Capacitor quality detection method and system based on big data

Abstract

The invention discloses a capacitor quality detection method and system based on big data, and relates to the technical field of power equipment intelligent monitoring, and the method comprises the steps: generating a data matrix through multi-source data, extracting each column in the data matrix, representing the column as a scalar function, simulating dynamic evolution through an asymmetric thermal diffusion equation, and calculating a feature field value after evolution; calculating a feature evolution index based on the evolved feature field value, converting a normalized index into a discretization level by using linear mapping, performing probability interval uniform division based on the discretization level, allocating each feature column to a corresponding equal probability interval, and endowing a discrete value; and calculating a quality score and detecting the quality of the capacitor. According to the method, by introducing the Shannon entropy and the asymmetric thermal diffusion model, the discriminability of the abnormal state is improved, and by combining structure tensor and anisotropic flux analysis, the misjudgment risk caused by the imbalance between the characteristics and the distribution skewness problem is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring of power equipment, and in particular to a capacitor quality detection method and system based on big data. Background Art

[0002] With the rapid development of smart grid and power electronics technology, capacitors, as important components widely used in power systems, have a health status that is directly related to the stability and safety of system operation. Traditional capacitor detection methods mostly rely on manual inspections or simple threshold judgments, which cannot meet the current needs of online equipment monitoring and fault warning. With the continuous maturity of sensor technology, edge computing and big data analysis, status assessment based on multi-source data has gradually become a research hotspot. Existing technologies usually use single or combined data sources such as voltage, current, and temperature to perform health judgments by setting thresholds, trend analysis, or traditional statistical models.

[0003] There are still some deficiencies in the existing capacitor quality detection technology. Currently, single-source or limited multi-source data analysis is mostly used, and there is a lack of modeling of the temporal dependency and dynamic evolution process between multi-source data, resulting in insufficient stability of the detection results. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a capacitor quality detection method and system based on big data to solve the problem that the current single-source or limited multi-source data analysis is mostly used, and there is a lack of modeling of the temporal dependency and dynamic evolution process between multi-source data, resulting in insufficient stability of the detection results.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a capacitor quality detection method based on big data, which comprises the following steps: Collect multi-source data and perform preprocessing, generate a data matrix from the multi-source data, extract each column in the data matrix, express it as a scalar function, calculate the coarse-grained value of the feature, calculate the Shannon entropy of the coarse-grained value, use the asymmetric heat diffusion equation to simulate the dynamic evolution, and calculate the characteristic field value after evolution; Calculate the first-order and second-order derivatives of the evolved eigenvalue values ​​to construct a structure tensor, perform eigenvalue decomposition on the structure tensor, calculate the anisotropic flux of the decomposed eigenvalues, calculate the characteristic evolution index based on the evolved eigenvalues, use linear mapping to convert the normalized index into a discretization level, evenly divide the probability interval based on the discretization level, assign each feature column to the corresponding equal probability interval, and assign a discrete value; Calculate the quality score and test the capacitor quality, and build a visual interface to display the quality test results.

[0007] As a preferred solution of the capacitor quality detection method based on big data described in the present invention, wherein: generating a data matrix from multi-source data, extracting each column in the data matrix, expressing it as a scalar function, calculating the coarse-grained value of the feature, calculating the Shannon entropy of the coarse-grained value, using the asymmetric heat diffusion equation to simulate dynamic evolution, and calculating the evolved feature field value, includes: Use matrix construction to generate data matrix from multi-source data, extract each column in the data matrix, define it as a one-dimensional time series feature field using time series representation, and express it as a scalar function; Use equal-interval time scale division to calculate candidate time scales, and use the sliding window method to calculate the coarse-grained value based on the candidate time scales; The Shannon entropy of the coarse-grained value is calculated using the Shannon entropy formula, the dynamic evolution is simulated using the asymmetric heat diffusion equation, the asymmetric heat diffusion equation is solved using fast Fourier transform, and the characteristic field value after evolution is calculated.

[0008] As a preferred solution of the capacitor quality detection method based on big data of the present invention, the calculation of the characteristic evolution index based on the evolved characteristic field value includes: The first-order derivative and second-order derivative of the evolved eigenfield value are calculated using the central difference method, the first-order and second-order derivatives are used to construct the structure tensor, and the eigenvalue decomposition of the structure tensor is performed to obtain the decomposed eigenvalue; Calculate the anisotropic flux of the decomposed eigenvalues, calculate the mean of the anisotropic flux, and generate anisotropic flux index; Based on the evolved feature field value, the feature evolution index is calculated, and the mean is calculated to generate the feature evolution index; The anisotropic flux index and characteristic evolution index are integrated to calculate the geometric-thermal evolution index.

[0009] As a preferred solution of the capacitor quality detection method based on big data of the present invention, wherein: the normalized index is converted into a discretization level using a linear mapping, the probability interval is evenly divided based on the discretization level, each feature column is assigned to a corresponding equal probability interval, and a discrete value is assigned, including: Normalizing the geometric-thermal evolution index, converting the normalized index into a discretization level using a linear mapping, and setting a range constraint on the discretization level; Extract each column in the data matrix, construct the empirical distribution function using kernel density estimation, evenly divide the probability interval based on the discretization level, and use quantiles to calculate the quantile of each subinterval; Use the interval mapping method to assign each feature column to the corresponding equal probability interval and assign discrete values.

[0010] As a preferred solution of the capacitor quality detection method based on big data of the present invention, the calculation of the quality score and the detection of the capacitor quality include: Collect the historical discrete values ​​with labels and average them, define them as the benchmark, and use the Mahalanobis distance formula to calculate the deviation between the discrete values ​​and the benchmark; The quality score was calculated using a weighted moving average; The detection threshold is set using statistical analysis methods, and the quality score is compared with the detection threshold. When the quality score is less than the detection threshold, it is judged as normal quality, otherwise it is judged as abnormal quality.

[0011] As a preferred solution of the capacitor quality detection method based on big data of the present invention, the step of constructing a visual interface to display the quality detection results includes: Use the front-end framework React.js to build a visual interface to visualize the quality inspection results; Users who have passed real-name verification are allowed to view the information.

[0012] As a preferred solution of the capacitor quality detection method based on big data of the present invention, the collecting of multi-source data and preprocessing thereof include: Use smart sensors to collect multi-source data and perform time stamp alignment, denoising, and normalization processing; The intelligent sensors include voltage, current, temperature and vibration sensors; The multi-source data includes voltage, current, temperature and vibration data.

[0013] In a second aspect, the present invention provides a capacitor quality detection system based on big data, comprising: The collection and evolution module is used to collect and preprocess multi-source data, generate a data matrix from the multi-source data, extract each column in the data matrix, express it as a scalar function, calculate the coarse-grained value of the feature, calculate the Shannon entropy of the coarse-grained value, simulate the dynamic evolution using the asymmetric heat diffusion equation, and calculate the feature field value after evolution; The indicator discretization module is used to calculate the first-order derivative and second-order derivative of the evolved characteristic field value to construct the structure tensor, perform eigenvalue decomposition on the structure tensor, calculate the anisotropic flux of the decomposed eigenvalue, calculate the characteristic evolution index based on the evolved characteristic field value, use linear mapping to convert the normalized index into a discretization level, evenly divide the probability interval based on the discretization level, assign each feature column to the corresponding equal probability interval, and assign a discrete value; The detection visualization module is used to calculate the quality score and detect the quality of capacitors, and build a visual interface to display the quality detection results.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the capacitor quality detection method based on big data as described in the first aspect of the present invention is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the capacitor quality detection method based on big data as described in the first aspect of the present invention is implemented.

[0016] The beneficial effects of the present invention are as follows: the present invention improves the discriminability of abnormal states by introducing Shannon entropy and asymmetric heat diffusion model, and reduces the risk of misjudgment caused by imbalance between features and distribution skewness problems by combining structural tensor and anisotropic flux analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 This is an operation flow chart of the capacitor quality detection method based on big data in Example 1.

[0019] Figure 2 This is a schematic diagram of the structure of the capacitor quality detection system based on big data in Example 1. DETAILED DESCRIPTION

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0021] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0023] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a capacitor quality detection method based on big data, comprising the following steps: S1. Collect multi-source data and perform preprocessing. Generate a data matrix from the multi-source data. Extract each column in the data matrix and express it as a scalar function. Calculate the coarse-grained value of the feature. Calculate the Shannon entropy of the coarse-grained value. Use the asymmetric heat diffusion equation to simulate the dynamic evolution and calculate the feature field value after evolution. Specifically, multi-source data is collected and preprocessed, including: Use smart sensors to collect multi-source data and perform time stamp alignment, denoising, and normalization processing; The intelligent sensors include voltage, current, temperature and vibration sensors; The multi-source data includes voltage, current, temperature and vibration data Multi-source data reduces modeling deviations caused by differences in sampling frequency, amplitude scale, etc. between heterogeneous features, builds a high-dimensional observation basis for the dynamic behavior of capacitors, and supports multi-level information extraction.

[0024] Furthermore, a data matrix is ​​generated from multi-source data, each column in the data matrix is ​​extracted and expressed as a scalar function, the coarse-grained value of the feature is calculated, the Shannon entropy of the coarse-grained value is calculated, the dynamic evolution is simulated using the asymmetric heat diffusion equation, and the characteristic field value after evolution is calculated, including: Use matrix construction to generate a data matrix from multi-source data. Each row of the matrix corresponds to a time point of a sensor, and each column corresponds to the measurement value of the smart sensor at the same time point. Extract each column in the data matrix and define it as a one-dimensional time series feature field using the time series representation method, expressed as a scalar function, and the formula is: , in is the characteristic field value of feature j at time point t, is the value of feature j at time point t in the data matrix, representing the collected multi-source data, t is a discrete time point, and j is the feature index; Using equally spaced time scales, we can calculate candidate time scales using the formula: , in is the i-th candidate time scale, i is the scale index, is the time sampling interval; Based on the candidate time scale, the coarse-grained value is calculated using the sliding window method, and the formula is: , in is the K-th coarse-grained value of feature j at the candidate time scale, s is the sampling frequency, and K is the coarse-grained sequence index; The Shannon entropy formula is used to calculate the Shannon entropy of the coarse-grained value. The formula is: , in is the Shannon entropy at the candidate time scale, is the probability of the coarse-grained value, estimated by the histogram; The second-order derivatives are calculated using the central difference method, and the dynamic evolution is simulated using the asymmetric heat diffusion equation, which is: , in is the Shannon entropy at the candidate time scale, is the probability of the coarse-grained value, estimated by the histogram; The second-order derivatives are calculated using the central difference method, and the dynamic evolution is simulated using the asymmetric heat diffusion equation, which is: , in is feature j at the candidate time scale Under the frequency domain representation of the characteristic field after asymmetric thermal diffusion evolution, FFT and are the fast Fourier transform and inverse transform, is feature j at the candidate time scale and the characteristic field value after evolution at time t.

[0025] Building a bridge from discrete data to mathematical functions not only improves the model's abstract expression capabilities, but also enables the subsequent direct application of physical modeling and transformation tools (such as derivatives, Fourier transforms, etc.), providing theoretical support for the introduction of dynamic evolution mechanisms. The data sequence is reconstructed at multiple scales through a sliding window mechanism to filter out short-term local disturbances, making the system more sensitive to long-term evolution trends. At the same time, multi-scale inputs are constructed to improve the model's robustness and cross-scale recognition capabilities for complex dynamic systems. By analyzing the Shannon entropy of characteristic sequences at different coarse-grained scales, the degree of uncertainty of the system at each scale can be quantified, and fragments with larger volatility and stronger structure in the sequence are found. The asymmetric thermal diffusion model is applied for the first time to the capacitor mass evolution field modeling, improving the model's response to small trend anomalies. The frequency domain reconstruction is combined with the partial differential modeling method, taking into account both timeliness and information density expression, to achieve a dual improvement in modeling efficiency and accuracy.

[0026] S2. Calculate the first-order derivative and second-order derivative of the evolved characteristic field value to construct a structure tensor, perform eigenvalue decomposition on the structure tensor, calculate the anisotropic flux of the decomposed eigenvalue, calculate the characteristic evolution index based on the evolved characteristic field value, use linear mapping to convert the normalized index into a discretization level, evenly divide the probability interval based on the discretization level, assign each characteristic column to the corresponding equal probability interval, and assign a discrete value; Specifically, based on the evolved feature field values, the feature evolution indicators are calculated, including: The first-order and second-order derivatives of the evolved characteristic field values ​​are calculated using the central difference method, and the structure tensor is constructed using the first-order and second-order derivatives. The formula is: , in is the jth feature at the candidate time scale and the structure tensor at time t, and are the jth feature at the candidate time scale and the first and second derivatives of the characteristic field values ​​after evolution at time t; Perform eigenvalue decomposition on the structure tensor to obtain the decomposition eigenvalue. The formula is: , Where U is the eigenvector matrix, T is the transpose operation, and To decompose the eigenvalues, they are represented as principal eigenvalues ​​and principal eigenvalues ​​respectively; Calculate the anisotropic flux of the decomposed eigenvalues ​​using the formula: , in is the jth feature at the candidate time scale and the anisotropic flux at time t; Calculate the mean of the anisotropic flux and generate the anisotropic flux index; Based on the evolved feature field value, the feature evolution index is calculated, and the mean is calculated to generate the feature evolution index. The formula is: , in is the jth feature at the candidate time scale The characteristic evolution index of n is the number of sampling points, is a very small constant to prevent the denominator from being zero; The anisotropic flux index and the characteristic evolution index are integrated to calculate the geometric-thermal evolution index. The formula is: , in is the geometric-thermal evolution index of the jth feature, is the anisotropic flux index of the jth feature, is the feature evolution index of the j-th feature.

[0027] Enhance the ability to identify subtle changes in equipment, especially for capturing early signs of degradation such as temperature changes and vibration fluctuations. Through gradient and curvature information, it can identify potential local abnormal areas inside the equipment, thereby helping to locate the potential location of faults in advance. Different equipment will exhibit different anisotropic characteristics when working. By calculating the anisotropic flux, the performance changes of the equipment can be analyzed in multiple dimensions and potential functional degradation can be identified. This evolution indicator can serve as the basis for equipment health prediction and help predict the possible failure time point or degradation rate of the equipment. The geometric-thermal evolution index provides a multi-dimensional and multi-level comprehensive assessment of the equipment's operating status, which can accurately identify subtle changes in equipment operation and respond in a timely manner.

[0028] Furthermore, a linear mapping is used to convert the normalized index into a discretization level, and the probability interval is evenly divided based on the discretization level. Each feature column is assigned to the corresponding equal probability interval and assigned a discrete value, including: The geometric-thermal evolution index is normalized and converted into a discretization level using a linear mapping. The formula is: , in is the discretization level of the jth feature, and are the minimum and maximum values ​​of the discretization level, respectively, and are set using the Monte Carlo simulation method. is the normalized index; The range constraint is imposed on the discretization level, and the formula is: , in is the discretization level after adjustment; Extract each column in the data matrix and use kernel density estimation to construct the empirical distribution function. The formula is: , , in is the kernel density estimation function of the jth feature, x is the input eigenvalue, which represents the collected multi-source data value, is the value of feature j in the data matrix at time point a, is the kernel density estimation bandwidth of the j-th feature, is the Gaussian kernel function, is the empirical distribution function of the jth feature, u is the integral variable, It is the lower bound in the calculation of the quantile of the empirical distribution function; The probability interval is evenly divided based on the discretization level, and the quantile is used to calculate the quantile of each sub-interval. The boundary is defined as , , the formula is: , in is the lth quantile of the jth feature, l is the quantile index, is the inverse function of the j-th feature empirical distribution function; Use the interval mapping method to assign each feature column to the corresponding equal probability interval and assign discrete values. The formula is: , , in is the discrete value of the jth feature at time point a, is an exponential function.

[0029] Standardizing the health status of equipment into discrete levels facilitates comparative analysis with historical data and enhances the automated processing capabilities of the monitoring system. Compared with traditional discretization methods, kernel density estimation can adaptively construct data distribution without prior assumptions, effectively improving the ability to accurately model the distribution of equipment characteristics. Through unbiased estimation and adaptive distribution modeling, it solves the problem of traditional methods' dependence on distribution assumptions and enhances the flexibility and applicability of analysis.

[0030] S3. Calculate the quality score and test the capacitor quality, and build a visual interface to display the quality test results; Specifically, the quality score is calculated and the capacitor quality is tested, including: Collect the historical discrete values ​​with labels and average them, which is defined as the benchmark. The Mahalanobis distance formula is used to calculate the deviation between the discrete value and the benchmark. The formula is: , , in is the Mahalanobis distance at time point o, is the discrete value at time point o, is the benchmark, is the covariance matrix of the benchmark, is the number of historical discrete values ​​under normal conditions, and R is the transposition operation; The quality score is calculated using a weighted moving average, using the formula: , , in is the quality score of time point o, T is the time window size, and the sliding window setting is used. is the weight of time offset m, m is the time offset, To traverse all time offsets within the time window, is the attenuation factor, which is set using the empirical parameter selection method; The detection threshold is set using statistical analysis methods, and the quality score is compared with the detection threshold. When the quality score is less than the detection threshold, it is judged as normal quality, otherwise it is judged as abnormal quality.

[0031] The Mahalanobis distance can identify anomalies caused by "co-variation" between dependent variables. Based on historical covariance matrix estimation, it can dynamically adapt to the normal behavior distribution under different devices or environments. For outliers caused by occasional noise or sensor jitter, the sliding average mechanism has a natural smoothing and filtering capability. Traditional fixed thresholds cannot adapt to differences between devices. This mechanism can significantly reduce false positives and missed negatives.

[0032] Furthermore, a visual interface is built to display the quality inspection results, including: Use the front-end framework React.js to build a visual interface to visualize the quality inspection results; Users who have passed real-name verification are allowed to view the information.

[0033] The front-end framework is used to visualize quality scores. Visual modules such as real-time curves, alarm prompts, and trend forecast charts are used to help technicians intuitively understand the operating status of the equipment. The interface supports a real-name verification mechanism to ensure that information access is secure and controllable.

[0034] Example 2, reference Figure 2 In a second embodiment of the present invention, a capacitor quality detection system based on big data includes: The collection and evolution module is used to collect and preprocess multi-source data, generate a data matrix from the multi-source data, extract each column in the data matrix, express it as a scalar function, calculate the coarse-grained value of the feature, calculate the Shannon entropy of the coarse-grained value, simulate the dynamic evolution using the asymmetric heat diffusion equation, and calculate the feature field value after evolution; The indicator discretization module is used to calculate the first-order derivative and second-order derivative of the evolved characteristic field value to construct the structure tensor, perform eigenvalue decomposition on the structure tensor, calculate the anisotropic flux of the decomposed eigenvalue, calculate the characteristic evolution index based on the evolved characteristic field value, use linear mapping to convert the normalized index into a discretization level, evenly divide the probability interval based on the discretization level, assign each feature column to the corresponding equal probability interval, and assign a discrete value; The detection visualization module is used to calculate the quality score and detect the quality of capacitors, and build a visual interface to display the quality detection results.

[0035] This embodiment also provides a computer device suitable for the capacitor quality detection method based on big data, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the capacitor quality detection method based on big data proposed in the above embodiment.

[0036] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0037] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the capacitor quality detection method based on big data as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk.

[0038] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A capacitor quality detection method based on big data, characterized by: The steps include: Collect multi-source data and perform preprocessing, generate a data matrix from the multi-source data, extract each column in the data matrix, express it as a scalar function, calculate the coarse-grained value of the feature, calculate the Shannon entropy of the coarse-grained value, use the asymmetric heat diffusion equation to simulate the dynamic evolution, and calculate the characteristic field value after evolution; Calculate the first-order and second-order derivatives of the evolved eigenvalue values ​​to construct a structure tensor, perform eigenvalue decomposition on the structure tensor, calculate the anisotropic flux of the decomposed eigenvalues, calculate the characteristic evolution index based on the evolved eigenvalues, use linear mapping to convert the normalized index into a discretization level, evenly divide the probability interval based on the discretization level, assign each feature column to the corresponding equal probability interval, and assign a discrete value; Calculate the quality score and test the capacitor quality, and build a visual interface to display the quality test results.

2. The capacitor quality detection method based on big data according to claim 1, characterized in that: The method generates a data matrix from multi-source data, extracts each column in the data matrix, expresses it as a scalar function, calculates the coarse-grained value of the feature, calculates the Shannon entropy of the coarse-grained value, simulates the dynamic evolution using the asymmetric heat diffusion equation, and calculates the feature field value after evolution, including: Use matrix construction to generate data matrix from multi-source data, extract each column in the data matrix, define it as a one-dimensional time series feature field using time series representation, and express it as a scalar function; Use equal-interval time scale division to calculate candidate time scales, and use the sliding window method to calculate the coarse-grained value based on the candidate time scales; The Shannon entropy of the coarse-grained value is calculated using the Shannon entropy formula, the dynamic evolution is simulated using the asymmetric heat diffusion equation, the asymmetric heat diffusion equation is solved using fast Fourier transform, and the characteristic field value after evolution is calculated.

3. The capacitor quality detection method based on big data according to claim 2, characterized in that: The calculation of the characteristic evolution index based on the evolved characteristic field value includes: The first-order derivative and second-order derivative of the evolved eigenfield value are calculated using the central difference method, the first-order and second-order derivatives are used to construct the structure tensor, and the eigenvalue decomposition of the structure tensor is performed to obtain the decomposed eigenvalue; Calculate the anisotropic flux of the decomposed eigenvalues, calculate the mean of the anisotropic flux, and generate anisotropic flux index; Based on the evolved feature field value, the feature evolution index is calculated, and the mean is calculated to generate the feature evolution index; The anisotropic flux index and characteristic evolution index are integrated to calculate the geometric-thermal evolution index.

4. The capacitor quality detection method based on big data according to claim 3, wherein: The method uses linear mapping to convert the normalized index into a discretization level, evenly divides the probability interval based on the discretization level, assigns each feature column to a corresponding equal probability interval, and assigns a discrete value, including: Normalizing the geometric-thermal evolution index, converting the normalized index into a discretization level using a linear mapping, and setting a range constraint on the discretization level; Extract each column in the data matrix, construct the empirical distribution function using kernel density estimation, evenly divide the probability interval based on the discretization level, and use quantiles to calculate the quantile of each subinterval; Use the interval mapping method to assign each feature column to the corresponding equal probability interval and assign discrete values.

5. The capacitor quality detection method based on big data according to claim 4, characterized in that: The method of calculating the quality score and detecting the quality of the capacitor includes: Collect the historical discrete values ​​with labels and average them, define them as the benchmark, and use the Mahalanobis distance formula to calculate the deviation between the discrete values ​​and the benchmark; The quality score was calculated using a weighted moving average; The detection threshold is set using statistical analysis methods, and the quality score is compared with the detection threshold. When the quality score is less than the detection threshold, it is judged as normal quality, otherwise it is judged as abnormal quality.

6. The capacitor quality detection method based on big data according to claim 5, characterized in that: The construction of a visual interface to display quality inspection results includes: Use the front-end framework React.js to build a visual interface to visualize the quality inspection results; Users who have passed real-name verification are allowed to view the information.

7. The capacitor quality detection method based on big data according to claim 1, wherein: The multi-source data collection and pre-processing includes: Use smart sensors to collect multi-source data and perform time stamp alignment, denoising, and normalization processing; The intelligent sensors include voltage, current, temperature and vibration sensors; The multi-source data includes voltage, current, temperature and vibration data.

8. A capacitor quality detection system based on big data, used to implement the method according to any one of claims 1 to 7, characterized in that: include: The collection and evolution module is used to collect and preprocess multi-source data, generate a data matrix from the multi-source data, extract each column in the data matrix, express it as a scalar function, calculate the coarse-grained value of the feature, calculate the Shannon entropy of the coarse-grained value, simulate the dynamic evolution using the asymmetric heat diffusion equation, and calculate the feature field value after evolution; The indicator discretization module is used to calculate the first-order derivative and second-order derivative of the evolved characteristic field value to construct the structure tensor, perform eigenvalue decomposition on the structure tensor, calculate the anisotropic flux of the decomposed eigenvalue, calculate the characteristic evolution index based on the evolved characteristic field value, use linear mapping to convert the normalized index into a discretization level, evenly divide the probability interval based on the discretization level, assign each feature column to the corresponding equal probability interval, and assign a discrete value; The detection visualization module is used to calculate the quality score and detect the quality of capacitors, and build a visual interface to display the quality detection results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the capacitor quality detection method based on big data according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the capacitor quality detection method based on big data according to any one of claims 1 to 7 are implemented.

Images