A low voltage fault arc detection method based on self-organizing brain emotion learning

Abstract

The present invention provides a low-voltage fault arc detection method based on self-organizing brain emotion learning, comprising: S1: calculating the mean square error of periodic sampling current and the high-frequency components of periodic current wavelet decomposition and reconstruction respectively through current detection data of low-voltage power supply and distribution lines; S2: constructing a fuzzy brain emotion learning neural network model with a function adjustment function of two channels and automatically optimizing the model structure and parameters; S3: setting the optimal learning rate of the model by using the Lyapunov function to make the model system stable, thereby correctly detecting the fault arc in the low-voltage power supply and distribution lines; using the mean square error of periodic sampling current and the high-frequency components of periodic current wavelet decomposition and reconstruction obtained by calculating the fault arc current data as feature data, inputting the model into the model to train the model until the model converges; S4: performing fault arc detection on the current to be measured through the low-voltage distribution line fault arc model based on self-organizing brain emotion learning.

Description

Technical Field

[0001] The present invention relates to the technical field of low-voltage fault arc detection, and in particular to a low-voltage fault arc detection method based on self-organizing brain emotion learning. Background Art

[0002] Most of the fault currents in low-voltage distribution networks are caused by arc faults. Arc is a continuous discharge phenomenon between two electrodes across the insulating medium, often accompanied by local heating of the electrodes. A typical arc is formed in the air gap between the two poles, and its center temperature is generally 5000 to 15000 degrees Celsius. In low-voltage distribution networks, lines and equipment have long working hours and complex operating environments. The insulation layer may age, break or have poor contact, resulting in a decrease in its insulation capacity, which in turn generates a fault arc. After a fault arc is generated, extremely high temperatures can be generated at a relatively low current, which is enough to ignite combustibles around the line or equipment.

[0003] The fault arc in the low-voltage power supply and distribution line has a small current value, which is not enough to drive the circuit breaker or fuse protection action, and it is very easy to cause electrical fires. The loads in the power supply and distribution line are diverse, and have characteristics similar to the fault arc waveform. Therefore, traditionally, a single criterion is used to distinguish arc faults, and the misjudgment rate is high. Summary of the invention

[0004] In view of this, the purpose of the present invention is to propose a low-voltage fault arc detection method based on self-organizing brain emotion learning, which can accurately judge the fault arc to ensure the safe operation of the low-voltage power supply and utilization system.

[0005] To solve the above problems, the present invention adopts the following technical solutions:

[0006] A low-voltage fault arc detection method based on self-organizing brain emotion learning comprises the following steps:

[0007] S1: Calculate the mean square error of periodic sampling current, the high-frequency component of periodic current wavelet decomposition and reconstruction through the current detection data of low-voltage power supply and distribution lines;

[0008] S2: Construct a fuzzy brain emotion learning neural network model with two-channel function adjustment function and automatic optimal adjustment of model structure and parameters;

[0009] S3: Using Lyapunov function to set the optimal learning rate of the model, so that the model system is stable, so as to correctly detect the fault arc in the low-voltage power supply and distribution line;

[0010] The mean square error of the periodic sampling current calculated from the fault arc current data, the high-frequency component of the periodic current wavelet decomposition and reconstruction are used as feature data and input into the model for training until the model converges, thus obtaining a low-voltage distribution line fault arc model based on self-organizing brain emotion learning.

[0011] S4: Fault arc detection of the current to be measured through a low-voltage distribution line fault arc model based on self-organizing brain emotion learning.

[0012] Furthermore, the mean square error of the periodic sampling current in S1 is calculated as follows:

[0013]

[0014] Among them, I n represents the current value sampled in the nth sampling cycle, i represents the i-th sampling cycle, represents the mean value of the current signal sampled in i sampling cycles, S i Represents the mean square error of the sampled current within i sampling periods;

[0015] The periodic current wavelet decomposition is calculated as follows:

[0016]

[0017] Among them, W f (a,b) represents the decomposed wavelet function, a represents the periodic scaling factor, and b represents the periodic translation factor; Represents the wavelet periodic function with independent variable *, and the expression is as follows:

[0018]

[0019] The periodic current wavelet reconstruction is calculated as follows:

[0020]

[0021] Among them, M i represents a wavelet transformed current signal having a sequence of i sampling periods, and γ is a given wavelet transform constant.

[0022] Further, S2 includes the following steps:

[0023] S2.1: Define the input data space dimensions Where i represents the sequence of periodic sampling;

[0024] The fuzzy brain emotion learning neural network model adopts a two-channel function regulation function, which has an enhanced transformation function with expanded dimension. Through the processing of the function regulation function, the outputs of the amygdala and orbitofrontal cortex model networks are:

[0025]

[0026] in, and are the weight values ​​of the functional regulation function corresponding to the amygdala and orbitofrontal cortex model networks, is the input data x i The dimension-enhanced transformation function, Q is the input dimension after the function is expanded, and p is the number of fuzzy rules of the fuzzy neural network;

[0027] S2.2: Establish a fuzzy brain neural network model, in which the fuzzy reasoning rules of the amygdala and orbitofrontal cortex are as follows:

[0028] R a :If S i yes M i yes

[0029] Then a=v hj ,h=1,2;j=1,2,…,p

[0030] R o :If S i yes M i yes

[0031] Then o=ω hj ,h=1,2;j=1,2,…,p

[0032] Among them, R a is the fuzzy inference rule of the amygdala model, R o is the fuzzy inference rule of the orbitofrontal cortex model, h is the dimension of the input vector, j is the fuzzy set of the jth layer, p is the number of fuzzy rules, and v hj is the fuzzy weight value of the h-th input of the amygdala model output at the j-th layer, ω hj is the fuzzy weight value of the h-th input of the orbitofrontal cortex model output at the j-th layer, a is the output of the amygdala model, and o is the output of the orbitofrontal cortex model;

[0033] The fuzzy brain neural network model has a 5-layer structure (5-layer signals are transmitted downward layer by layer), which are as follows:

[0034] Layer 1: Pass the signal directly to the next layer. The expression is as follows:

[0035]

[0036] Among them, x ihrepresents the h-th dimension current signal sampled in the i-th cycle;

[0037] Layer 2: Fuzzy reasoning is used, and each node is used as a membership function. The membership function has the expression of a Gaussian function as follows:

[0038]

[0039] in, are the means of the Gaussian functions during the blurring of the amygdala and orbitofrontal cortex, respectively. are the standard deviations of the Gaussian functions during blurring of the amygdala and orbitofrontal cortex, respectively; the symbol exp represents the exponential operation of e.

[0040] Layer 3: The neurons in this layer represent the preprocessing part of the fuzzy rules. The output is obtained by multiplication. The output is expressed as follows:

[0041]

[0042] Among them, the symbol ∏ represents the cumulative multiplication operator, i represents the sequence of periodic sampling, h is the dimension of the input vector, j is the fuzzy set of the jth layer, and p is the number of fuzzy rules;

[0043] Layer 4: Multiply the signal transmitted to this layer by the output signal of the function, expressed as follows:

[0044]

[0045] in, and is the adjustment value of the weight; ∑ is the accumulation symbol;

[0046] Layer 5: The output Out of the fuzzy brain neural network model is expressed as:

[0047] Out=a i -o i

[0048]

[0049] Among them, a i and i Represent the output of the amygdala and orbitofrontal cortex models, respectively, and v i and ω i are the weights of the amygdala and orbitofrontal cortex models respectively, and the weight values ​​are updated by the following formula:

[0050]

[0051] Among them, ρ i and σ i They are Δvi and Δω i ES is the learning rate, which reflects people’s satisfaction with the system performance. The symbol max indicates the maximum value.

[0052] S2.3: Fuzzy brain neural network model, with self-organizing layer-adding rules of sensory cortex, defining the average measurement distance D ij ,j=1,…,p:

[0053] D ij =||x i -m j ||2

[0054] Among them, m j =[m 1j ,m 2j ] T , m j is the preset reference value; where ‖·‖2 represents the 2-norm.

[0055] For j = 1, 2, ..., p, let the objective function D ij Take the variable value j when the minimum value is obtained, and then assign this variable value to The expression is as follows;

[0056]

[0057] For h = 1, 2, the objective function The variable h when taking the maximum value is expressed as follows:

[0058]

[0059] Among them, K g is the set layer adding threshold;

[0060] S2.4: Fuzzy brain neural network model, with self-organizing layer reduction rules for receiving domain layers, defining the objective function value M ij :

[0061]

[0062] Among them,

[0063] For j = 1, 2, ..., p, let the objective function M ij Take the variable value j when the minimum value is obtained, and then assign this variable value to The expression is as follows;

[0064]

[0065] Among them, M ij ≤K c; K c is the set layer reduction threshold.

[0066] Furthermore, the learning rate of the model in S3 includes the amygdala model learning rate Learning rate with orbitofrontal cortex model

[0067] The optimal learning rate of the model is set through the Lyapunov function stability theory. The setting of the optimal learning rate is as follows:

[0068]

[0069] Furthermore, S4 is specifically:

[0070] The mean square error of the periodic sampling current calculated from the current data to be tested, the high-frequency components of the periodic current wavelet decomposition and reconstruction are used as feature data and input into the trained low-voltage distribution line fault arc model based on self-organizing brain emotion learning for arc fault detection.

[0071] Based on the above scheme, the present invention also provides a computer-readable storage medium, in which at least one instruction, at least one program, code set or instruction set is stored, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the low-voltage fault arc detection method described above.

[0072] Beneficial effects:

[0073] Compared with the traditional single indicator, the low-voltage fault arc detection method proposed in the present invention uses the calculation of the mean square error of the periodic current and the calculation of the high-frequency component of the periodic current wavelet decomposition and reconstruction as the judgment criterion, which improves the accuracy of fault arc judgment, can grasp the health status of the low-voltage power supply and distribution lines in real time, ensure the safe operation of the low-voltage power supply and use system, and is suitable for further promotion and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the embodiments of the present invention 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 only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0075] Figure 1 It is a brief schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0076] The present invention will be further described in detail below in conjunction with the accompanying drawings and examples. It is particularly noted that the following examples are only used to illustrate the present invention, but are not intended to limit the scope of the present invention. Similarly, the following examples are only partial embodiments of the present invention rather than all embodiments, and all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0077] like Figure 1 As shown, this embodiment provides a low-voltage fault arc detection method based on self-organizing brain emotion learning, comprising the following steps:

[0078] S1: Based on the historical current detection data of the low-voltage power supply and distribution line, the mean square error of the periodic sampling current, the high-frequency component of the periodic current wavelet decomposition and reconstruction are calculated respectively;

[0079] Among them, the mean square error of the periodic sampling current is calculated as follows:

[0080]

[0081] Among them, I n represents the current value sampled in the nth sampling cycle, i represents the i-th sampling cycle, represents the mean value of the current signal sampled in i sampling cycles, S i Represents the mean square error of the sampled current within i sampling periods;

[0082] The periodic current wavelet decomposition is calculated as follows:

[0083]

[0084] Among them, W f (a,b) represents the decomposed wavelet function, a represents the periodic scaling factor, and b represents the periodic translation factor; Represents the wavelet periodic function with independent variable *, and the expression is as follows:

[0085]

[0086] The periodic current wavelet reconstruction is calculated as follows:

[0087]

[0088] Among them, M i represents a wavelet transformed current signal having a sequence of i sampling periods, and γ is a given wavelet transform constant.

[0089] S2: Construct a fuzzy brain emotion learning neural network model, which has a two-channel function adjustment function and can automatically optimize the structure and parameters of the model;

[0090] The specific steps include:

[0091] S2.1: Define the input data space dimensions Where i represents the sequence of periodic sampling;

[0092] The fuzzy brain emotion learning neural network model adopts a two-channel function adjustment function, which has an enhanced transformation function with expanded dimension. The expression is as follows:

[0093]

[0094] The symbol T represents the transpose of the matrix.

[0095] Therefore, through the processing of the functional regulation function, the outputs of the amygdala and orbitofrontal cortex model networks are:

[0096]

[0097] in, and are the weight values ​​of the functional regulation function corresponding to the amygdala and orbitofrontal cortex model networks, is the input data x i The dimension expansion and enhancement transformation function is , Q is the input dimension after the function is expanded, and p is the number of layers of the amygdala and orbitofrontal cortex model network.

[0098] S2.2: A fuzzy brain neural network model is established, in which the fuzzy reasoning rules of the amygdala and orbitofrontal cortex are as follows:

[0099] R a :If S i yes M i yes

[0100] Then a=v hj ,h=1,2;j=1,2,…,p

[0101] R o :If S i yes M i yes

[0102] Then o=ω hj ,h=1,2;j=1,2,…,p

[0103] Among them, R a is the fuzzy inference rule of the amygdala model, R ois the fuzzy inference rule of the orbitofrontal cortex model, h is the dimension of the input vector, j is the fuzzy set of the jth layer, p is the number of fuzzy rules, and v hj is the fuzzy weight value of the h-th input of the amygdala model output at the j-th layer, ω hj is the fuzzy weight value of the h-th input of the orbitofrontal cortex model output at the j-th layer, a is the output of the amygdala model, and o is the output of the orbitofrontal cortex model;

[0104] The fuzzy brain neural network model has a 5-layer structure (5 layers of signals are transmitted downward layer by layer), which are as follows:

[0105] Layer 1: Pass the signal directly to the next layer. The expression is as follows:

[0106]

[0107] Among them, x ih represents the h-th dimension current signal sampled in the i-th cycle;

[0108] Layer 2: This layer uses fuzzy reasoning, and each node is used as a membership function. The membership function has the expression of a Gaussian function as follows:

[0109]

[0110] in, are the means of the Gaussian functions during the blurring of the amygdala and orbitofrontal cortex, respectively. They are represented as the standard deviation of the Gaussian function in the blurring process of the amygdala and orbitofrontal cortex, respectively; the symbol exp represents the exponential operation of e; and Represented as the results of running Gaussian functions for the amygdala and orbitofrontal cortex, respectively.

[0111] Layer 3: The neurons in this layer represent the preprocessing part of the fuzzy rules. The output is obtained by multiplication. The output is expressed as follows:

[0112]

[0113] Among them, the symbol ∏ represents the cumulative multiplication operator, i represents the sequence of periodic sampling, h is the dimension of the input vector, j is the fuzzy set of the jth layer, and p is the number of fuzzy rules; and They are the cumulative multiplication of the Gaussian functions of the amygdala and orbitofrontal cortex, respectively.

[0114] Layer 4: Multiply the signal transmitted to this layer by the output signal of the function, expressed as follows:

[0115]

[0116] in, and are the adjustment values ​​of weights respectively; ∑ is the accumulation symbol; and They are the weighted outputs of the amygdala and orbitofrontal cortex.

[0117] Layer 5: The output Out of the fuzzy brain neural network model is expressed as:

[0118] Out=a i -o i

[0119]

[0120] Among them, a i and i Represent the output of the amygdala and orbitofrontal cortex models, respectively, and v i and ω i are the weights of the amygdala and orbitofrontal cortex models respectively, and the weight values ​​are updated by the following formula:

[0121]

[0122] Among them, ρ i and σ i They are Δv i and Δω i The learning rate, ES is the emotional signal, which reflects people’s satisfaction with the system performance; the symbol max indicates the maximum value.

[0123] S2.3: Fuzzy brain neural network model, with self-organizing layer-adding rules of sensory cortex, defining the average measurement distance D ij ,j=1,…,p:

[0124] D ij =||x i -m j ||2

[0125] Among them, m j =[m 1j ,m 2j ] T , m j is the preset reference value; where ‖·‖2 represents the 2-norm.

[0126] For j = 1, 2, ..., p, let the objective function D ij Take the variable value j when the minimum value is obtained, and then assign this variable value to The expression is as follows;

[0127]

[0128] For h = 1, 2, the objective function The variable h when taking the maximum value is expressed as follows:

[0129]

[0130] Among them, K g is the set layer adding threshold;

[0131] S2.4: Fuzzy brain neural network model, with self-organizing layer reduction rules for receiving domain layers, defining the objective function value M ij :

[0132]

[0133] Among them,

[0134] For j = 1, 2, ..., p, let the objective function M ij Take the variable value j when the minimum value is obtained, and then assign this variable value to The expression is as follows;

[0135]

[0136] Among them, M ij ≤K c ; K c is the set layer reduction threshold.

[0137] The above steps can realize self-adjustment of the model structure and automatic optimization.

[0138] S3: Using Lyapunov function to set the optimal learning rate of the model, so that the model system is stable, so as to correctly detect the fault arc in the low-voltage power supply and distribution line;

[0139] The objective function is defined as:

[0140]

[0141] Among them, y d is the low voltage fault arc reference output, E i is the square of the error.

[0142] The error term to be transmitted at the fourth layer is:

[0143]

[0144] Among them, the symbol represents differential; is the objective function Find the partial derivative, is the objective function Find the partial derivative.

[0145] The error term to be transmitted in the third layer is:

[0146]

[0147] in, is the objective function Find the partial derivative, is the objective function Find the partial derivative.

[0148] The connection weight values ​​in the functional connection function are updated as:

[0149]

[0150] in, and are the learning rates of the amygdala model and orbitofrontal cortex model, and The weight values ​​are and The increment in each sampling period i. Then the connection weight value is updated by the following formula:

[0151]

[0152] Among them, i is the sampling time series.

[0153] The Lyapunov function is established as follows:

[0154]

[0155] Among them, e(i)=y d (i)-Out(i), V(i) is a positive definite energy function.

[0156] Define the increment of the Lyapunov function as:

[0157]

[0158] The error equation can be expressed as:

[0159]

[0160]

[0161] From formulas (1) to (8), we can get:

[0162]

[0163] therefore,

[0164]

[0165] ΔV(i) can be expressed as:

[0166]

[0167] Therefore, the learning rate of the brain emotion model is finally set as follows, so that ΔV(i)<0, ensuring the convergence of the tracking error e(i);

[0168]

[0169] Referring to the model learning rate set above, the mean square error of the periodic sampling current calculated from the fault arc current data, the high-frequency component of the periodic current wavelet decomposition and reconstruction are used as feature data, and input into the model to train the model until the model converges, thereby obtaining a low-voltage distribution line fault arc model based on self-organizing brain emotion learning;

[0170] S4: Fault arc detection is performed on the current to be measured through the low-voltage distribution line fault arc model based on self-organizing brain emotion learning; specifically:

[0171] The mean square error of the periodic sampling current calculated from the current data to be tested, the high-frequency components of the periodic current wavelet decomposition and reconstruction are used as feature data and input into the trained low-voltage distribution line fault arc model based on self-organizing brain emotion learning for arc fault detection.

[0172] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0173] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0174] The above descriptions are only some embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any equivalent device or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A low voltage fault arc detection method based on self-organizing brain emotion learning, characterized in that: The steps include: S1: Calculate the mean square error of periodic sampling current, the high-frequency component of periodic current wavelet decomposition and reconstruction through the current detection data of low-voltage power supply and distribution lines; S2: Construct a fuzzy brain emotion learning neural network model with two-channel function adjustment function and automatic optimal adjustment of model structure and parameters; S3: Using Lyapunov function to set the optimal learning rate of the model, so that the model system is stable, so as to correctly detect the fault arc in the low-voltage power supply and distribution line; The mean square error of the periodic sampling current calculated from the fault arc current data, the high-frequency component of the periodic current wavelet decomposition and reconstruction are used as feature data and input into the model for training until the model converges, thus obtaining a low-voltage distribution line fault arc model based on self-organizing brain emotion learning. S4: Fault arc detection of the current to be measured through a low-voltage distribution line fault arc model based on self-organizing brain emotion learning.

2. A low voltage fault arc detection method based on self-organizing brain emotion learning according to claim 1, characterized in that: The mean square error of the periodic sampling current in S1 is calculated as follows: Among them, I n represents the current value sampled in the nth sampling cycle, i represents the i-th sampling cycle, represents the mean value of the current signal sampled in i sampling cycles, S i Represents the mean square error of the sampled current within i sampling periods; The periodic current wavelet decomposition is calculated as follows: Among them, W f (a,b) represents the decomposed wavelet function, a represents the periodic scaling factor, and b represents the periodic translation factor; Represents the wavelet periodic function with independent variable *, and the expression is as follows: The periodic current wavelet reconstruction is calculated as follows: Among them, M i represents a wavelet transformed current signal having a sequence of i sampling periods, and γ is a given wavelet transform constant.

3. A low voltage fault arc detection method based on self-organizing brain emotion learning according to claim 2, characterized in that: S2 includes the following steps: S2.1: Define the input data space dimensions Where i represents the sequence of periodic sampling; The fuzzy brain emotion learning neural network model adopts a two-channel function regulation function, which has an enhanced transformation function with expanded dimension. Through the processing of the function regulation function, the outputs of the amygdala and orbitofrontal cortex model networks are: in, and are the weight values ​​of the functional regulation function corresponding to the amygdala and orbitofrontal cortex model networks, is the input data x i The dimension-enhanced transformation function, Q is the input dimension after the function is expanded, and p is the number of fuzzy rules of the fuzzy brain emotion learning neural network model; S2.2: Establish a fuzzy brain emotion learning neural network model, where the fuzzy reasoning rules of the amygdala and orbitofrontal cortex models are as follows: R a :If S i yes M i yes Then a=v hj ,h=1,2;j=1,2,…,p R o :If S i yes M i yes Then o=ω hj ,h=1,2;j=1,2,…,p Among them, R a is the fuzzy inference rule of the amygdala model, R o is the fuzzy inference rule of the orbitofrontal cortex model, h is the dimension of the input vector, j is the fuzzy set of the jth layer, p is the number of fuzzy rules, and v hj is the fuzzy weight value of the h-th input of the amygdala model output at the j-th layer, ω hj is the fuzzy weight value of the h-th input of the orbitofrontal cortex model output at the j-th layer, a is the output of the amygdala model, and o is the output of the orbitofrontal cortex model; The fuzzy brain emotion learning neural network model has a 5-layer structure, which are as follows: Layer 1: Pass the signal directly to the next layer. The expression is as follows: Among them, x ih represents the h-th dimension current signal sampled in the i-th cycle; Layer 2: Fuzzy reasoning is used, and each node is used as a membership function. The membership function has the expression of a Gaussian function as follows: in, are respectively represented as the mean of the Gaussian function in the fuzzification process of the amygdala and orbitofrontal cortex models, They are expressed as the standard deviation of the Gaussian function during the fuzzification process of the amygdala and orbitofrontal cortex models, respectively; Layer 3: The neurons in this layer represent the preprocessing part of the fuzzy rules. The output is obtained by "multiplication". The output is expressed as follows: Among them, i represents the sequence of periodic sampling, h is the dimension of the input vector, j is the fuzzy set of the jth level, and p is the number of fuzzy rules; Layer 4: Multiply the signal transmitted to this layer by the output signal of the function, expressed as follows: Layer 5: The output Out of the fuzzy brain emotion learning neural network model is expressed as: Out=a i -o i Among them, a i and i Represent the output of the amygdala and orbitofrontal cortex models, respectively, and v i and ω i are the weights of the amygdala and orbitofrontal cortex models respectively, and the weight values ​​are updated by the following formula: Among them, ρ i and σ i They are Δv i and Δω i The learning rate, ES is an emotional signal that reflects people’s satisfaction with the system performance; S2.3: Fuzzy brain emotion learning neural network model, with self-organizing layer-adding rules of sensory cortex, defining the average measurement distance D ij ,j=1,…,p: D ij =||x i -m j ||2 Among them, m j =[m 1j ,m 2j ] T , m j is the preset reference value; For j = 1, 2, ..., p, the objective function D ij The variable value j when taking the minimum value is assigned to The expression is as follows; For h = 1, 2, the objective function is The variable h at its maximum value is expressed as follows: Among them, K g is the set layer adding threshold; S2.4: Fuzzy brain emotion learning neural network model, with self-organizing layer reduction rules of receiving domain layer, defining the objective function value M ij : Among them, For j = 1, 2, ..., p, let the objective function M ij Take the variable value j when the minimum value is obtained, and then assign this variable value to The expression is as follows; Among them, M ij ≤K c ; K c is the set layer reduction threshold.

4. A low voltage fault arc detection method based on self-organizing brain emotion learning according to claim 3, characterized in that: The learning rate of the model in S3 includes the amygdala model learning rate Learning rate with orbitofrontal cortex model The optimal learning rate of the model is set through the Lyapunov function stability theory. The setting of the optimal learning rate is as follows:

5. A low voltage fault arc detection method based on self-organizing brain emotion learning according to claim 1, characterized in that: S4 is specifically: The mean square error of the periodic sampling current calculated from the current data to be tested, the high-frequency components of the periodic current wavelet decomposition and reconstruction are used as feature data and input into the trained low-voltage distribution line fault arc model based on self-organizing brain emotion learning for arc fault detection.

6. A computer-readable storage medium, characterized in that: The storage medium stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the low-voltage fault arc detection method as described in any one of claims 1 to 5.

Images