Method and system for state monitoring and fault diagnosis of direct-current storage battery

Abstract

The invention discloses a DC storage battery state monitoring and fault diagnosis method and system, and the method comprises the steps: building a cable line fault positioning model, and determining a fault point. And fault information is extracted according to the wavelet transform model, and fault feature indexes are calculated. And establishing an optimized neural network algorithm to locate the cable line fault, obtaining a non-linear relationship between the fault sample and the fault position, and outputting the fault position. According to the direct-current storage battery state monitoring and fault diagnosis method and system provided by the invention, the real-time positioning strategy of the BP neural network cable line fault optimized based on the firefly group algorithm and fusion of stationary wavelet transform can accurately predict the fault position by determining key indexes. The designed fault prediction strategy can reduce the average value of the fault position offset rate by about 35.758%, and has a wide application prospect.

Description

Technical Field

[0001] The present invention relates to the technical field of cable line fault location, and particularly to a method and system for monitoring the state and diagnosing faults of a DC battery. Background Art

[0002] With the rapid development of the power system, the power grid is more closely connected, which puts forward higher requirements for the efficient and stable operation of the power transmission line cables. Different from overhead lines, cable lines are generally laid underground and are greatly affected by factors such as harsh underground environments, external forces, moisture, and chemical pollution. During long-term operation of cable lines, the probability of failures surges, which will greatly increase the workload of maintenance and defect elimination. Quickly, reliably, timely, and accurately determining the fault location of cable lines is of great significance for improving the efficiency of line maintenance work, reducing the blindness of fault line patrols, shortening the repair and outage time, effectively reducing the congestion of power transmission channels, and reducing the duration of power limit violations.

[0003] Existing methods for cable fault location can be divided into offline location and online location. Among them, offline fault location mainly determines the fault location by calculating the arrival time difference between the initial signal and the reflected signal, and has advantages such as simple principle and convenient operation [8]. However, offline fault location needs to be tested in a power-off environment, and the time cost of fault location is relatively high, which is difficult to meet the development needs of the power system. Therefore, researchers have gradually begun to pay attention to online location methods for cable faults, and online location methods include impedance method, traveling wave method, etc. The impedance method realizes fault location by establishing the corresponding relationship between the steady-state phasor and the fault distance, and has advantages such as low cost and strong robustness. The prior art has realized cable aging and fault detection based on broadband radio frequency impedance testing by using the impedance method, and has a good cable fault location effect. The prior art improves the traditional impedance method and applies it to the fault location prediction of underground cables in coal mines. The traveling wave method uses the traveling wave signal propagated in the cable to determine the cable fault location, and short pulse voltage signals need to be transmitted at both ends of the cable, and the fault distance is calculated by measuring the time taken by the cable traveling wave signal to pass through, and has advantages such as high accuracy and wide measurement range. The prior art proposes a cable fault location strategy based on the variational mode decomposition strategy, and uses the traveling wave method to obtain the wavefront of the fault wave for feature analysis. The prior art proposes a double-ended fault location method for flexible DC cables based on improved local mean decomposition, and uses the double-ended traveling wave fault ranging principle to determine the fault location. The above methods have achieved good results, but still need to be improved and optimized by means of advanced algorithms in terms of fault location accuracy and efficiency.

[0004] Wavelet transform can characterize the mutation characteristics in signal transmission. Compared with traditional transformation methods such as Fourier transform, wavelet transform has advantages in the resolution of both frequency domain and time domain, and can provide more comprehensive signal information. Discrete wavelet transform requires convolution and downsampling operations for each stage, consuming huge computing resources. In addition, discrete wavelet transform has disadvantages such as poor directionality and lack of phase information, and its application effect is not significant. Stationary wavelet transform improves the decomposition process, keeps the approximation coefficients and detail coefficients consistent with the length of the original signal, can effectively ensure the integrity of information, and has been widely used in the identification and extraction of cable line fault characteristics.

[0005] With the rapid development of artificial intelligence technology, the prediction and analysis of the parameters of any nonlinear system can be realized by virtue of the outstanding advantages of neural networks in pattern recognition and parameter estimation. BP neural network has good generalization ability, anti-interference characteristics, robustness and fault tolerance ability, and is suitable for the field of cable fault identification with complex working environment characteristics. Since the weights and thresholds of the BP neural network have a great influence on its convergence performance, using swarm intelligence optimization algorithm to determine its specific parameters by optimization can significantly improve the performance of the BP neural network. Summary of the Invention

[0006] In view of the above problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by the present invention is that although the prior art has achieved good results, advanced algorithms are still needed for improvement and optimization in terms of fault location accuracy and efficiency.

[0008] To solve the above technical problem, the present invention provides the following technical solution: A method for monitoring the state and diagnosing faults of a DC battery, including: establishing a cable line fault location model to determine the fault point.

[0009] Extract fault information according to the wavelet transform model and calculate fault characteristic indexes.

[0010] Establish an optimized neural network algorithm to locate the cable line fault, obtain the nonlinear relationship between the fault sample and the fault location, and output the fault location.

[0011] As a preferred solution of the method for monitoring the state and diagnosing faults of a DC battery according to the present invention, wherein: the establishment of the cable line fault location model includes that the types of cable line faults are divided into three situations: short circuit, open circuit and poor contact.

[0012] The voltage and current parameters of the cable line are characterized by the cable line parameters, and the cable line fault location is realized according to the change amount of the voltage and current. In case of a fault, the relationship between the voltage, current of different segments of the cable and the cable line parameters is:

[0013]

[0014] Among them, R represents the resistance of the cable line per unit length. L represents the inductance of the cable line per unit length. C represents the capacitance of the cable line per unit length. G represents the conductance of the cable line per unit length. x represents the length (km) from the fault information measurement end to the fault information receiving end. v represents the voltage of the cable line. i represents the current of the cable line.

[0015] The voltage and current transfer impedance z c and the corresponding transfer constant κ are expressed as:

[0016]

[0017] The voltage and current at the fault information receiving end are represented by V r and I r respectively. When the distance from the fault information measurement end to the fault information receiving end is x, the voltage V x and current I x at the fault information measurement end are:

[0018]

[0019] The voltage and current at the fault information sending end are represented by V s and I s respectively. When the total length of the cable line is l, the voltage V x and current I x at the fault information measurement end are:

[0020]

[0021] When a fault occurs at any point on the cable line, the cable line is regarded as being divided into two similar parts by the fault point, and these two parts are regarded as normal cables without faults. When a fault occurs, the fault location is determined according to the voltage and current data of the fault information measurement section and the fault information receiving end.

[0022] As a preferred solution of the DC battery state monitoring and fault diagnosis method described in the present invention, among them: the determination of the fault point includes that the voltage at the fault point of the cable line is represented by V f respectively. When the distance from the position of the fault point to the fault receiving end is Q, V f is:

[0023] V f =V r cosh(κQ)+z c I r sinh(κQ)

[0024] =V scosh[κ(l - Q)] + z c I s sinh[κ(l - Q)]

[0025] Based on the voltage and current values at the cable line fault point and the corresponding values at the fault information receiving and sending ends, determine the location of the fault point, expressed as:

[0026]

[0027] Among them, Q represents the location of the fault point.

[0028] As a preferred scheme of the DC battery state monitoring and fault diagnosis method described in the present invention, where: the extraction of fault information according to the wavelet transform model includes analyzing the wavelet transform results of the cable line operation data from different dimensions using wavelet transform, extracting the characteristic data of the faulty cable line, and realizing fault location and ranging.

[0029] When the square-integrable function The corresponding Fourier transform Satisfies the following formula, this square-integrable function is called Base wavelet:

[0030]

[0031] The base wavelet of the square-integrable function Is translated and expressed as:

[0032]

[0033] Among them, a represents the scaling factor in the translation process. b represents the translation factor in the translation process.

[0034] Perform power-level discretization and uniform discretization processing on the scaling factor a and the translation factor b:

[0035]

[0036] Among them, a0 > 1, b0 > 0, both are constants.

[0037] According to the definition of wavelet transform, we get:

[0038]

[0039] The voltage and current change characteristics of the cable line are represented by f(t), f(t) ∈ L 2 (R), the discrete wavelet transform of f(t) is expressed as:

[0040]

[0041] Among them, denote the conjugate complex number of

[0042] The stationary wavelet transform is used to extract the cable line fault information. Downsampling is cancelled during the decomposition process, and upsampling is cancelled during the reconstruction part. The stationary wavelet transform discretizes the dilation factor a while ensuring the continuity of the translation factor b, which is expressed as:

[0043]

[0044] The formula of the stationary wavelet transform for the voltage and current change characteristics f(t) of the cable is expressed as:

[0045]

[0046] where i represents the decomposition degree, and j represents the separation degree. λ ij (t) represents the dilation function. c represents the approximation coefficient of the stationary wavelet transform, and d represents the detail coefficient.

[0047] The stationary wavelet transform obtains the cable line fault signal. The low-frequency approximation coefficient c is used to characterize the fundamental frequency component of the low-frequency part, and the high-frequency detail coefficient d is used to characterize the transient characteristics of the high-frequency part. During the stationary wavelet transform process, the approximation coefficient and detail coefficient of each layer are expressed as:

[0048]

[0049] where H represents the low-pass filter corresponding to the i-th layer, and G respectively represents the low-pass and high-pass filters corresponding to the i-th layer. n represents the sampling window serial number. N represents the total number of samplings.

[0050] As a preferred scheme of the DC battery state monitoring and fault diagnosis method described in the present invention, among them: the calculated fault characteristic indexes include equally dividing each sub-band according to the working frequency, and extracting characteristics for each time period. The selected characteristics include energy value, average value, variance and entropy value.

[0051] The energy value is used to characterize the energy distribution characteristics of the actually measured propagation data of the cable line. For the q-th sub-band, the energy value is expressed as:

[0052]

[0053] where N t represents the number of sampling points.

[0054] The average value is used to characterize the centralized characteristics of the actually measured cable parameter propagation data so as to reflect the stability degree of the object, and is expressed as:

[0055]

[0056] The variance is used to characterize the degree of dispersion of the measured cable parameter propagation data, and is expressed as:

[0057]

[0058] The entropy value is correlated with the degree of chaos of the data, and is used to characterize the chaotic characteristics of the energy sequence of the measured cable parameter propagation data, and is expressed as:

[0059]

[0060] Among them, E represents the total energy value of the sub-band.

[0061] To avoid the influence of eigenvalue dimension differences on the results, the above four characteristic parameters are normalized, and are expressed as:

[0062]

[0063] Among them, x c represents the value after normalization of the data sequence, x r represents the initial value, x min represents the minimum value, x max represents the maximum value.

[0064] As a preferred solution of the DC battery state monitoring and fault diagnosis method described in the present invention, wherein: the establishment of an optimized neural network algorithm for cable line fault location includes that the mapping relationship of the BP neural network is expressed as:

[0065]

[0066] Among them, c j represents the weight value from the hidden layer to the output layer. b j represents the output of the hidden layer node. ε represents the threshold of the output layer.

[0067] The correlation degree between the outputs of the hidden layer nodes is calculated using the tansig activation function, and the correlation degree between the c i th node and the c j th node is expressed as:

[0068]

[0069] The output b j of the hidden layer node is expressed as:

[0070]

[0071] The error between the c i th node and the c j th node in the hidden layer is expressed as:

[0072]

[0073] Among them, w ij represents the weight between the input layer and the hidden layer. θ j represents the threshold of the hidden layer nodes.

[0074] As a preferred solution of the DC battery state monitoring and fault diagnosis method described in the present invention, wherein: the output fault location includes using the firefly swarm optimization algorithm to optimize and solve the weights and thresholds of the BP neural network, making up for the random defects in the selection of connection weights and thresholds of the BP neural network. Assume that the number of fireflies is N, and the positions and objective functions corresponding to the i-th firefly are (x i , y i ) and f(x i , y i ), and the position corresponding to the luciferin value is T i . The decision radius of each firefly is expressed as:

[0075]

[0076] Among them, R s represents the perception radius. β represents the control parameter. N i (t) is expressed as:

[0077]

[0078] Among them, X i (t) represents the position of the i-th firefly in the t-th generation, and I i (t) represents the luciferin value of the i-th firefly in the t-th generation.

[0079] The position update formula of each firefly is:

[0080]

[0081] Among them, s represents the moving step size.

[0082] After globally optimizing the weights and thresholds of the BP neural network using the firefly swarm algorithm, a transmission line cable fault location strategy with better performance in convergence ability and prediction accuracy is obtained.

[0083] A DC battery state monitoring and fault diagnosis system, characterized in that it includes,

[0084] A model establishment module, which establishes a cable line fault location model and determines the fault point.

[0085] A feature calculation module, which extracts fault information according to the wavelet transform model and calculates fault feature indexes.

[0086] Optimization module, which establishes an optimized neural network algorithm to locate cable line faults, obtains the non-linear relationship between fault samples and fault locations, and outputs the fault locations.

[0087] A computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the method described above are implemented.

[0088] A computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described above are implemented.

[0089] Advantages of the present invention: The real-time cable line fault location strategy that combines stationary wavelet transform and BP neural network optimized by firefly swarm algorithm proposed by the present invention can determine key indicators and accurately predict fault locations. The designed fault prediction strategy can reduce the average value of the fault location deviation rate by about 35.758%, and has broad application prospects. Description of the Drawings

[0090] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. Among them:

[0091] Figure 1 It is the overall flowchart of a method and system for monitoring the state and diagnosing faults of a DC battery provided by the first embodiment of the present invention.

[0092] Figure 2 It is the transmission characteristic diagram of the cable data measurement point position, voltage and current for a method and system for monitoring the state and diagnosing faults of a DC battery provided by the first embodiment of the present invention.

[0093] Figure 3 It is the schematic diagram of the stationary wavelet transform decomposition for a method and system for monitoring the state and diagnosing faults of a DC battery provided by the first embodiment of the present invention.

[0094] Figure 4 It is the on-site survey map of the cable for a method and system for monitoring the state and diagnosing faults of a DC battery provided by the second embodiment of the present invention.

[0095] Figure 5 It is the sample data diagram of the stationary wavelet transform for a method and system for monitoring the state and diagnosing faults of a DC battery provided by the second embodiment of the present invention.

[0096] Figure 6 The curve graph showing the variation of the training error of a method and system for monitoring the state and diagnosing faults of a DC battery provided in the second embodiment of the present invention with the number of iterations.

[0097] Figure 7 The trend graph of the offset rate of a method and system for monitoring the state and diagnosing faults of a DC battery provided in the second embodiment of the present invention.

[0098] Figure 8 The average offset rate graph of the test samples of a method and system for monitoring the state and diagnosing faults of a DC battery provided in the second embodiment of the present invention. Detailed implementation manners

[0099] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following provides a detailed description of the specific implementation manners of the present invention with reference to the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0100] Embodiment 1, referring to Figures 1 to 3 , which is an embodiment of the present invention, provides a method for monitoring the state and diagnosing faults of a DC battery, including:

[0101] S1: Establish a cable line fault location model to determine the fault point.

[0102] It should be noted that in order to accurately locate the cable line fault and distinguish the fault type, it is necessary to ensure that the cable line is a power transmission line with uniform texture, and the conductance, capacitance, resistance, and inductance parameters distributed along it do not change at any position of the cable. When a fault occurs at a certain point on the cable line, the entire cable can be regarded as composed of two normal cables divided by the fault point.

[0103] The types of cable line faults can be divided into three situations: short circuit, open circuit, and poor contact. A short circuit means an unrestricted electrical connection occurs between two circuit elements, resulting in a significant decrease in voltage, causing phenomena such as circuit heating, breakdown, and burnout, which is a relatively common fault type. An open circuit means that a certain position on the cable line is disconnected, and at this time, the cable cannot transmit electrical energy, forming an open loop in the circuit. Poor contact means problems such as loose, oxidized contact between wires and cables, resulting in an increase in the resistance of the cable and affecting the electrical energy transmission efficiency.

[0104] The voltage and current parameters of the cable line are characterized by the cable line parameters, and the cable line fault location is realized according to the change amounts of voltage and current. In the case of a fault, the relationship between the voltage, current of different segments of the cable and the cable line parameters is:

[0105]

[0106] Among them, R represents the resistance of the cable line per unit length. L represents the inductance of the cable line per unit length. C represents the capacitance of the cable line per unit length. G represents the conductance of the cable line per unit length. x represents the length (km) from the fault information measurement end to the fault information receiving end. v represents the voltage of the cable line. i represents the current of the cable line.

[0107] The distance between the measurement end and the data receiving end of the voltage and current fault information of the cable line, as well as the transmission characteristics of the voltage and current, are as Figure 2 shown.

[0108] The voltage and current transfer impedance z c and the corresponding transfer constant κ are expressed as:

[0109]

[0110] According to Figure 1 , the voltage and current at the fault information sending end are represented by V s and I s respectively. When the total length of the cable line is l, the voltage V x and current I x at the fault information measurement end are:

[0111]

[0112] The voltage and current at the fault information sending end are represented by V s and I s respectively. When the total length of the cable line is l, the voltage V x and current I x at the fault information measurement end are:

[0113]

[0114] When a fault occurs at any point on the cable line, the cable line is regarded as being divided into two similar parts by the fault point, and these two parts are regarded as normal cables without faults. When a fault occurs, the fault location is determined according to the voltage and current data at the fault information measurement section and the fault information receiving end.

[0115] The voltage at the fault point of the cable line is represented by V f . When the length from the fault point to the fault receiving end is Q, V f is:

[0116] V f = V r cosh(κQ) + z c I r sinh(κQ)

[0117] = V s cosh[κ(l - Q)] + z c I s sinh[κ(l - Q)]

[0118] Based on the voltage and current values at the cable line fault point and their corresponding values at the fault information receiving and sending ends, determine the location of the fault point, expressed as:

[0119]

[0120] where Q represents the location of the fault point.

[0121] S2: Extract fault information according to the wavelet transform model and calculate the fault characteristic index.

[0122] It should be noted that wavelet transform has a powerful ability to process non - stationary signals and can capture the instantaneous characteristics of signals. At the same time, when wavelet transform is used for signal denoising and compression, it can ensure the reconstruction accuracy and signal - to - noise ratio of the signal. By using wavelet transform, the wavelet transform results of cable line operation data in different dimensions can be analyzed to extract the characteristic data of the faulty cable line, realizing fault location and ranging.

[0123] Use wavelet transform to analyze the wavelet transform results of cable line operation data in different dimensions, extract the characteristic data of the faulty cable line, and realize fault location and ranging.

[0124] When the square - integrable function whose corresponding Fourier transform satisfies the following equation, this square - integrable function is called the mother wavelet:

[0125]

[0126] Translate the square - integrable function mother wavelet to be expressed as:

[0127]

[0128] where a represents the scaling factor in the translation process. b represents the translation factor in the translation process.

[0129] Perform power - level discretization and uniform discretization on the scaling factor a and the translation factor b:

[0130]

[0131] where a0 > 1, b0 > 0, both are constants.

[0132] According to the definition of wavelet transform, we get:

[0133]

[0134] The voltage and current variation characteristics of the cable line are represented by f(t), where f(t) ∈ L 2 (R), and the discrete wavelet transform of f(t) is expressed as:

[0135]

[0136] where, denotes the conjugate complex number of

[0137] Furthermore, since the discrete wavelet transform requires convolution and downsampling operations at each stage, it consumes a large amount of computing resources. Due to the frequency and correlation differences, each layer of wavelet coefficients obtained by the discrete wavelet transform may lose important cable line fault characteristics. The present invention uses stationary wavelet transform to extract cable line fault information, cancels downsampling in the decomposition process, and cancels upsampling in the reconstruction part. The stationary wavelet transform can ensure the consistency of the approximate coefficients and detail coefficients of each layer obtained by signal decomposition with the length of the initial signal. The stationary wavelet transform discretizes the scaling factor a while ensuring the continuity of the translation factor b, which is expressed as:

[0138]

[0139] The formula for the stationary wavelet transform of the voltage and current variation characteristics f(t) of the cable is expressed as:

[0140]

[0141] where i represents the decomposition degree, and j represents the separation degree. λ ij (t) represents the scaling function. c represents the approximate coefficient of the stationary wavelet transform, and d represents the detail coefficient.

[0142] The stationary wavelet transform obtains the cable line fault signal, uses the low-frequency approximate coefficient c to characterize the fundamental frequency component of the low-frequency part, and uses the high-frequency detail coefficient d to characterize the transient characteristics of the high-frequency part. During the stationary wavelet transform process, the approximate coefficient and detail coefficient of each layer are expressed as:

[0143]

[0144] where H represents the low-pass filter corresponding to the i-th layer, and G respectively represents the low-pass and high-pass filters corresponding to the i-th layer. n represents the sampling window number. N represents the total number of samplings.

[0145] It should be noted that the decomposition principle of the stationary wavelet transform is as Figure 3 shown.

[0146] Figure 3 After acquiring the voltage and current signal data of the faulty cable line, perform 5-layer decomposition on it using stationary wavelet transform, and a total of 6 sub-frequency bands can be obtained.

[0147] Divide each sub-frequency band into equal lengths according to the working frequency, and extract features for each time period. The selected features include energy value, average value, variance, and entropy value.

[0148] The energy value is used to characterize the energy distribution characteristics of the actually measured propagation data of the cable line. For the q-th sub-frequency band, the energy value is expressed as:

[0149]

[0150] Among them, N t represents the number of sampling points.

[0151] The average value is used to characterize the central tendency of the actually measured cable parameter propagation data and thus reflect the stability of the object, and is expressed as:

[0152]

[0153] The variance is used to characterize the degree of dispersion of the actually measured cable parameter propagation data, and is expressed as:

[0154]

[0155] The entropy value is correlated with the degree of chaos of the data and is used to characterize the chaotic characteristics of the energy sequence of the actually measured cable parameter propagation data, and is expressed as:

[0156]

[0157] Among them, E represents the total energy value of the sub-frequency band.

[0158] To avoid the influence of the feature value dimension difference on the result, normalize the above four feature parameters, and it is expressed as:

[0159]

[0160] Among them, x c represents the value after normalizing the data sequence, x r represents the initial value, x min represents the minimum value, x max represents the maximum value.

[0161] S3: Establish an optimized neural network algorithm to locate the cable line fault, obtain the non-linear relationship between the fault sample and the fault location, and output the fault location.

[0162] It should be noted that the BP neural network is an artificial intelligence method based on the multi-layer perceptron structure that can effectively handle non-linear relationships. It has strong adaptive learning ability and can improve the performance of the network by adjusting weights and thresholds to obtain high accuracy and precision. Using the BP neural network for cable line fault location has good results.

[0163] When a short-circuit grounding fault occurs in the cable line, the grounding resistance corresponding to the fault point is relatively small. When the cable line has poor contact or is disconnected, resulting in a grounding fault, the grounding resistance corresponding to the fault point is relatively large. In addition, the initial angle of the signal corresponding to the cable line fault also has great uncertainty, which affects the traveling wave amplitude of the faulty cable. In order to study the influence of different grounding resistances and fault angles and expand the diversity of neural network training samples, this paper establishes a neural network database including different fault occurrence positions, different fault resistance values, and different fault angles, as shown in Table 1.

[0164] Table 1 Fault neural network database table

[0165]

[0166] Furthermore, the BP neural network can approximate any model with uncertainty with arbitrary precision and has strong modeling ability for uncertain models. The BP neural network algorithm includes two processes: forward propagation of signals and backpropagation of errors. During forward propagation, signals are propagated from the input layer to the output layer through the hidden layer. If the result of the output layer does not meet the expectation, the signal will be propagated backward, and the sum of squared network errors will be minimized by adjusting the weights and thresholds of the network.

[0167] The mapping relationship of the BP neural network is expressed as:

[0168]

[0169] Among them, c j represents the weight from the hidden layer to the output layer. b j represents the output of the hidden layer nodes. ε represents the threshold of the output layer.

[0170] The tansig activation function is used to calculate the correlation degree between the outputs of the hidden layer nodes. The correlation degree between the c i th node and the c j th node is expressed as:

[0171]

[0172] The output b j of the hidden layer nodes is expressed as:

[0173]

[0174] The error between the c-th node of the hidden layer and the c-th node is expressed as: i and the c-th node j is:

[0175]

[0176] where w ij represents the weight from the input layer to the hidden layer. θ j represents the threshold of the hidden layer node.

[0177] It should be noted that when using the BP neural network to locate cable line faults, overfitting or underfitting may occur during the training process, and corresponding measures need to be taken for adjustment and optimization. In this paper, the firefly swarm optimization algorithm is used to optimize and solve the weights and thresholds of the BP neural network, making up for the random defects in the selection of connection weights and thresholds of the BP neural network, so that the BP neural network has better generalization mapping ability and stronger learning ability.

[0178] The firefly swarm algorithm simulates the search and optimization process as the attraction and movement process of firefly individuals, and quantifies the optimal position of firefly individuals by solving the objective function problem. In the firefly swarm algorithm, each firefly is distributed in a predefined objective function interval, and its brightness depends on the objective function value at its location. Assuming the number of fireflies is N, the position and objective function corresponding to the i-th firefly are (x i , y i ) and f(x i , y i ), and the position corresponding to the luciferin value is T i . The decision radius of each firefly is expressed as:

[0179]

[0180] where R s represents the perception radius. β represents the control parameter. N i (t) is expressed as:

[0181]

[0182] where X i (t) represents the position of the i-th firefly at the t-th generation, and I i (t) represents the luciferin value of the i-th firefly at the t-th generation.

[0183] The position update formula of each firefly is:

[0184]

[0185] where s represents the moving step size.

[0186] After globally optimizing the weights and thresholds of the BP neural network using the firefly swarm algorithm, a transmission line cable fault location strategy with better performance in convergence ability and prediction accuracy is obtained.

[0187] Furthermore, after globally optimizing the weights and thresholds of the BP neural network using the firefly swarm algorithm, a transmission line cable fault location strategy with better performance in convergence ability and prediction accuracy can be obtained. Based on the database containing different fault types established in Table 1, the non-linear relationship between the fault samples and the fault locations can be directly obtained by the BP neural network, and the output neuron directly outputs the fault location (represented by a numerical value) corresponding to the fault sample data.

[0188] In the above embodiments, there is also a DC battery state monitoring and fault diagnosis system, specifically:

[0189] Model establishment module, establishing a cable line fault location model to determine the fault point.

[0190] Calculation feature module, extracting fault information according to the wavelet transform model and calculating fault feature indexes.

[0191] Optimization module, establishing an optimized neural network algorithm to locate the cable line fault, obtaining the non-linear relationship between the fault samples and the fault locations, and outputting the fault location.

[0192] The computer device can be a server. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the data cluster data of the power monitoring system. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through a network connection. The computer program, when executed by the processor, implements a DC battery state monitoring and fault diagnosis method.

[0193] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0194] Embodiment 2, referring to Figures 4 to 8 , which is an embodiment of the present invention, provides a method and system for monitoring the state and diagnosing faults of a DC battery. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0195] In order to verify the effectiveness and feasibility of the cable line fault location method based on stationary wavelet transform and improved BP neural network proposed in this paper, sample data is collected for the common short-circuit grounding faults of cable lines in this paper.

[0196] The total length of the cable line sample is selected to be 51 km, and the on-site survey of the cable is as Figure 4 shown.

[0197] On-site survey enables the collection of a large number of real fault sample data. These data are derived from various real faults, including high-temperature faults, low-temperature faults, voltage instability, mechanical component wear, etc. To obtain as diverse data as possible, this paper simulates the operating conditions under which various cables may fail, and selects a series of representative fault types. Integrate all the faults that occur in the cables in real scenarios to construct a fault sample database. This database serves as the training basis for the neural network model. Collect more data in actual scenarios, continuously expand the sample database, further improve the diversity of data, and enhance the credibility of the model.

[0198] 1000 cable fault data samples are obtained through on-site survey to avoid the subjective influence of the artificial selection process on the results. Then, 150 samples are randomly selected for verifying the performance of the neural network model after training, and 50 samples are used for testing. Among the 1000 selected samples, after performing undecimated decomposition using stationary wavelet transform to obtain six sub-bands, each sub-band is divided at a certain time interval, and the values of the four characteristic parameters corresponding to Section 2.3 are calculated for each divided sub-band. After normalization, they are used as the input of the BP neural network. The distribution characteristics of 24 normalized characteristic values corresponding to 5 groups of sample data are as Figure 5 shown.

[0199] According to Figure 5 the 5 groups of sample data shown: After decomposing the cable fault data using stationary wavelet transform and obtaining the characteristic values corresponding to the sub-bands, each group of characteristic parameters varies due to their corresponding different fault positions. The BP neural network can calculate the inherent non-linear relationship between the characteristic values and the fault positions, thereby realizing the prediction of the fault positions. The change curve of the error of the BP neural network during the training process with the number of iterations is randomly selected as Figure 6 shown.

[0200] According to Figure 6 it can be known that: Both the BP neural network and the BP neural network optimized by the firefly swarm algorithm can reach relatively low training errors after the iteration is completed. Thanks to the optimization calculation ability of the firefly swarm algorithm for the weight and threshold parameters of the BP neural network, the training error of the optimized BP neural network is reduced. Therefore, under the same training error requirements, the number of iterations and the iteration time of the BP neural network optimized by the firefly swarm algorithm can be significantly reduced, and it has better performance.

[0201] Use the selected 50 samples to test the BP neural network model optimized by the firefly swarm algorithm, and analyze the test results of 10 randomly selected samples, as shown in Table 2. The accuracy change curves of the two cable fault location strategies in Table 2 are as Figure 7 shown.

[0202] Test Results of the Model in Table 2

[0203]

[0204] As can be seen from Table 2: After using the firefly swarm optimization algorithm to optimize the weights and thresholds of the BP neural network, the prediction accuracy (offset rate) of the cable line fault location is within 2%. Due to factors such as unreasonable selection of weights and thresholds, the traditional BP neural network is prone to falling into the overfitting or underfitting process, and its prediction accuracy for the cable line fault location is relatively poor.

[0205] According to Figure 7 it can be seen that when testing the two cable line fault location strategies of the BP neural network optimized by the firefly swarm and the traditional BP neural network, the offset rate of the optimized BP neural network is always less than that of the traditional BP neural network. When using the traditional BP neural network to predict the cable fault location, its maximum offset rate is as high as 3.836%. After using the firefly swarm algorithm to optimize the weights and thresholds to improve the convergence performance and prediction accuracy, the corresponding offset rate of the fault location prediction is only 1.906% at most, which is 50.313% less than that of the traditional BP neural network. In addition, corresponding Figure 7 to the 10 groups of data, the average value of the offset rates corresponding to the fault prediction positions of all test samples without optimization is 2.598%. After being optimized by the firefly swarm algorithm, this average value is reduced to 1.669%, a reduction of 35.758%. Therefore, the idea proposed in this paper of using the firefly swarm to optimize the BP neural network for cable fault location prediction is effective and can significantly improve the prediction accuracy. The fault test samples of the cable line are randomly divided into 5 groups, each group contains 10 test samples, and the corresponding average offset rate of each group is calculated. The results obtained are as Figure 8 shown.

[0206] According to Figure 8 it can be seen that the maximum average prediction offset rate of the 5 groups of test data selected without optimization is 2.534%, while the maximum average prediction offset rate after optimization is only 1.902%. Comparing the two cable fault location strategies of the BP neural network optimized by the firefly swarm and the traditional BP neural network, the average offset rate of the cable line fault location corresponding to the BP neural network optimized by the firefly swarm is smaller, and it has a more accurate fault location performance. The unreasonable selection of weights and thresholds in the traditional BP neural network leads to a slight decrease in its prediction accuracy. The optimization effect of the fourth group of data is the most obvious. The intervention of the firefly swarm algorithm reduces the average value of the fault prediction offset rate of the BP neural network by 33.458%. The BP neural network optimized by the firefly swarm algorithm proposed in this paper for cable line fault location has good application prospects.

[0207] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for monitoring the state and diagnosing faults of a DC storage battery, characterized in that, Including: Establish a cable line fault location model to determine the fault point; Extract fault information according to the wavelet transform model and calculate fault characteristic indexes; Establish an optimized neural network algorithm to locate cable line faults, obtain the non-linear relationship between fault samples and fault locations, and output the fault location.

2. The method for monitoring the state and diagnosing faults of a DC battery according to claim 1, characterized in that: The establishment of the cable line fault location model includes that the types of cable line faults are divided into three cases: short circuit, open circuit, and poor contact; The voltage and current parameters of the cable line are characterized by the cable line parameters, and cable line fault location is realized according to the change amounts of voltage and current; in case of a fault, the relationship between the voltage and current of different segments of the cable and the cable line parameters is: Wherein, R represents the resistance of the cable line per unit length; L represents the inductance of the cable line per unit length; C represents the capacitance of the cable line per unit length; G represents the conductance of the cable line per unit length; x represents the length (km) from the fault information measurement end to the fault information receiving end; v represents the cable line voltage; i represents the cable line current; Voltage, current transfer impedance z c and the corresponding transfer constant κ are expressed as: The voltage and current of the fault information receiving end are represented by V r and I r respectively. When the distance between the fault information measuring end and the fault information receiving end is x, the voltage V x and current I x at the fault information measuring end are as follows: The voltage and current of the fault information sending end are represented by V s and I s respectively. When the total length of the faulty cable line is l, the voltage V x and current I x at the fault information measurement end are as follows: When a fault occurs at any point on the cable line, the cable line is regarded as being divided into two similar parts by the fault point, and these two parts are regarded as normal cables without faults; When a fault occurs, determine the fault occurrence location according to the voltage and current data of the fault information measurement section and the fault information receiving end.

3. The method for monitoring the state and diagnosing faults of a DC battery according to claim 2, wherein: The determination of the fault point includes that the voltage of the fault point on the cable line is represented by V f When the length from the position of the fault point to the fault receiving end is Q, V f is as follows: V f = V r cosh(κQ) + z c I r sinh(κQ) = V s cosh[κ(l - Q)] + z c I s sinh[κ(l - Q)] Determine the location of the fault point according to the voltage and current values of the fault point on the cable line and the corresponding values of the fault information receiving end and the sending end, which is expressed as: Wherein, Q represents the location of the fault point.

4. The method for monitoring the state and diagnosing faults of a DC battery according to claim 3, characterized in that: The extraction of fault information according to the wavelet transform model includes analyzing the wavelet transform results of cable line operation data from different dimensions by using wavelet transform, extracting the characteristic data of the faulty cable line, and realizing fault location and ranging; When the square-integrable function the corresponding Fourier transform satisfies the following formula, the square-integrable function is called a mother wavelet: The square-integrable function-based wavelet is translated as follows: Wherein, a represents the scaling factor in the translation process; b represents the translation factor in the translation process; Perform power-level discretization and uniform discretization processing on the scaling factor a and the translation factor b: Wherein, a0 > 1, b0 > 0, both are constants; According to the definition of wavelet transform, it is obtained that: The voltage and current variation characteristics of the cable line are represented by f(t), where f(t) ∈ L 2 (R), and the discrete wavelet transform of f(t) is expressed as: Among them, denotes the conjugate complex number of; Adopt stationary wavelet transform to extract cable line fault information, cancel downsampling in the decomposition process, and cancel upsampling in the reconstruction part; stationary wavelet transform performs discretization processing on the scaling factor a, and at the same time ensures the continuity of the translation factor b, which is expressed as: The stationary wavelet transform formula of the voltage and current change characteristic f(t) of the cable is expressed as: where i represents the decomposition degree and j represents the separation degree; λ ij (t) represents the dilation function; c represents the approximation coefficient of the stationary wavelet transform, and d represents the detail coefficient; Stationary wavelet transform obtains the cable line fault signal, uses the low-frequency approximation coefficient c to characterize the fundamental frequency component of the low-frequency part, and uses the high-frequency detail coefficient d to characterize the transient characteristics of the high-frequency part; in the process of stationary wavelet transform, each layer of approximation coefficient and detail coefficient is expressed as: Wherein, H represents the low-pass filter corresponding to the i-th layer, and G respectively represents the low-high pass filter corresponding to the i-th layer; n represents the sampling window serial number; N represents the total number of samplings.

5. The DC battery state monitoring and fault diagnosis method according to claim 4, characterized in that: The calculation of the fault characteristic indexes includes equally dividing each sub-band according to the working frequency, extracting features for each time period, and the selected features include energy value, average value, variance, and entropy value; The energy value is used to characterize the energy distribution characteristics of the actual measured propagation data of the cable line. For the q-th sub-band, the energy value is expressed as: Among them, N t represents the number of sampling points; The average value is used to characterize the central tendency of the measured cable parameter propagation data, thereby reflecting the stability of the object, and is expressed as: The variance is used to characterize the dispersion degree of the measured cable parameter propagation data, and is expressed as: The entropy value is correlated with the degree of chaos of the data and is used to characterize the chaotic characteristics of the energy sequence of the measured cable parameter propagation data, and is expressed as: Among them, E represents the total energy value of the sub-band; To avoid the influence of eigenvalue dimension differences on the results, the above four characteristic parameters are normalized, and are expressed as: Among them, x c represents the value after normalization of the data sequence, x r represents the initial value, x min represents the minimum value, x max represents the maximum value.

6. The method for monitoring the state and diagnosing faults of a DC battery according to claim 5, characterized in that: The establishment of the optimized neural network algorithm for cable line fault location includes that the mapping relationship of the BP neural network is expressed as: Among them, c j represents the weight from the hidden layer to the output layer; b j represents the output of the hidden layer nodes; ε represents the threshold of the output layer; Calculate the correlation between the outputs of the hidden layer nodes using the tansig activation function. The correlation between the c i -th node and the c j -th node is expressed as: The output b of the hidden layer nodes j It is expressed as: The error between the c-th node of the hidden layer and the c-th node is expressed as: i the c-th j node Among them, w ij represents the weight between the input layer and the hidden layer; θ j represents the threshold of the hidden layer nodes.

7. The method for monitoring the state and diagnosing faults of a DC battery according to claim 6, characterized in that: The output fault location includes using the firefly swarm optimization algorithm to optimize and solve the weights and thresholds of the BP neural network, making up for the random defects of the BP neural network in the selection of connection weights and thresholds; assuming that the number of fireflies is N, the position and objective function corresponding to the i-th firefly are (x i , y i ) and f(x i , y i ), and the position corresponding to the luciferin value is T i . The decision radius of each firefly is expressed as: Among them, R s represents the sensing radius; β represents the control parameter; N i (t) is expressed as: Among them, X i (t) represents the position of the i-th firefly at the t-th generation, I i (t) represents the fluorescence value of the i-th firefly at the t-th generation; The position update formula for each firefly is: Among them, s represents the moving step size; After using the firefly swarm algorithm to globally optimize the weights and thresholds of the BP neural network, a transmission line cable fault location strategy with better performance in convergence ability and prediction accuracy is obtained.

8. A DC battery state monitoring and fault diagnosis system using the method according to any one of claims 1-7, characterized in that: A model establishment module that establishes a cable line fault location model and determines the fault point; A calculation feature module that extracts fault information according to the wavelet transform model and calculates fault feature indicators; An optimization module that establishes an optimized neural network algorithm to locate cable line faults, obtains the non-linear relationship between the fault sample and the fault location, and outputs the fault location.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images