjustment on the iterative gamma value according to the difference characteristics between the iterative gamma value and the model error degree of the initial gamma value to obtain a final gamma value;

[0008] A voltage monitoring module, which is used to obtain a predicted temperature influence coefficient at any temperature according to the SVM model of the final gamma value; obtain a corrected monitored voltage at the any temperature according to the predicted temperature influence coefficient.

[0009] Further, the step of obtaining the temperature influence coefficient of the preset temperature according to the preset temperature and the difference characteristics between the monitored voltage and the actual voltage at the preset temperature includes:

[0010] Calculate the difference between any preset temperature and the preset normal temperature to obtain a temperature deviation value; calculate the difference between the monitored voltage and the actual voltage at the any preset temperature to obtain a voltage deviation value; calculate the ratio of the voltage deviation value to the temperature deviation value to obtain the temperature influence coefficient of the any preset temperature.

[0011] Further, the step of obtaining the temperature discrete characteristic value according to the distribution characteristics of the preset temperature includes:

[0012] Perform a standardization process on the preset temperature to obtain a preset temperature standard value; calculate the average value of the absolute values of the differences between any two preset temperature standard values among all the preset temperature standard values to obtain a temperature average difference value; calculate the product of the standard deviation of the preset temperature standard value and the temperature average difference value to obtain the temperature discrete characteristic value.

[0013] Further, the step of obtaining the influence coefficient discrete characteristic value according to the distribution characteristics of the temperature influence coefficient includes:

[0014] Perform a standardization process on the temperature influence coefficient to obtain an influence coefficient standard value; calculate the average value of the absolute values of the differences between any two influence coefficient standard values among all the influence coefficient standard values to obtain an influence coefficient average difference value; calculate the product of the standard deviation of the influence coefficient standard value and the influence coefficient average difference value to obtain the influence coefficient discrete characteristic value.

[0015] Further, the step of obtaining the relative discreteness degree according to the difference characteristics between the temperature discrete characteristic value and the influence coefficient discrete characteristic value includes:

[0016] Calculate the ratio of the discrete eigenvalue of the influence coefficient to the discrete eigenvalue of the temperature to obtain the relative discreteness degree.

[0017] Further, the step of obtaining the initial gamma value by adjusting the preset default gamma value of the SVM support vector machine according to the relative discreteness degree includes:

[0018] Calculate the product of the reciprocal of the relative discreteness degree and the preset default gamma value to obtain the initial gamma value.

[0019] Further, the step of obtaining the model error degree of the initial gamma value according to the prediction error characteristics of the temperature influence coefficient in the SVM model includes:

[0020] Calculate the difference between the temperature influence coefficient and the predicted value of the SVM model at the same preset temperature to obtain the prediction difference; calculate the average value of the absolute values of all prediction differences to obtain the average error; calculate the product of the absolute value of the skewness of the prediction difference and the average error to obtain the model error degree of the initial gamma value.

[0021] Further, the step of obtaining the iterative gamma value by performing an increase or decrease adjustment of a preset multiple on the initial gamma value includes:

[0022] The preset multiple includes a preset first multiple and a preset second multiple. The preset first multiple is less than the constant 1, and the preset second multiple is greater than the constant 1; calculate the product of the initial gamma value and the preset first multiple to obtain the iterative gamma value ; calculate the product of the initial gamma value and the preset second multiple to obtain the iterative gamma value . Further, the step of obtaining the final gamma value by continuing to perform an increase or decrease adjustment on the iterative gamma value according to the difference characteristics of the model error degree between the iterative gamma value and the initial gamma value includes:

[0023] When the difference between the model error degrees of the initial gamma value and the iterative gamma value exceeds the constant 0, calculate the product of the iterative gamma value and the preset first multiple to obtain a new iterative gamma value , compare the model error degrees of the iterative gamma value and the iterative gamma value and continue the iteration; when the difference between the model error degrees of the initial gamma value and the iterative gamma value exceeds the constant 0, calculate the product of the iterative gamma value and the preset second multiple to obtain a new iterative gamma value , compare the model error degrees of the iterative gamma value and the iterative gamma value and continue the iteration;

[0024] If the model error degree of the latest round of iterative gamma value is greater than that of the previous round of iterative gamma value, the average value of the latest round of iterative gamma value and the previous round of iterative gamma value is used as the final gamma value.

[0025] Further, the step of obtaining the corrected monitoring voltage of any temperature according to the predicted temperature influence coefficient includes:

[0026] Calculate the difference between any temperature and the preset normal temperature to obtain the actual temperature difference; calculate the product of the actual temperature difference and the preset temperature influence coefficient to obtain the voltage difference amount; calculate the difference between the monitoring voltage at any temperature and the voltage difference amount to obtain the corrected monitoring voltage of any temperature.

[0027] The present invention has the following beneficial effects:

[0028] In the present invention, since temperature has a non-linear influence on the resistance characteristics in the monitoring voltage device, obtaining the temperature influence coefficient of the preset temperature can determine the influence degree of temperature on voltage measurement at some temperatures during the experiment, which is convenient for predicting and fitting the temperature influence coefficients at different temperatures later and improving the accuracy of overvoltage monitoring. Obtaining the temperature discrete characteristic value can characterize the distribution discrete degree of the preset temperature, and obtaining the influence coefficient discrete characteristic value can characterize the discrete degree of the temperature influence coefficient; furthermore, obtaining the relative discrete degree can accurately characterize the distribution discrete degree of the temperature influence coefficient without the influence of temperature distribution. Since the selection of the gamma value in the SVM model is related to the discrete degree of the samples, obtaining the initial gamma value according to the relative discrete degree can quickly determine a more appropriate gamma hyperparameter, improving the construction efficiency of the SVM model without reducing the prediction and fitting accuracy. Obtaining the model error degree can evaluate the prediction accuracy of the model, which is convenient for selecting an iterative gamma value with better prediction effect based on the initial gamma value. Obtaining the final gamma value can improve the construction efficiency of the SVM model while maintaining good model prediction accuracy, and further improve the efficiency of predicting and fitting the temperature influence coefficient and overvoltage monitoring. Finally, obtaining the corrected monitoring voltage can more accurately monitor the overvoltage situation of the device. Description of the Drawings

[0029] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0030] Figure 1 It is a block diagram of an overvoltage fault identification system for a multiplexing circuit based on a sensor provided by an embodiment of the present invention. Detailed Implementation Manner

[0031] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation manner, structure, features, and effects of a sensor-based multiplexing circuit overvoltage fault identification system proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0033] The following specifically describes the specific solution of a sensor-based multiplexing circuit overvoltage fault identification system provided by the present invention with reference to the accompanying drawings.

[0034] Please refer to Figure 1 , which shows a block diagram of a sensor-based multiplexing circuit overvoltage fault identification system provided by an embodiment of the present invention. The system includes the following modules:

[0035] A data acquisition module S1, configured to acquire the monitored voltage of the actual voltage in the experiment at different preset temperatures.

[0036] Since the operating temperature of the device or the ambient temperature will cause changes in the resistance characteristics in the overvoltage monitoring device, resulting in a difference between the monitored voltage of the overvoltage monitoring device and the actual voltage of the device; in order to improve the accuracy of overvoltage fault identification of the multiplexing circuit, it is necessary to calculate the deviation value between the monitored voltage and the actual voltage at different temperatures, so as to correct the monitored voltage. Since the temperature range is relatively wide, it is difficult to obtain the deviation values at all temperatures through experiments. Therefore, the influence degree of partial temperatures on the monitored voltage can be obtained, and the influence degree at all temperatures can be predicted and fitted. First, acquire the monitored voltage of the actual voltage in the experiment at different preset temperatures, measure the temperature at the overvoltage monitoring device through a temperature sensor, and combine the temperature signal and the voltage signal monitored by the overvoltage monitoring device onto a signal channel through multiplexing circuit technology and transmit it to the analysis module to improve the data transmission efficiency and reliability. In the embodiment of the present invention, a constant voltage is used to input a stable voltage value into the current circuit system as the actual voltage, and the monitored voltage of the overvoltage monitoring device is read at 30 different preset temperatures; thus, the monitored voltages of the actual voltage at 30 different temperatures are obtained; the implementer can determine the preset temperature and the actual voltage according to the implementation scenario.

[0037] The temperature influence analysis module S2 is used to obtain the temperature influence coefficient of a preset temperature according to the preset temperature and the difference characteristics between the monitored voltage and the actual voltage at the preset temperature; obtain the temperature discrete characteristic value according to the distribution characteristics of the preset temperature; obtain the influence coefficient discrete characteristic value according to the distribution characteristics of the temperature influence coefficient; and obtain the relative discrete degree according to the difference characteristics between the temperature discrete characteristic value and the influence coefficient discrete characteristic value.

[0038] Since the resistance characteristics in the overvoltage monitoring device show non-linear changes due to temperature, after obtaining the monitored voltages at different preset temperatures, the difference between the monitored voltage and the actual voltage can be analyzed to determine the temperature influence coefficient. Therefore, the temperature influence coefficient of the preset temperature is obtained according to the preset temperature and the difference characteristics between the monitored voltage and the actual voltage at the preset temperature; preferably, in the embodiment of the present invention, the steps of obtaining the temperature influence coefficient of the preset temperature include: calculating the difference between any preset temperature and the preset normal temperature to obtain the temperature deviation value; the preset normal temperature is the temperature at which the monitored voltage is the same as the actual voltage when the resistance characteristics are normal; the temperature deviation value characterizes the difference between the preset temperature and the preset normal temperature, and the greater the temperature deviation value, the greater the influence of temperature on the resistance characteristics. Calculate the difference between the monitored voltage and the actual voltage at any preset temperature to obtain the voltage deviation value; the greater the voltage deviation value, the greater the influence of the any preset temperature on the accuracy of voltage monitoring. Calculate the ratio of the voltage deviation value to the temperature deviation value to obtain the temperature influence coefficient of the any preset temperature; the greater the temperature influence coefficient, the more likely it is to cause voltage monitoring errors at the any preset temperature, and the greater the influence on voltage monitoring.

[0039] Furthermore, during the actual voltage monitoring process, if the temperature and the temperature influence coefficient of the overvoltage monitoring device are obtained, the voltage deviation value can be obtained reversely, and the monitored voltage can be corrected according to the voltage deviation value, so that the corrected monitored voltage is closer to the actual voltage of the device. Therefore, a prediction fitting model can be constructed according to the preset temperature and the corresponding temperature influence coefficient to obtain the temperature influence coefficients at different temperatures; since the relationship between temperature and the temperature influence coefficient is non-linear, the existing commonly used SVM (Support Vector Machine) is usually used for prediction fitting. SVM is an existing regression analysis method that can handle non-linear relationships well, and the specific steps will not be elaborated here. SVM usually uses the radial basis kernel RBF as the kernel function in model training. The RBF kernel maps the data to a higher-dimensional space, making the samples that could not be linearly separated originally become separable in the new space. There is an important hyperparameter gamma in this RBF kernel , which can determine the influence range of a single training sample, and thus determine the accuracy of the subsequent model for training sample data. However, the selection range of the gamma value is relatively large, usually between and Among them, if the gamma value is traversed and selected within this range for accuracy verification, it will take a lot of time, affecting the efficiency of model construction and overvoltage monitoring and identification. Therefore, it is necessary to provide the selection speed of the gamma value and determine the value with high fitting accuracy in a short time.

[0040] Since the gamma value determines the influence range of a single sample on the model decision boundary, the selection of its size is negatively correlated with the distribution dispersion degree of the data. First, obtain the temperature dispersion characteristic value according to the distribution characteristics of the preset temperature. Preferably, in the embodiment of the present invention, the steps of obtaining the temperature dispersion characteristic value include: performing standardization processing on the preset temperature to obtain the preset temperature standard value; in the embodiment of the present invention, standard deviation standardization is used to normalize the data to remove the dimension. Calculate the average value of the absolute values of the differences between any two preset temperature standard values among all the preset temperature standard values to obtain the temperature average difference value; when the temperature average difference value is larger, it means that the selection of the preset temperature is more random. Calculate the product of the standard deviation of the preset temperature standard value and the temperature average difference value to obtain the temperature dispersion characteristic value; when the standard deviation and the temperature average difference value are larger, it means that the distribution of the preset temperature is more discrete, the temperature dispersion characteristic value is larger, and the change correlation is smaller.

[0041] Furthermore, obtain the influence coefficient dispersion characteristic value according to the distribution characteristics of the temperature influence coefficient. Preferably, in the embodiment of the present invention, the steps of obtaining the influence coefficient dispersion characteristic value include: performing standardization processing on the temperature influence coefficient to obtain the influence coefficient standard value; using standard deviation standardization for normalization. Calculate the average value of the absolute values of the differences between any two influence coefficient standard values among all the influence coefficient standard values to obtain the influence coefficient average difference value; when the influence coefficient average difference value is larger, it means that the distribution of the temperature influence coefficient is more discrete. Calculate the product of the standard deviation of the influence coefficient standard value and the influence coefficient average difference value to obtain the influence coefficient dispersion characteristic value; when the standard deviation and the influence coefficient average difference value are larger, it means that the distribution of the temperature influence coefficients under different preset temperatures is more discrete, the influence coefficient dispersion characteristic value is larger, and the change correlation is smaller.

[0042] Since the change of the temperature influence coefficient is related to the temperature change, if the temperature dispersion characteristic value is larger, the corresponding influence coefficient dispersion characteristic value is also larger. Therefore, in order to avoid the influence of the temperature change amount on the change characteristics of the temperature influence coefficient, the relative dispersion degree can be obtained according to the difference characteristics of the temperature dispersion characteristic value and the influence coefficient dispersion characteristic value. Preferably, in the embodiment of the present invention, the steps of obtaining the relative dispersion degree include: calculating the ratio of the influence coefficient dispersion characteristic value to the temperature dispersion characteristic value to obtain the relative dispersion degree. When the relative dispersion degree is larger, it means that the change amplitude of the temperature influence coefficient is relatively larger, and the dispersion degree of the temperature influence coefficient is more obvious.

[0043] The model construction module S3 is used to adjust the preset default gamma value of the SVM (Support Vector Machine) according to the relative discreteness degree to obtain the initial gamma value; construct an SVM model of the preset temperature and the temperature influence coefficient according to the initial gamma value; obtain the model error degree of the initial gamma value according to the prediction error characteristics of the temperature influence coefficient in the SVM model; perform addition and subtraction adjustment on the initial gamma value by a preset multiple to obtain the iterative gamma value; continue to perform addition and subtraction adjustment on the iterative gamma value according to the difference characteristics between the model error degrees of the iterative gamma value and the initial gamma value to obtain the final gamma value.

[0044] Since the value of the gamma value is negatively correlated with the distribution discreteness degree of the data, the greater the relative discreteness degree, the smaller the gamma value required; therefore, the preset default gamma value of the SVM is adjusted according to the relative discreteness degree to obtain the initial gamma value; preferably, in the embodiment of the present invention, the step of obtaining the initial gamma value includes: calculating the product of the reciprocal of the relative discreteness degree and the preset default gamma value to obtain the initial gamma value. The common default gamma value is the constant 1. In the embodiment of the present invention, the preset default gamma value takes the constant 1, and the implementer can determine it according to the implementation scenario. When the relative discreteness degree corresponding to the temperature influence coefficient is large, a smaller initial gamma value is required to avoid overfitting and enhance the generalization ability of the SVM model; when the relative discreteness degree is small, a larger initial gamma value is required, so as to balance between the complexity and generalization ability of the model and finally accurately predict the temperature influence coefficient at any temperature. Further, an SVM model of the preset temperature and the temperature influence coefficient can be constructed according to the initial gamma value; it should be noted that constructing an SVM model according to the temperature influence coefficients and the initial gamma value at different preset temperatures belongs to the prior art, and the specific steps will not be elaborated.

[0045] Furthermore, the SVM model obtained according to the initial gamma value can relatively accurately predict the temperature influence coefficients at different temperatures. Compared with traversing and selecting appropriate gamma values in a relatively wide range of gamma values for model construction, it can greatly improve the efficiency of model construction and overpressure monitoring. In order to further improve the prediction fitting accuracy of the SVM model, adjustment can be performed in the neighborhood of the initial gamma value to select a gamma value with higher prediction accuracy; therefore, it is necessary to judge the error degree of the SVM model predictions corresponding to different gamma values, so the model error degree of the initial gamma value is obtained according to the prediction error characteristics of the temperature influence coefficient in the SVM model.

[0046] Preferably, in the embodiments of the present invention, the step of obtaining the model error degree includes: calculating the difference between the temperature influence coefficient at the same preset temperature and the predicted value of the SVM model to obtain a prediction difference; the larger the prediction difference, the more inaccurate the prediction result. Calculate the average value of the absolute values of all prediction differences to obtain the average error; the larger the average error, the more the prediction result of the model deviates from the actual value. Calculate the product of the absolute value of the skewness of the prediction difference and the average error to obtain the model error degree of the initial gamma value. It should be noted that the calculation of skewness belongs to the prior art, and the specific steps will not be elaborated. The smaller the skewness, the closer the distribution of the prediction difference is to the Gaussian distribution, the more random the error, and the better the prediction effect; on the contrary, the larger the skewness, the worse the prediction effect. The smaller the model error degree, the more accurate the prediction of the SVM model with the initial gamma value; the formula for obtaining the model error degree includes:

[0047]

[0048] In the formula, D represents the model error degree, F represents the skewness of the prediction difference, N represents the number of preset temperatures, represents the prediction difference at the nth preset temperature, represents the average error.

[0049] Further, after determining the prediction error degree of the SVM model, the initial gamma value can be adjusted by a preset multiple to increase or decrease to obtain an iterative gamma value; preferably, in the embodiments of the present invention, the step of obtaining the iterative gamma value includes: the preset multiple includes a preset first multiple and a preset second multiple, the preset first multiple is less than the constant 1, and the preset second multiple is greater than the constant 1; in the embodiments of the present invention, the preset first multiple is 0.8 and the preset second multiple is 1.2, which can increase or decrease the initial gamma value, and the implementer can determine it according to the implementation scenario. Calculate the product of the initial gamma value and the preset first multiple to obtain the iterative gamma value , which is smaller than the initial gamma value; calculate the product of the initial gamma value and the preset second multiple to obtain the iterative gamma value , which is larger than the initial gamma value.

[0050] After obtaining different iterative gamma values, it is necessary to judge a more appropriate gamma value for model construction; the more the selected gamma value conforms to the relationship between the temperature influence coefficient and the temperature, the smaller the model error degree; on the contrary, the larger the model error degree. In the correlation model between the gamma value and the model error degree, at the most appropriate gamma value, the value of the model error degree is smaller, and the larger the difference from the most appropriate gamma value, the larger the value of the model error degree. Furthermore, the iterative gamma value can be adjusted to increase or decrease according to the difference characteristics of the model error degrees of the iterative gamma value and the initial gamma value to obtain the final gamma value.

[0051] Preferably, in the embodiments of the present invention, the step of obtaining the final gamma value includes: when the difference between the initial gamma value and the model error degree of the iterative gamma value exceeds a constant 0, it means that the prediction accuracy of the iterative gamma value is higher than that of the initial gamma value, then it is necessary to further reduce the initial gamma value and determine whether there is a smaller gamma value that makes the prediction accuracy higher; therefore, calculate the product of the iterative gamma value and a preset first multiple to obtain a new iterative gamma value , and then compare the model error degrees of the iterative gamma value and the iterative gamma value and continue the iteration. Similarly, when the difference between the initial gamma value and the model error degree of the iterative gamma value exceeds a constant 0, it means that the prediction accuracy of the iterative gamma value is higher than that of the initial gamma value, then it is necessary to further increase the initial gamma value; calculate the product of the iterative gamma value and a preset second multiple to obtain a new iterative gamma value , compare the model error degrees of the iterative gamma value and the iterative gamma value and continue the iteration. It should be noted that since both too large or too small gamma values will cause an increase in the model error degree, the two situations where the difference between the initial gamma value and the model error degree of the iterative gamma value exceeds a constant 0 and the difference between the initial gamma value and the model error degree of the iterative gamma value exceeds a constant 0 will not occur simultaneously. If the model error degree of the latest round of iterative gamma value is greater than that of the previous round of iterative gamma value, it means that the most suitable gamma value is between the latest round of iterative gamma value and the previous round of iterative gamma value, and the numerical difference between the two rounds of iterative gamma values is small. Therefore, the difference in the model error degrees corresponding to the gamma values within this range interval is small. To balance the selection efficiency of the gamma value and the prediction accuracy of the SVM model, the average value of the latest round of iterative gamma value and the previous round of iterative gamma value can be used as the final gamma value. The prediction accuracy of the SVM model with the final gamma value is higher than that of the initial gamma value. Therefore, obtaining the final gamma value can improve the construction efficiency of the SVM model on the premise of a relatively high model prediction accuracy compared with traversing and selecting gamma values to construct the SVM model, and make the prediction speed of the temperature influence coefficient faster.

[0052] The voltage monitoring module S4 is used to obtain the predicted temperature influence coefficient at any temperature according to the SVM model with the final gamma value; and obtain the corrected monitoring voltage at any temperature according to the predicted temperature influence coefficient.

[0053] The SVM model based on the final gamma value can accurately predict and fit the predicted temperature influence coefficient at any temperature; furthermore, the corrected monitoring voltage at any temperature can be obtained based on the predicted temperature influence coefficient. The specific steps include: calculating the difference between any temperature and the preset normal temperature to obtain the actual temperature difference; calculating the product of the actual temperature difference and the preset temperature influence coefficient to obtain the voltage difference; calculating the difference between the monitoring voltage at any temperature and the voltage difference to obtain the corrected monitoring voltage at that any temperature. The formula for obtaining the corrected monitoring voltage is as follows:

[0054]

[0055] In the formula, T represents any temperature, represents the corrected monitoring voltage at temperature T, represents the monitoring voltage at temperature T, represents the predicted temperature influence coefficient at temperature T, represents the preset normal temperature, represents the actual temperature difference, represents the voltage difference. The corrected monitoring voltage can more accurately determine whether the device has an overvoltage fault compared to the measured monitoring voltage, reducing the influence of temperature on measuring the device voltage by the overvoltage monitoring device and improving the stability and safety of the device operation. So far, first, the temperature influence coefficients at different preset temperatures are obtained through experiments, a relatively appropriate initial gamma value is obtained based on the distribution and dispersion characteristics of the temperature influence coefficients, and the SVM model obtained based on the initial gamma value has a good prediction effect on the temperature influence coefficient; subsequently, the initial gamma value is iterated to obtain the final gamma value with the best prediction effect, which can significantly reduce the SVM model construction time compared to traversing and selecting the final gamma value in a relatively large range of gamma values and improve the efficiency of predicting and fitting the temperature influence coefficient and overvoltage monitoring.

[0056] In summary, the embodiment of the present invention provides an overvoltage fault identification system for a multiplexing circuit based on a sensor; obtaining the temperature influence coefficient of the preset temperature according to the preset temperature and the difference characteristics between the monitoring voltage and the actual voltage; obtaining the relative dispersion degree according to the distribution and dispersion characteristics of the preset temperature and the temperature influence coefficient; adjusting the preset default gamma value according to the relative dispersion degree to obtain the initial gamma value and the corresponding SVM model; obtaining the model error degree according to the prediction error characteristics of the temperature influence coefficient; adjusting the initial gamma value to obtain the iterative gamma value. The present invention iteratively adjusts according to the model error degrees of the iterative gamma value and the initial gamma value to obtain the final gamma value; obtaining the predicted temperature influence coefficient according to the SVM model of the final gamma value; obtaining the corrected monitoring voltage according to the predicted temperature influence coefficient; improving the efficiency of overvoltage monitoring and predicting and fitting the predicted temperature influence coefficient.

[0057] It should be noted that the above sequence of embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0058] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the key point of each embodiment is to illustrate the differences from other embodiments.

Claims

1. A sensor-based multiplexing circuit overvoltage fault identification system, characterized in that, The system includes the following modules: A data acquisition module, configured to acquire the monitored voltages of the actual voltage at different preset temperatures in the experiment; A temperature influence analysis module, configured to obtain the temperature influence coefficient of the preset temperature according to the preset temperature and the difference characteristics between the monitored voltage and the actual voltage at the preset temperature; obtain the temperature discrete characteristic value according to the distribution characteristic of the preset temperature; obtain the influence coefficient discrete characteristic value according to the distribution characteristic of the temperature influence coefficient; obtain the relative discrete degree according to the difference characteristics between the temperature discrete characteristic value and the influence coefficient discrete characteristic value; A model construction module, configured to adjust the preset default gamma value of the SVM support vector machine according to the relative discrete degree to obtain the initial gamma value; construct an SVM model of the preset temperature and the temperature influence coefficient according to the initial gamma value; obtain the model error degree of the initial gamma value according to the prediction error characteristics of the temperature influence coefficient in the SVM model; perform an adjustment of increasing or decreasing the initial gamma value by a preset multiple to obtain the iterative gamma value; continue to adjust the iterative gamma value by increasing or decreasing according to the difference characteristics between the iterative gamma value and the model error degree of the initial gamma value to obtain the final gamma value; A voltage monitoring module, configured to obtain the predicted temperature influence coefficient at any temperature according to the SVM model of the final gamma value; obtain the corrected monitored voltage at the any temperature according to the predicted temperature influence coefficient.

2. The overvoltage fault identification system for a multiplexing circuit based on a sensor according to claim 1, wherein The step of obtaining the temperature influence coefficient of the preset temperature according to the preset temperature and the difference characteristics between the monitored voltage and the actual voltage at the preset temperature includes: Calculating the difference between any preset temperature and the preset normal temperature to obtain the temperature deviation value; calculating the difference between the monitored voltage and the actual voltage at the any preset temperature to obtain the voltage deviation value; calculating the ratio of the voltage deviation value to the temperature deviation value to obtain the temperature influence coefficient of the any preset temperature.

3. The overvoltage fault recognition system for a multiplexing circuit based on a sensor according to claim 1, characterized in that, The step of obtaining the temperature discrete characteristic value according to the distribution characteristic of the preset temperature includes: Performing a normalization process on the preset temperature to obtain the preset temperature standard value; calculating the average value of the absolute values of the differences between any two preset temperature standard values among all the preset temperature standard values to obtain the temperature average difference value; calculating the product of the standard deviation of the preset temperature standard value and the temperature average difference value to obtain the temperature discrete characteristic value.

4. A sensor-based multiplexing circuit overvoltage fault identification system according to claim 1, characterized in that, The step of obtaining the influence coefficient discrete characteristic value according to the distribution characteristic of the temperature influence coefficient includes: Performing a normalization process on the temperature influence coefficient to obtain the influence coefficient standard value; calculating the average value of the absolute values of the differences between any two influence coefficient standard values among all the influence coefficient standard values to obtain the influence coefficient average difference value; calculating the product of the standard deviation of the influence coefficient standard value and the influence coefficient average difference value to obtain the influence coefficient discrete characteristic value.

5. A sensor-based multiplexing circuit overvoltage fault identification system according to claim 1, characterized in that The step of obtaining the relative discrete degree according to the difference characteristics between the temperature discrete characteristic value and the influence coefficient discrete characteristic value includes: Calculating the ratio of the influence coefficient discrete characteristic value to the temperature discrete characteristic value to obtain the relative discrete degree.

6. The overvoltage fault identification system for a multiplexing circuit based on a sensor according to claim 1, characterized in that, The step of adjusting the preset default gamma value of the SVM support vector machine according to the relative discrete degree to obtain the initial gamma value includes: Calculate the product of the reciprocal of the relative dispersion degree and a preset default gamma value to obtain the initial gamma value.

7. The overvoltage fault identification system for a multiplexing circuit based on sensors according to claim 1, characterized in that, The step of obtaining the model error degree of the initial gamma value according to the prediction error characteristics of the temperature influence coefficient in the SVM model includes: Calculate the difference between the temperature influence coefficient at the same preset temperature and the predicted value of the SVM model to obtain a prediction difference; calculate the average value of the absolute values of all prediction differences to obtain an average error; calculate the product of the absolute value of the skewness of the prediction difference and the average error to obtain the model error degree of the initial gamma value.

8. A sensor-based multiplexing circuit overvoltage fault identification system according to claim 1, wherein, The step of obtaining an iterative gamma value by increasing or decreasing the initial gamma value by a preset multiple includes: The preset multiples include a preset first multiple and a preset second multiple, where the preset first multiple is less than the constant 1 and the preset second multiple is greater than the constant 1; calculate the product of the initial gamma value and the preset first multiple to obtain an iterative gamma value ; calculate the product of the initial gamma value and the preset second multiple to obtain an iterative gamma value .

9. The overvoltage fault recognition system of a sensor-based multiplexing circuit according to claim 8, wherein The step of continuing to increase or decrease the iterative gamma value according to the difference characteristics between the iterative gamma value and the model error degree of the initial gamma value to obtain a final gamma value includes: When the difference between the initial gamma value and the model error degree of the iterative gamma value exceeds a constant 0, calculate the product of the iterative gamma value and a preset first multiple to obtain a new iterative gamma value , compare the model error degrees of the iterative gamma value and the iterative gamma value and continue the iteration; when the difference between the initial gamma value and the model error degree of the iterative gamma value exceeds a constant 0, calculate the product of the iterative gamma value and a preset second multiple to obtain a new iterative gamma value , compare the model error degrees of the iterative gamma value and the iterative gamma value and continue the iteration; If the model error degree of the latest iterative gamma value is greater than that of the previous iterative gamma value, take the average value of the latest iterative gamma value and the previous iterative gamma value as the final gamma value.

10. A sensor-based multiplexing circuit overvoltage fault identification system according to claim 1, wherein, The step of obtaining the corrected monitoring voltage at any temperature according to the predicted temperature influence coefficient includes: Calculate the difference between the any temperature and the preset normal temperature to obtain an actual temperature difference; calculate the product of the actual temperature difference and the preset temperature influence coefficient to obtain a voltage difference amount; calculate the difference between the monitoring voltage at the any temperature and the voltage difference amount to obtain the corrected monitoring voltage at the any temperature.

Images