Monitoring method, device, equipment and medium for life of metallized film capacitor

Abstract

The present invention is applicable to the technical field of pulse capacitors, and discloses a monitoring method, device, equipment and medium for the life of metallized film capacitors. The discharge signal is subjected to noise reduction processing by Mallat algorithm and discrete wavelet transform, and feature extraction is performed on the processed discharge signal based on the multi-scale spatial energy distribution of wavelet transform, and a discharge signal feature database of different discharge types is established. A discharge type identification model of metallized film capacitors is constructed based on the feature database and machine learning methods, and the model is trained. The trained model is used to predict the real-time discharge signal generated during the operation cycle of the metallized film capacitor, and the life of the metallized film capacitor is determined according to the prediction result, and a warning is issued. The technical solution of the present invention can realize online monitoring of the operation status of the metallized film capacitor and active warning of the expiration of the life, so as to achieve the purpose of safety protection, and has practical application value of being simple, reliable and having high prediction accuracy.

Description

Technical Field

[0001] The present disclosure relates to the technical field of pulse capacitors, and in particular to a method, device, equipment and medium for monitoring the life of metallized film capacitors. Background Art

[0002] Metallized film capacitors are widely used in high-voltage direct current transmission systems and pulse power systems due to their advantages such as fast charging and discharging speed, high operating reliability, and large capacitance. Energy storage capacitors working in pulse power systems will face extreme environments such as high electric fields and large temperature rises. An important reason why metallized film capacitors have a longer cycle life and higher operating reliability than other insulating dielectric capacitors is that they have a self-healing mechanism. Although this mechanism improves the operating reliability of energy storage dielectrics under high field strength, the degree of self-healing inside the capacitor increases sharply under extreme environments. At the same time, the edge position without electrodes at one end of the metallized film will generate an electric field component along the surface during operation, causing surface discharge or even flashover. The above types of discharge will cause the loss of electrodes on the surface of the metal film and the capacitance of the capacitor will decrease. At the same time, the electrical and thermal effects generated by the discharge will change the dielectric structure, causing irreversible insulation degradation, introducing new electrical weaknesses, making the above discharge more likely to occur, and accelerating the process of capacitance reduction. Metallized film capacitors may also experience sudden failure or even extremely strong explosions under high-frequency and high-power density charging and discharging conditions.

[0003] In engineering applications, it is necessary to monitor the operating status and life of metallized film capacitors through certain monitoring methods. The current main monitoring method still requires removing the capacitor, measuring the capacitance of the metallized film capacitor, and determining the life of the capacitor by manually calculating the capacitance value. Removing the capacitor will affect the reliable operation of the pulse power system. Therefore, the current monitoring methods are inefficient and unreliable, and it is difficult to monitor the operating status and remaining life of the metallized film capacitor in real time. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a method, device, equipment and medium for monitoring the life of a metallized film capacitor, which can realize online monitoring of the operating status of the metallized film capacitor and active warning of the expiration of the life, thereby achieving the purpose of safety protection. It has simple, reliable and practical application value with high prediction accuracy.

[0005] To achieve the above objectives, the present disclosure adopts the following technical solutions:

[0006] In a first aspect, the present invention provides a method for monitoring the life of a metallized film capacitor, comprising:

[0007] Acquire a first discharge signal and voltage information of a metallized film capacitor operating in a pulse power system;

[0008] Based on the voltage information of the metallized film capacitor during the discharge process and the surface morphology of the metallized film, a second discharge signal corresponding to the discharge type is identified from the first discharge signal;

[0009] Based on the multi-scale spatial energy distribution of wavelet transform, feature extraction is performed on the second discharge signal to obtain a discharge signal feature database;

[0010] Based on the discharge signal feature database, a discharge type recognition model is constructed;

[0011] The real-time discharge signal of the metallized film capacitor is input into the discharge type recognition model for prediction;

[0012] The life of the metallized film capacitor is determined based on the prediction results of the discharge type identification model.

[0013] Further, after obtaining the first discharge signal and voltage information of the metallized film capacitor working in the pulse power system, it also includes:

[0014] The first discharge signal is subjected to noise reduction processing by using Mallat algorithm and discrete wavelet transform.

[0015] Furthermore, based on the multi-scale spatial energy distribution of wavelet transform, feature extraction is performed on the second discharge signal to obtain a discharge signal feature database, including:

[0016] Obtaining frequency domain characteristics of the second discharge signal through Fourier transform;

[0017] Calculate the energy of each frequency band corresponding to the multi-scale space based on wavelet transform to obtain a feature vector corresponding to the second discharge signal;

[0018] Based on the feature vector, a discharge signal feature database of metallized film capacitors is constructed.

[0019] Furthermore, based on the discharge signal feature database, a discharge type recognition model is constructed, including:

[0020] Based on the linear discriminant analysis method and the discharge signal feature database, a discharge type recognition model is constructed; the optimization objectives of the discharge type recognition model meet the following requirements:

[0021]

[0022] in, is the projection vector, is the between-class scatter matrix, is the intra-class scatter matrix, and T represents the transpose of the matrix.

[0023] Furthermore, after constructing the discharge type recognition model based on the discharge signal feature database, it also includes:

[0024] The discharge type recognition model is trained based on the Gaussian radial kernel function and the discriminant function of the optimal classification surface.

[0025] Furthermore, based on the prediction result of the discharge type identification model, the life of the metallized film capacitor is determined, including:

[0026] Determine the discharge type of the real-time discharge signal according to the prediction result of the discharge type recognition model;

[0027] If the real-time discharge signal is a discharge signal when the metallized film capacitor fails suddenly, the life of the metallized film capacitor is deemed to have ended;

[0028] If the real-time discharge signal is a discharge signal that causes loss of metal electrodes of the metallized film capacitor, the life of the metallized film capacitor is determined by the amount of capacitance loss of the metallized film capacitor during the operation cycle.

[0029] Furthermore, the life of the metallized film capacitor is determined by the capacitance loss of the metallized film capacitor during the operation cycle, including:

[0030] Extract the energy generated when the discharge signal causes the loss of the metal electrode of the metallized film capacitor;

[0031] Based on the energy, the accumulated energy of the discharge signal causing the metal electrode loss during the operation of the metal film capacitor is obtained;

[0032] Based on the accumulated energy of the discharge signal causing the metal electrode loss, the area of ​​the metal electrode loss is obtained;

[0033] Based on the area of ​​metal electrode loss, the capacitance loss of the metallized film capacitor during the operation cycle is obtained;

[0034] The life of the metallized film capacitor is determined based on the amount of capacitance loss of the metallized film capacitor.

[0035] In a second aspect, the present invention further provides a device for monitoring the life of a metallized film capacitor, comprising:

[0036] An acquisition module, used to acquire a first discharge signal and voltage information of a metallized film capacitor working in a pulse power system;

[0037] An identification module, used to identify a second discharge signal corresponding to the discharge type from the first discharge signal based on voltage information of the metallized film capacitor during discharge and the surface morphology of the metallized film;

[0038] An extraction module, used for extracting features of the second discharge signal based on multi-scale spatial energy distribution of wavelet transform to obtain a discharge signal feature database;

[0039] A construction module, used for constructing a discharge type recognition model based on a discharge signal feature database;

[0040] A prediction module, used for inputting the acquired real-time discharge signal of the metallized film capacitor into a discharge type recognition model for prediction;

[0041] The determination module is used to determine the life of the metallized film capacitor based on the prediction result of the discharge type identification model.

[0042] Further, after obtaining the first discharge signal and voltage information of the metallized film capacitor working in the pulse power system, it also includes:

[0043] The first discharge signal is subjected to noise reduction processing by using Mallat algorithm and discrete wavelet transform.

[0044] Furthermore, the extraction module is also used to:

[0045] Obtaining frequency domain characteristics of the second discharge signal through Fourier transform;

[0046] Calculate the energy of each frequency band corresponding to the multi-scale space based on wavelet transform to obtain a feature vector corresponding to the second discharge signal;

[0047] Based on the feature vector, a discharge signal feature database of metallized film capacitors is constructed.

[0048] Furthermore, the building blocks are also used to:

[0049] Based on the linear discriminant analysis method and the discharge signal feature database, a discharge type recognition model is constructed; the optimization objectives of the discharge type recognition model meet the following requirements:

[0050]

[0051] in, is the projection vector, is the between-class scatter matrix, is the intra-class scatter matrix, and T represents the transpose of the matrix.

[0052] Furthermore, after constructing the discharge type recognition model based on the discharge signal feature database, it also includes:

[0053] The discharge type recognition model is trained based on the Gaussian radial kernel function and the discriminant function of the optimal classification surface.

[0054] Furthermore, the module is determined to be further used for:

[0055] Determine the discharge type of the real-time discharge signal according to the prediction result of the discharge type recognition model;

[0056] If the real-time discharge signal is a discharge signal when the metallized film capacitor fails suddenly, the life of the metallized film capacitor is deemed to have ended;

[0057] If the real-time discharge signal is a discharge signal that causes loss of metal electrodes of the metallized film capacitor, the life of the metallized film capacitor is determined by the amount of capacitance loss of the metallized film capacitor during the operation cycle.

[0058] Furthermore, the life of the metallized film capacitor is determined by the capacitance loss of the metallized film capacitor during the operation cycle, including:

[0059] Extract the energy generated when the discharge signal causes the loss of the metal electrode of the metallized film capacitor;

[0060] Based on the energy, the accumulated energy of the discharge signal causing the metal electrode loss during the operation of the metal film capacitor is obtained;

[0061] Based on the accumulated energy of the discharge signal causing the metal electrode loss, the area of ​​the metal electrode loss is obtained;

[0062] Based on the area of ​​metal electrode loss, the capacitance loss of the metallized film capacitor during the operation cycle is obtained;

[0063] The life of the metallized film capacitor is determined based on the amount of capacitance loss of the metallized film capacitor.

[0064] In a third aspect, the present invention further provides an electronic device, comprising: a processor and a memory;

[0065] The processor is coupled to the memory;

[0066] The processor is used to read and execute the program or instructions stored in the memory, so that the device executes the method of the first aspect.

[0067] In a fourth aspect, the present invention further provides a computer-readable storage medium storing a computer program, which implements the method of the first aspect when executed by a processor.

[0068] In summary, the technical solution provided by this application has at least the following technical effects or advantages:

[0069] The technical solution of the present invention monitors the discharge signal generated during the operation cycle of the metallized film capacitor in the pulse power system, performs noise reduction processing on the discharge signal through the Mallat algorithm and discrete wavelet transform, extracts features of the processed discharge signal based on the multi-scale spatial energy distribution of the wavelet transform, establishes a discharge signal feature database of different discharge types, constructs a discharge type recognition model of the metallized film capacitor based on the feature database and machine learning method, and trains the model, uses the trained model to predict the real-time discharge signal generated during the operation cycle of the metallized film capacitor, determines the life of the metallized film capacitor according to the prediction result, and issues a warning. The technical solution of the present invention can realize online monitoring of the operation status of the metallized film capacitor and active warning of the expiration of the life, achieves the purpose of safety protection, has simple and reliable practical application value with high prediction accuracy, and is of great significance for monitoring the working status of the metallized film capacitor and maintaining the stable operation of the pulse power system.

[0070] Other features and advantages of the present disclosure will be described in the following description, and partly become apparent from the description, or be understood by implementing the present disclosure. The purpose and other advantages of the present disclosure can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0072] Figure 1 A schematic diagram of a process for monitoring the life of a metallized film capacitor in an embodiment of the present disclosure;

[0073] Figure 2 It is a structural schematic diagram of a device for monitoring the life of a metallized film capacitor in an embodiment of the present disclosure;

[0074] Figure 3 It is a further structural schematic diagram of a device for monitoring the life of a metallized film capacitor in an embodiment of the present disclosure;

[0075] Figure 4 is a schematic diagram of a typical discharge signal that causes metal electrode loss in an embodiment of the present disclosure;

[0076] Figure 5 A schematic diagram of a typical discharge signal when a metallized film capacitor fails suddenly in an embodiment of the present disclosure;

[0077] Figure 6 Schematic diagram of time domain and frequency domain characteristics of a discharge signal in an embodiment of the present disclosure;

[0078] Figure 7 A schematic diagram of the discharge type pattern recognition of the metallized film capacitor in the embodiment of the present disclosure;

[0079] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0080] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0081] Figure 1 The figure is a flow chart of a method for monitoring the life of a metallized film capacitor in an embodiment of the present disclosure. As shown in the figure, the method includes:

[0082] S101, obtaining a first discharge signal and voltage information of a metallized film capacitor working in a pulse power system;

[0083] For metallized film capacitors operating in pulse power systems, the first discharge signal and capacitor voltage information within a complete working cycle are measured. The first discharge signal refers to the discharge signal when partial discharge, self-healing discharge, edge flashover, capacitor sudden failure and other states occur, including signal waveform, frequency, amplitude and other information. The voltage information includes the voltage value during the capacitor discharge process.

[0084] The acquired first discharge signal is subjected to noise reduction processing by Mallat algorithm and discrete wavelet transform to remove interference signals such as leakage circuit from high-frequency power supply and Gaussian white noise from surrounding environment. The first discharge signal is reduced in order by using Mallat tower algorithm, and the first discharge signal is decomposed by inverse transform in discrete wavelet transform domain, and the first discharge signal is decomposed into approximate component and detail component at each scale. The approximate component represents the high scale of the signal, that is, the low-frequency information; the detail component represents the low scale of the signal, that is, the high-frequency information. For a signal containing noise, the main energy of the noise component is concentrated in the detail component of wavelet decomposition. The expression of discrete wavelet transform (DWT) is:

[0085]

[0086] in, is the wavelet basis function, S is the scale parameter, t is the displacement parameter, and j represents the imaginary unit. S0=2, τ0=1, j and k are both integers starting from 0. As the value of j increases, the scale parameter of the sub-wavelet increases by a multiple of 2, and the corresponding detected frequency value will be reduced by a multiple of 1 / 2.

[0087] After signal decomposition is completed, threshold processing is performed, that is, threshold processing is performed on each layer of decomposed coefficients to obtain the denoised wavelet coefficients. Wavelet reconstruction is performed based on the denoised wavelet coefficients to complete signal reconstruction and obtain the denoised first discharge signal.

[0088] S102, identifying a second discharge signal corresponding to a discharge type from the first discharge signal based on voltage information of the metallized film capacitor during discharge and the surface morphology of the metallized film;

[0089] The metallized film capacitor core was disassembled, and the location of the metallized film electrode loss was observed. The surface element distribution in the metal electrode loss area was further studied by combining a scanning electron microscope (SEM) with an energy dispersive spectrometer (EDS). If the Zn and Al elements at the location of the metallized film electrode loss are significantly reduced, and there is no aggregation of O elements. This indicates that the loss of the metal electrode is caused by discharge. Therefore, combining the voltage information and the change in surface morphology during the discharge process of the metallized film capacitor, the second discharge signal corresponding to the discharge type is identified from the first discharge signal. The characteristics of the surface morphology of the metallized film can be obtained through experimental analysis. The discharge types include: discharge signals that will not cause the loss of the metal electrode of the metallized film capacitor, discharge signals that cause the loss of the metal electrode of the metallized film capacitor, and discharge signals when the metallized film capacitor fails suddenly. The discharge signals that cause the loss of metal electrodes in metallized film capacitors include discharge signals when self-healing discharge, surface discharge (a type of local discharge), and edge flashover occur. At this time, the discharge process is accompanied by the evaporation of the metal electrode, which will cause the capacitance of the metal film capacitor to decrease. The metallized film capacitor directly loses its insulation performance after violent discharge. The discharge signal at this time is the discharge signal when the metallized film capacitor fails suddenly.

[0090] S103, extracting features of the second discharge signal based on multi-scale spatial energy distribution of wavelet transform to obtain a discharge signal feature database;

[0091] The frequency domain characteristics of the second discharge signal can be obtained by Fourier transform, and the expression of Fourier transform is:

[0092]

[0093] in, is the frequency domain signal; is the time domain signal; is the discharge duration, is the exponential function in complex form.

[0094] By processing the frequency domain signals, it is found that the frequency of the discharge signals that do not cause electrode loss is concentrated at 10 MHz, while the discharge signals that cause metal electrode loss have obvious low-frequency components. Therefore, the signal can be classified through the characteristics of the frequency domain.

[0095] The spectrum analysis of the second discharge signal is performed, and the energy of each frequency band corresponding to the multi-scale space is calculated based on wavelet transform, and the obtained energy is used as the feature vector. The j-layer wavelet decomposition of the signal f(t) can be expressed as:

[0096]

[0097] Among them, cA j and cD i They are the approximate signal of the j-th layer and the detail signal of the ith layer respectively. Decomposing the signal f(t) to the j-th layer can obtain the approximate signal of the j-th layer and the detail signals of each layer.

[0098] The wavelet coefficients of the signal are squared and summed to obtain the wavelet coefficient energy. The total energy of the signal after decomposition into j layers is expressed as:

[0099]

[0100] Where E is the total energy of the signal after decomposition at j layers, is the energy of the approximate signal after decomposing the jth layer, is the energy of the detail signal of the i-th layer.

[0101] Select the energy of the approximate signal of the jth layer and the detail signal of each layer as features, and normalize them in scale order to construct the feature vector F:

[0102]

[0103] in, is the energy of the approximate signal after decomposing the jth layer, is the energy of the detail signal of the i-th layer.

[0104] Based on the characteristic vector data representing the energy of the discharge signal extracted by wavelet transform, a discharge signal characteristic database of metallized film capacitors is established. The discharge signal is decomposed into the third layer (j=3) using the sym8 wavelet function, and the wavelet coefficients of the three levels are obtained respectively. The energy values ​​of each layer are obtained using these wavelet coefficients, and these energy values ​​are used as the characteristic data of the discharge signal. For the case of j=3, the constructed characteristic vector is three detail components and one approximate component, which is the characteristic vector data for constructing the discharge signal characteristic database of metallized film capacitors. The characteristic data in the characteristic database also includes three types of data labels: discharge signals that will not cause metal electrode loss, discharge signals that cause metal electrode loss, and discharge signals when metallized film capacitors fail suddenly, that is, three types of data labels corresponding to the discharge types.

[0105] S104, constructing a discharge type recognition model based on a discharge signal feature database;

[0106] Based on the Linear Discriminant Analysis (LDA) method and the discharge signal feature database, a discharge type pattern recognition model corresponding to the metallized film capacitor is constructed, namely the discharge type recognition model. The LDA method is used to reduce the dimension of the feature vector in the discharge signal feature database, and the labeled high-dimensional samples are projected into a one-dimensional space. Points of the same category will be closer in the projected space. When the "minimum variance within the class and maximum variance between classes" are met, a decision boundary is obtained so that the discharge type pattern has the best separability in the space.

[0107] Based on the above metallized film capacitor discharge signal feature database, the entire sample space is divided into three categories, namely 1, 2, and 3. N1, N2, and N3 represent the number of samples in category 1, 2, and 3, respectively; the samples are x , set z to the mapped coordinate (i.e. the projected coordinate).

[0108] The sample data x Projecting onto the w vector (setting the modulus of w to 1), we have:

[0109]

[0110] Among them, zi is the mapped coordinate, w is the modulus direction vector, and xi is the sample data.

[0111] The intra-class scatter matrix is ​​defined as , the inter-class scatter matrix is , the global divergence matrix is :

[0112] The expression is:

[0113]

[0114] The expression is:

[0115]

[0116] The expression is:

[0117]

[0118] Among them, µ is the mean of the three categories of samples, µ1, µ1, and µ1 are the means of the three categories of samples, and N1, N2, and N3 represent the number of samples in categories 1, 2, and 3, respectively.

[0119] In order to obtain a good classification model, the distance within the class should be small and the distance between classes should be large. Based on this idea, the optimization goal is adopted:

[0120]

[0121] in, is the projection vector (set the modulus to 1), is the between-class scatter matrix, is the intra-class scatter matrix, and T represents the transpose of the matrix.

[0122] The final decision boundary can be obtained by satisfying this optimization goal.

[0123] The feature vector after dimension reduction is used as the input vector of the discharge type recognition model to train the discharge type recognition model. The Gaussian radial kernel function is used to map the input vector from one-dimensional space to high-dimensional space. The Gaussian radial kernel function is expressed as:

[0124]

[0125] Where: x is the sample data, z is the center of the kernel function (i.e., the support vector); σ is a constant parameter, and e is the base of the natural logarithm.

[0126] Based on the discriminant function of the optimal classification surface, linear separability is achieved in high-dimensional space. The discriminant function of the optimal classification surface is:

[0127]

[0128] Where: n is the index set of Lagrange multipliers corresponding to the training set; is the connection weight, when it is non-zero, the corresponding ( ) is the support vector; is bias; is a symbolic function.

[0129] An optimal classification plane with the maximum interval between different samples is constructed, the training of the discharge type recognition model corresponding to the metallized film capacitor is completed, and the accuracy of the model prediction is verified.

[0130] Through the discharge type recognition model, the intelligent recognition and differentiation of different discharge types of the discharge signal of the metallized film capacitor operating in the pulse power system can be achieved.

[0131] S105, inputting the acquired real-time discharge signal of the metallized film capacitor into a discharge type recognition model for prediction;

[0132] A real-time discharge signal of a metallized film capacitor working in a pulse power system is obtained, and the specific information contained in the real-time discharge signal is the same as the first discharge signal. After the real-time discharge signal is subjected to the above-mentioned noise reduction processing and feature extraction, the feature data corresponding to the real-time discharge signal is input into the discharge type recognition model for prediction to obtain a prediction result.

[0133] S106: Determine the life of the metallized film capacitor based on the prediction result of the discharge type identification model.

[0134] According to the prediction result of the discharge type identification model, the discharge type of the identified real-time discharge signal is judged, and then the life of the metallized film capacitor is determined.

[0135] If the real-time discharge signal is a discharge signal of a sudden failure of the metallized film capacitor, that is, the metallized film capacitor loses its insulation performance after severe discharge, the life of the metallized film capacitor is deemed to have expired and an alarm signal is issued.

[0136] If the real-time discharge signal is a discharge signal that causes the loss of the metal electrode of the metallized film capacitor, that is, the discharge process is accompanied by the evaporation of the metal electrode, which will cause the capacitance of the metal film capacitor to decrease, then the energy generated by the discharge signal that causes the loss of the metal electrode of the metallized film capacitor is extracted by calculation, and the calculation formula is:

[0137]

[0138] in, The energy corresponding to the discharge that causes electrode loss in the metallized film capacitor to be tested; is the discharge duration; is the voltage of the capacitor during the discharge process; It is the pulse current generated during discharge.

[0139] When calculating the discharge signal that causes the metal electrode loss of the metallized film capacitor, the cumulative energy of the discharge signal that causes the metal electrode loss during the operation of the metal film capacitor is calculated as follows:

[0140]

[0141] in, It is the accumulated energy of discharge signals that cause electrode loss n times in a film capacitor during a period of operation.

[0142] Based on the accumulated energy of the discharge signal that causes the metal electrode loss, the area of ​​the metal electrode loss is calculated using the following formula:

[0143]

[0144] Among them, the thickness d of the metal electrode is a constant, k is the proportional coefficient, is the area lost by the metal electrode.

[0145] Based on the area of ​​metal electrode loss, the capacitance loss of the metallized film capacitor during the operation cycle is calculated using the following formula:

[0146]

[0147] in, is the capacitance loss, is the area lost by the metal electrode, is the relative dielectric constant, Dielectric constant of vacuum.

[0148] When the capacitance of the metallized film capacitor drops by 5%, it is considered that the life of the capacitor has ended and the system will issue an alarm.

[0149] According to the above processing of the prediction results of the discharge type identification model, the demand for online monitoring of the operating status and remaining life of the metallized film capacitor is realized.

[0150] In summary, the technical solutions in the embodiments of the present application have at least the following technical effects or advantages:

[0151] The technical solution of the present invention monitors the discharge signal generated during the operation cycle of the metallized film capacitor in the pulse power system, performs noise reduction processing on the discharge signal through the Mallat algorithm and discrete wavelet transform, extracts features of the processed discharge signal based on the multi-scale spatial energy distribution of the wavelet transform, establishes a discharge signal feature database of different discharge types, constructs a discharge type recognition model of the metallized film capacitor based on the feature database and machine learning method, and trains the model, uses the trained model to predict the real-time discharge signal generated during the operation cycle of the metallized film capacitor, determines the life of the metallized film capacitor according to the prediction result, and issues a warning. The technical solution of the present invention can realize online monitoring of the operation status of the metallized film capacitor and active warning of the expiration of the life, achieves the purpose of safety protection, has simple and reliable practical application value with high prediction accuracy, and is of great significance for monitoring the working status of the metallized film capacitor and maintaining the stable operation of the pulse power system.

[0152] Figure 2 This is a schematic diagram of the structure of a device for monitoring the life of a metallized film capacitor provided in an embodiment of the present disclosure. As shown in the figure, the device includes:

[0153] An acquisition module, used to acquire a first discharge signal and voltage information of a metallized film capacitor working in a pulse power system;

[0154] an identification module, configured to identify a second discharge signal corresponding to a discharge type from the first discharge signal based on voltage information of a discharge process of the metallized film capacitor and a surface morphology of the metallized film;

[0155] An extraction module, configured to extract features of the second discharge signal based on multi-scale spatial energy distribution of wavelet transform to obtain a discharge signal feature database;

[0156] A construction module, used to construct a discharge type recognition model based on the discharge signal feature database;

[0157] A prediction module, used for inputting the acquired real-time discharge signal of the metallized film capacitor into the discharge type recognition model for prediction;

[0158] The determination module is used to determine the life of the metallized film capacitor based on the prediction result of the discharge type identification model.

[0159] It should be noted that, for the sake of convenience, Figure 2 For example, only the main modules of the monitoring device structure of the metallized film capacitor life are shown. In practical applications, the system may also include modules or components not shown in the figure; the system is not limited to the above module structure, but may also be other module structures that implement the above method embodiments.

[0160] like Figure 3 As shown, a monitoring device for the life of a metallized film capacitor also includes: a capacitor Cs to be tested, a programmable DC power supply, a sampling resistor, a high-frequency current mutual inductance coil (HFCT), an oscilloscope, and a monitoring host; the programmable DC power supply simulates the working environment of the pulse capacitor, the sampling resistor collects the pulse current signal generated when the metallized film capacitor is discharged, and the HFCT collects the voltage signal generated when the metallized film capacitor is discharged, and the two discharge signals and the voltage signal of the film capacitor during discharge are input into the oscilloscope, and the oscilloscope inputs the collected signals into the data processing module, performs wavelet noise reduction processing through the Mallat algorithm and extracts feature information through discrete wavelet transform, constructs a discharge signal feature information data set, and transmits the data set to a server, and the server uses a metallized film capacitor discharge type recognition model to predict the discharge feature information data set, and identifies and extracts the discharge signal that will cause the loss of the metallized film surface electrode from all discharge signals, that is, the discharge process is accompanied by the evaporation of the metal electrode; and the discharge signal when the metallized film capacitor fails suddenly, that is, the metallized film capacitor loses its insulation performance after severe discharge.

[0161] Exemplarily, by disassembling the film capacitor core and observing the location of the metallized film electrode loss, combining the capacitor voltage information of the discharge process with the surface morphology of the metallized film after the test, it is determined that the discharge signals that may cause the metal electrode loss include self-healing, surface discharge and edge flashover.

[0162] For example, Figure 4 The discharge signal that causes metal electrode loss is shown in the figure. The results show that the discharge signal waveform is a multi-cycle decaying sinusoidal signal with a duration of 0.5-1µs. The voltage amplitude of the discharge signal can reach 30V, and it will cause a voltage drop of hundreds of volts on the capacitor, resulting in the loss of the surface metal electrode and the decrease of the capacitance of the film capacitor.

[0163] For example, Figure 5 The figure shows the discharge signal that causes the sudden failure of metallized film capacitors. Along with the violent discharge, the capacitor voltage continues to drop to zero, the insulation performance is lost, and the life span ends.

[0164] For example, Figure 6 As shown in the figure, the characteristic vectors of the above discharge signals in the time domain and frequency domain are extracted. The time domain waveforms of the discharge signals that cause metal electrode loss are all reflected as multi-cycle attenuated sinusoidal signals. Through the processing of the frequency domain signals, it is found that the frequency of the discharge signals that do not cause electrode loss is concentrated at 10MHz, while the discharge signals that cause metal electrode loss have obvious low-frequency components, and as the discharge amplitude increases, the amplitude of the low-frequency signal gradually increases. Therefore, the characteristics of the frequency domain can be used to classify the signal.

[0165] For example, Figure 7 As shown in the figure, the use of machine learning-based discharge type pattern recognition has achieved intelligent recognition and differentiation of three types of discharge signals: discharge signals that will not cause metal electrode loss, discharge signals that cause metal electrode loss, and discharge signals when metallized film capacitors fail suddenly. The accuracy of the model has been verified through test data sets.

[0166] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure.

[0167] like Figure 8 As shown, the electronic device includes: a processor and a memory;

[0168] The processor is used to read and execute programs and instructions stored in the memory, so that the electronic device executes the above method embodiment.

[0169] It should be noted that, for the sake of convenience, Figure 8 Only the main components of the electronic device are shown. In practical applications, the electronic device may also include components or assemblies not shown in the figure.

[0170] The present disclosure also provides a computer-readable storage medium storing a program or instruction. When a computer reads and executes the program or instruction, the computer executes the above method embodiment.

[0171] Although the present disclosure has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. A method for monitoring the life of a metallized film capacitor, characterized in that: include: Acquire a first discharge signal and voltage information of a metallized film capacitor operating in a pulse power system; Based on the voltage information of the metallized film capacitor during the discharge process and the surface morphology of the metallized film, identifying a second discharge signal corresponding to the discharge type from the first discharge signal; Based on the multi-scale spatial energy distribution of wavelet transform, feature extraction is performed on the second discharge signal to obtain a discharge signal feature database; Based on the discharge signal feature database, construct a discharge type recognition model; The real-time discharge signal of the metallized film capacitor obtained is input into the discharge type recognition model for prediction; The life of the metallized film capacitor is determined based on the prediction result of the discharge type identification model.

2. The method for monitoring the life of a metallized film capacitor according to claim 1, characterized in that: After obtaining the first discharge signal and voltage information of the metallized film capacitor working in the pulse power system, the method further includes: The first discharge signal is subjected to noise reduction processing by using Mallat algorithm and discrete wavelet transform.

3. The method for monitoring the life of a metallized film capacitor according to claim 1, characterized in that: The multi-scale spatial energy distribution based on wavelet transform performs feature extraction on the second discharge signal to obtain a discharge signal feature database, including: Obtaining frequency domain characteristics of the second discharge signal through Fourier transform; Calculate the energy of each frequency band corresponding to the multi-scale space based on wavelet transform to obtain a feature vector corresponding to the second discharge signal; Based on the characteristic vector, a discharge signal characteristic database of metallized film capacitors is constructed.

4. The method for monitoring the life of a metallized film capacitor according to claim 1, characterized in that: The step of constructing a discharge type recognition model based on the discharge signal feature database includes: Based on the linear discriminant analysis method and the discharge signal feature database, a discharge type recognition model is constructed; the optimization target of the discharge type recognition model satisfies: in, W represents the projection matrix, is the projection vector, is the between-class scatter matrix, is the intra-class scatter matrix, and T represents the transpose of the matrix.

5. The method for monitoring the life of a metallized film capacitor according to claim 4, characterized in that: After constructing the discharge type recognition model based on the discharge signal feature database, the method further includes: The discharge type recognition model is trained based on the Gaussian radial kernel function and the discriminant function of the optimal classification surface.

6. The method for monitoring the life of a metallized film capacitor according to any one of claims 1 to 5, characterized in that: The method of determining the life of the metallized film capacitor based on the prediction result of the discharge type identification model includes: Determining the discharge type of the real-time discharge signal according to the prediction result of the discharge type identification model; If the real-time discharge signal is a discharge signal when the metallized film capacitor fails suddenly, the life of the metallized film capacitor is deemed to have ended; If the real-time discharge signal is a discharge signal that causes loss of metal electrodes of the metallized film capacitor, the life of the metallized film capacitor is determined by the amount of capacitance loss of the metallized film capacitor during the operation cycle.

7. The method for monitoring the life of a metallized film capacitor according to claim 6, characterized in that: The method of determining the life of the metallized film capacitor by the capacitance loss of the metallized film capacitor during the operation cycle includes: Extract the energy generated when the discharge signal causes the loss of the metal electrode of the metallized film capacitor; Based on the energy, the accumulated energy of the discharge signal causing the metal electrode loss during the operation of the metal film capacitor is obtained; Based on the accumulated energy of the discharge signal causing the metal electrode loss, the area of ​​the metal electrode loss is obtained; Based on the area of ​​metal electrode loss, the capacitance loss of the metallized film capacitor during the operation cycle is obtained; The life of the metallized film capacitor is determined based on the loss of capacitance of the metallized film capacitor.

8. A device for monitoring the life of a metallized film capacitor, characterized in that: include: An acquisition module, used to acquire a first discharge signal and voltage information of a metallized film capacitor working in a pulse power system; an identification module, configured to identify a second discharge signal corresponding to a discharge type from the first discharge signal based on voltage information of a discharge process of the metallized film capacitor and a surface morphology of the metallized film; An extraction module, configured to extract features of the second discharge signal based on multi-scale spatial energy distribution of wavelet transform to obtain a discharge signal feature database; A construction module, used to construct a discharge type recognition model based on the discharge signal feature database; A prediction module, used for inputting the acquired real-time discharge signal of the metallized film capacitor into the discharge type recognition model for prediction; The determination module is used to determine the life of the metallized film capacitor based on the prediction result of the discharge type identification model.

9. The device for monitoring the life of a metallized film capacitor according to claim 8, characterized in that: After obtaining the first discharge signal and voltage information of the metallized film capacitor working in the pulse power system, the method further includes: The first discharge signal is subjected to noise reduction processing by using Mallat algorithm and discrete wavelet transform.

10. The device for monitoring the life of a metallized film capacitor according to claim 8, characterized in that: The extraction module is further used for: Obtaining frequency domain characteristics of the second discharge signal through Fourier transform; Calculate the energy of each frequency band corresponding to the multi-scale space based on wavelet transform to obtain a feature vector corresponding to the second discharge signal; Based on the characteristic vector, a discharge signal characteristic database of metallized film capacitors is constructed.

11. The device for monitoring the life of a metallized film capacitor according to claim 8, characterized in that: The building blocks are also used to: Based on the linear discriminant analysis method and the discharge signal feature database, a discharge type recognition model is constructed; the optimization target of the discharge type recognition model satisfies: in, W represents the projection matrix, is the projection vector, is the between-class scatter matrix, is the intra-class scatter matrix, and T represents the transpose of the matrix.

12. The device for monitoring the life of a metallized film capacitor according to claim 8, characterized in that: After constructing the discharge type recognition model based on the discharge signal feature database, the method further includes: The discharge type recognition model is trained based on the Gaussian radial kernel function and the discriminant function of the optimal classification surface.

13. The device for monitoring the life of a metallized film capacitor according to any one of claims 8 to 12, characterized in that: The determining module is further used for: Determining the discharge type of the real-time discharge signal according to the prediction result of the discharge type identification model; If the real-time discharge signal is a discharge signal when the metallized film capacitor fails suddenly, the life of the metallized film capacitor is deemed to have ended; If the real-time discharge signal is a discharge signal that causes loss of metal electrodes of the metallized film capacitor, the life of the metallized film capacitor is determined by the amount of capacitance loss of the metallized film capacitor during the operation cycle.

14. The device for monitoring the life of a metallized film capacitor according to claim 13, characterized in that: The method of determining the life of the metallized film capacitor by the capacitance loss of the metallized film capacitor during the operation cycle includes: Extract the energy generated when the discharge signal causes the loss of the metal electrode of the metallized film capacitor; Based on the energy, the accumulated energy of the discharge signal causing the metal electrode loss during the operation of the metal film capacitor is obtained; Based on the accumulated energy of the discharge signal causing the metal electrode loss, the area of ​​the metal electrode loss is obtained; Based on the area of ​​metal electrode loss, the capacitance loss of the metallized film capacitor during the operation cycle is obtained; The life of the metallized film capacitor is determined based on the loss of capacitance of the metallized film capacitor.

15. An electronic device, characterized in that: include: Processor and memory; The processor is coupled to a memory; The processor is used to read and execute the program or instruction stored in the memory, so that the device executes the method according to any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that: A computer program is stored, and when the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images