Single-phase earth fault diagnosis method based on fusion of steady-state features and transient-state features

Abstract

The invention discloses a steady-state feature and transient feature fused single-phase earth fault diagnosis method. The method comprises the following steps: acquiring online real-time data of a fault line and a non-fault line; amplitude and phase information of the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic and the zero-sequence current active component is extracted, and fault line steady-state information is obtained; calculating the amplitude and phase information of the fault first half-wave signal based on the equally-spaced sampling signal, and taking the amplitude and phase information as the transient information of the single-phase earth fault; and constructing a deep neural network structure model by using the extracted fault feature values as an input layer, optimizing the deep neural network structure model, storing optimal model parameters after training is finished, and inputting the verification set into a fault diagnosis model of the optimal model parameters to finish single-phase earth fault diagnosis. According to the method, the technical concept of combining the steady-state characteristics and the transient characteristics is adopted, the modal aliasing phenomenon of the zero-sequence current signals in the empirical mode decomposition process can be effectively solved, and the single-phase earth fault recognition accuracy and the anti-interference performance are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power fault diagnosis, and in particular to a single-phase grounding fault diagnosis method integrating steady-state characteristics with transient characteristics. Background Art

[0002] In my country's distribution network system, 3-35kV distribution network system generally adopts neutral point ungrounded system and neutral point grounded system through arc suppression coil. According to statistics, the probability of single-phase grounding fault is the highest, accounting for more than 80% of distribution network faults. my country's power companies allow the distribution network system to maintain operation for 1-2 hours when a single-phase grounding fault occurs. However, when the distribution network system operates with a single-phase grounding fault for a long time, it is easy to cause the fault situation to increase. If it is not maintained in time, it will affect the normal operation of the power system and endanger the lives of the people. Therefore, it is of great significance to quickly and accurately locate the fault line and cut it off.

[0003] At present, the fault line selection methods in small current grounding systems are mainly divided into two categories. One is the traditional method, which uses manual analysis of fault data to extract fault features, manually sets fault line selection standards, and completes the fault line selection work. The other is to use machine learning methods to automatically learn and classify fault features from a large number of fault data samples to detect fault lines. Most of the fault line selection methods in small current grounding systems use recorded data for analysis, which has problems such as low line selection accuracy and high cost. In addition, traditional fault analysis methods are too dependent on the judgment of the effective order of singular values; and when the signal does not meet the absolute integrability condition, Fourier transform will be powerless; at the same time, its analysis accuracy is limited by the data window and cannot reflect the damping characteristics and instantaneous frequency of the oscillation.

[0004] In view of this, this application is hereby filed. Summary of the invention

[0005] The technical problem to be solved by the present invention is that the traditional fault analysis method is too dependent on the judgment of the effective order of singular values; and when the signal does not meet the absolute integrability condition, the Fourier transform will be powerless; at the same time, its analysis accuracy is limited by the data window and cannot reflect the damping characteristics and instantaneous frequency of the oscillation. The purpose is to provide a single-phase grounding fault diagnosis method that integrates steady-state characteristics and transient characteristics. The technical concept of combining steady-state characteristics with transient characteristics is adopted. Steady-state information is extracted through frequency deviation processing, and transient information is calculated based on equally spaced sampling signals. A deep neural network structure model is constructed as the input layer, which can effectively solve the modal aliasing phenomenon of zero-sequence current signals in the process of empirical mode decomposition, and greatly improve the accuracy of single-phase grounding fault identification and anti-interference.

[0006] The present invention is achieved through the following technical solutions:

[0007] A single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics comprises the following steps:

[0008] S1, obtaining online real-time data of faulty lines and non-faulty lines in the power dispatching automation system;

[0009] S2, extracting the amplitude and phase information of the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic and the zero-sequence current active component to obtain the steady-state information of the fault line;

[0010] S3, calculating the amplitude and phase information of the first half-wave signal of the fault based on the equally spaced sampling signal as the transient information of the single-phase grounding fault;

[0011] S4, uses the fault feature values ​​extracted by S2 and S3 as the input layer to build a deep neural network structure model, and optimizes its learning function and learning rate. After the model training is completed, the optimal model parameters are saved, and the verification set is input into the single-phase grounding fault diagnosis model with the optimal model parameters to complete the single-phase grounding fault diagnosis.

[0012] In a specific implementation, in step S1, the online real-time data includes zero-sequence current and first half-wave current of the fault line and non-fault line collected by the power dispatching automation system.

[0013] In a specific implementation, in step S2, the amplitude and phase angle information of the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic and the zero-sequence current active component are extracted based on the frequency deviation processing method to obtain the steady-state information of the fault line.

[0014] In a specific implementation, the specific steps of step S2 are as follows:

[0015] S201, filtering and denoising the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic and the zero-sequence current active component signal x(t) to obtain a noise-filtered signal s(t);

[0016] S202, performing Fourier transform on the signal s(t) to obtain the frequency of any mode and the amplitude of the corresponding Fourier spectrum;

[0017] S203, determining whether the mode has a modal aliasing phenomenon, if so, performing frequency deviation processing on the signal s(t) using a frequency deviation processing method; if not, no processing is performed;

[0018] S204, restore the intrinsic mode function, and then perform Hilbert transformation on the dominant intrinsic mode function component to obtain its instantaneous amplitude and instantaneous phase angle information.

[0019] In a specific implementation, in step S203, the specific calculation of the frequency deviation processing method is as follows:

[0020] There is a signal f(t) containing two oscillation modes x1 and x2, whose frequency components are f1 and f2 respectively. After the signal s(t) is transformed by Hilbert, the signal f is obtained. H (t);

[0021] The frequency component of s(t) is deviated by the carrier signal with a carrier frequency of F = 2f1-f2, and the single-sideband modulated signal is obtained as

[0022] f LSSB (t) = Re[(f(t)+jf H (t))e -j2πFt ] (11)

[0023] f USSB (t) = Re[(f(t)+jf H (t))e j2πFt ] (12)

[0024] Among them, f LSSB (t) is the lower sideband signal expression of the single sideband modulation signal, f USSB (t) is the upper sideband signal expression of the single sideband modulation signal, j is the imaginary part calculation unit;

[0025] After the frequency deviation, the frequency of the oscillation mode x1 is

[0026] f 11 =F-f1=f1-f2 (13)

[0027] The frequency of the oscillation mode x2 is

[0028] f 12 =F-f2=2(f1-f2)=2f 11 (14)

[0029] In a specific implementation, the specific steps of step S3 are as follows:

[0030] S301, assuming that the mathematical model sequence of the first half-wave signal of the fault is

[0031]

[0032] In the formula, ω i =2πf i , k is the number of harmonics and A i 、f i and θ i Respectively represent the amplitude, frequency and phase of the ith harmonic, and x(n) represents the noise signal in the first half-wave signal of the fault

[0033] S302, the signal y(n) is subjected to noise reduction processing, and the sampling signal sequence y'(n) is reconstructed. The mathematical model of the first half-wave signal of the fault after noise reduction processing is expressed as follows:

[0034]

[0035] In the formula, A i 、f i and θ i represent amplitude, frequency and phase respectively, * Indicates taking the conjugate of this quantity, m i * Represented as m i The conjugate of i * Represented as v i The conjugation of the quantity, j is the unit for calculating the imaginary part;

[0036] S303, define the sample function as

[0037]

[0038] Wherein, i and j represent the i-th row and j-th column of the matrix respectively, y*(x) represents the conjugate of the function y(x), k represents the number of input measurement signal estimation values, n=k represents the summation starts from n equals k, and N-1 represents the summation stops when n=N-1;

[0039] According to the signal after filtering and the effective order L (2k=L) of the obtained signal, a sample matrix is ​​constructed using the sample function, and a normal equation is constructed through the sample matrix;

[0040] S304, according to the constructed normal equation, solve the homogeneous linear equation system with constant coefficients, and the solution of the equation system is the coefficient Let a 2k-i =a i ,i=1,…,k-1 and

[0041] S305, construct characteristic polynomial:

[0042]

[0043] In the formula, v i is the unit modulus root, a i is the coefficient, and a i =a 2k-i , i=0,1,…k;

[0044] Solve the characteristic polynomial (8) with the obtained coefficients and obtain the conjugate root pair (v i,v i * ), i = 1, 2, ... k, and then substitute the value of the conjugate root pair into equation (6) to obtain the parameter m i (i=1,2,…k);

[0045] S306, the amplitude and phase information of the first half-wave signal of the fault are calculated by the following formula:

[0046]

[0047] In the formula, A i is the amplitude of the signal, θ i is the phase of the signal, i=1,2,…k.

[0048] In a specific implementation, in step S303, the normal equation is as follows:

[0049]

[0050] In the formula, r(i,j) is the sample function, is the coefficient.

[0051] In a specific implementation, the specific steps of step S4 are as follows:

[0052] S401, establish a deep neural network with three hidden layers, the input value of the input layer is the m fault feature values ​​extracted, recorded as column vector {x m};

[0053] S402, using the three steady-state characteristic quantities of step S2: zero-sequence current fundamental amplitude and phase information, zero-sequence current fifth harmonic amplitude and phase information, and zero-sequence current active component amplitude and phase information, and the transient characteristic quantity of step S3: fault first half-wave amplitude and phase information, as input values ​​of the deep network to train the deep neural network;

[0054] S403, after the training is completed, the best model training parameters are saved, and the verification set is input into the single-phase grounding fault diagnosis model of the best model parameters to complete the single-phase grounding fault diagnosis.

[0055] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0056] 1. A single-phase grounding fault diagnosis method that integrates steady-state characteristics and transient characteristics provided by an embodiment of the present invention adopts a technical method that combines steady-state characteristics and transient characteristics, extracts steady-state information through frequency deviation processing, and calculates transient information based on equally spaced sampling signals, and constructs a deep neural network structure model as an input layer, which can effectively solve the problems existing in traditional fault analysis, such as over-reliance on singular value effective order judgment, inability to implement Fourier transform when the signal does not meet the absolute integrability, and inability to reflect the damping characteristics and instantaneous frequency of oscillation due to the limitation of data window in analysis accuracy;

[0057] 2. A single-phase ground fault diagnosis method that integrates steady-state characteristics and transient characteristics provided by an embodiment of the present invention can effectively solve the modal aliasing phenomenon of zero-sequence current signals in the process of empirical mode decomposition by extracting steady-state information through frequency deviation processing;

[0058] 3. A single-phase grounding fault diagnosis method that integrates steady-state characteristics and transient characteristics provided by an embodiment of the present invention calculates transient information based on equally spaced sampling signals, calculates the amplitude and phase of the filtered signal, and has strong real-time performance;

[0059] 4. A single-phase grounding fault diagnosis method that integrates steady-state characteristics and transient characteristics provided by an embodiment of the present invention can largely eliminate the influence of noise on monitoring data, realize synchronous tracking of characteristic quantities of distribution network current signals, and use a deep neural network structure model to identify characteristic quantities, thereby increasing the accuracy and anti-interference ability of identifying single-phase grounding faults in distribution network cables;

[0060] 5. A single-phase grounding fault diagnosis method that integrates steady-state characteristics and transient characteristics provided in an embodiment of the present invention can improve the anti-interference capability of the fault diagnosis system and filter out the influence of noise on the monitored current data; integrating steady-state characteristics and transient characteristics as the input of a deep neural network system greatly improves the fault identification accuracy of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 A flow chart of a distribution network cable single-phase grounding fault diagnosis method that integrates steady-state characteristics and transient characteristics provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0064] In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is apparent to those of ordinary skill in the art that these specific details do not have to be employed to practice the present invention. In other embodiments, in order to avoid confusing the present invention, well-known structures are not specifically described.

[0065] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment," "an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. In addition, it will be appreciated by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0066] In the description of the present invention, the directions or positional relationships indicated by terms such as “front”, “rear”, “left”, “right”, “up”, “down”, “vertical”, “horizontal”, “high”, “low”, “inside” and “outside” are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the scope of protection of the present invention.

[0067] Example 1

[0068] like Figure 1 As shown, an embodiment of the present invention provides a single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics, comprising the following steps:

[0069] S1, obtaining online real-time data of faulty lines and non-faulty lines in the power dispatching automation system;

[0070] S2, extracting the amplitude and phase information of the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic and the zero-sequence current active component based on frequency deviation processing to obtain the steady-state information of the fault line;

[0071] S3, calculating the amplitude and phase information of the first half-wave signal of the fault based on the equally spaced sampling signal as the transient information of the single-phase grounding fault;

[0072] S4, uses the fault feature values ​​extracted by S2 and S3 as the input layer to build a deep neural network structure model, and optimizes its learning function and learning rate. After the model training is completed, the optimal model parameters are saved, and the verification set is input into the single-phase grounding fault diagnosis model with the optimal model parameters to complete the single-phase grounding fault diagnosis.

[0073] In one or more embodiments of the present invention, in step S1, the online real-time data includes zero-sequence current and first half-wave current of the fault line and non-fault line collected by the power dispatching automation system.

[0074] In one or more embodiments of the present invention, in step S2, the amplitude and phase angle information of the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic, and the zero-sequence current active component are extracted based on the frequency deviation processing to obtain the steady-state information of the fault line. The specific steps are as follows:

[0075] S201, filtering and denoising the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic and the zero-sequence current active component signal x(t) to obtain a noise-filtered signal s(t);

[0076] S202, performing Fourier transform on the signal s(t) to obtain the frequency of any mode and the amplitude of the corresponding Fourier spectrum;

[0077] S203, determining whether the mode has a modal aliasing phenomenon, if so, performing frequency deviation processing on the signal s(t) using a frequency deviation processing method; if not, no processing is performed;

[0078] S204, restore the intrinsic mode function, and then perform Hilbert transformation on the dominant intrinsic mode function component to obtain its instantaneous amplitude and instantaneous phase angle information.

[0079] In one or more embodiments of the present invention, in step S203, the specific calculation of the frequency deviation processing method is as follows:

[0080] There is a signal f(t) containing two oscillation modes x1 and x2, whose frequency components are f1 and f2 respectively. After the signal s(t) is transformed by Hilbert, the signal f is obtained. H (t);

[0081] The frequency component of s(t) is deviated by the carrier signal with a carrier frequency of F = 2f1-f2, and the single-sideband modulated signal is obtained as

[0082] f LSSB(t) = Re[(f(t)+jf H (t))e -j2πFt ] (twenty one)

[0083] f USSB (t) = Re[(f(t)+jf H (t))e j2πFt ] (twenty two)

[0084] Among them, f LSSB (t) is the lower sideband signal expression of the single sideband modulation signal, f USSB (t) is the upper sideband signal expression of the single sideband modulation signal, j is the imaginary part calculation unit;

[0085] After the frequency deviation, the frequency of the oscillation mode x1 is

[0086] f 11 =F-f1=f1-f2 (23)

[0087] The frequency of the oscillation mode x2 is

[0088] f 12 =F-f2=2(f1-f2)=2f 11 (twenty four)

[0089] In one or more embodiments of the present invention, the specific steps of step S3 are as follows:

[0090] S301, assuming that the mathematical model sequence of the first half-wave signal of the fault is

[0091]

[0092] In the formula, ω i =2πf i , k is the number of harmonics and A i 、f i and θ i Respectively represent the amplitude, frequency and phase of the ith harmonic, and x(n) represents the noise signal in the first half-wave signal of the fault

[0093] S302, the signal y(n) is subjected to noise reduction processing, and the sampling signal sequence y'(n) is reconstructed. The mathematical model of the first half-wave signal of the fault after noise reduction processing is expressed as follows:

[0094]

[0095] In the formula, A i 、f i and θ i represent amplitude, frequency and phase respectively, * Indicates taking the conjugate of this quantity, mi * Represented as m i The conjugate of i * Represented as v i The conjugation of the quantity, j is the unit for calculating the imaginary part;

[0096] S303, define the sample function as

[0097]

[0098] Wherein, i and j represent the i-th row and j-th column of the matrix respectively, y*(x) represents the conjugate of the function y(x), k represents the number of input measurement signal estimation values, n=k represents the summation starts from n equals k, and N-1 represents the summation stops when n=N-1;

[0099] According to the signal after filtering and the effective order L (2k=L) of the obtained signal, a sample matrix is ​​constructed using the sample function, and a normal equation is constructed through the sample matrix;

[0100] S304, according to the constructed normal equation, solve the homogeneous linear equation system with constant coefficients, and the solution of the equation system is the coefficient make and

[0101] S305, construct characteristic polynomial:

[0102]

[0103] In the formula, v i is the unit modulus root, a i is the coefficient, and a i =a 2k-i , i=0,1,…k;

[0104] Solve the characteristic polynomial (8) with the obtained coefficients and obtain the conjugate root pair (v i ,v i * ), i = 1, 2, ... k, and then substitute the value of the conjugate root pair into equation (6) to obtain the parameter m i (i=1,2,…k);

[0105] S306, the amplitude and phase information of the first half-wave signal of the fault are calculated by the following formula:

[0106]

[0107] In the formula, A i is the amplitude of the signal, θ i is the phase of the signal, i=1,2,…k.

[0108] In one or more embodiments of the present invention, in step S303, the normal equation is as follows:

[0109]

[0110] In the formula, r(i,j) is the sample function, is the coefficient.

[0111] In one or more embodiments of the present invention, the specific steps of step S4 are as follows:

[0112] S401, establish a deep neural network with three hidden layers, the input value of the input layer is the m fault feature values ​​extracted, recorded as column vector {x m};

[0113] S402, using the three steady-state characteristic quantities of step S2: zero-sequence current fundamental amplitude and phase information, zero-sequence current fifth harmonic amplitude and phase information, and zero-sequence current active component amplitude and phase information, and the transient characteristic quantity of step S3: fault first half-wave amplitude and phase information, as input values ​​of the deep network to train the deep neural network;

[0114] S403, after the training is completed, the best model training parameters are saved, and the verification set is input into the single-phase grounding fault diagnosis model of the best model parameters to complete the single-phase grounding fault diagnosis.

[0115] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics, characterized in that: The steps include: S1, obtain online real-time data of faulty lines and non-faulty lines in the power system; S2, extracting the amplitude and phase information of the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic and the zero-sequence current active component based on the frequency deviation processing method to obtain the steady-state information of the fault line; S3, calculating the amplitude and phase information of the first half-wave signal of the fault based on the equally spaced sampling signal as the transient information of the single-phase grounding fault; S4, uses the fault feature values ​​extracted by S2 and S3 as the input layer to build a deep neural network structure model, and optimizes the learning function and learning rate. After the model training is completed, the optimal model parameters are saved, and the verification set is input into the single-phase grounding fault diagnosis model with the optimal model parameters to complete the single-phase grounding fault diagnosis.

2. The single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics according to claim 1 is characterized in that: In step S1, the online real-time data includes zero-sequence current of the fault line and the non-fault line and the first half-wave current of the fault collected by the power system.

3. The single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics according to claim 1 is characterized in that: The specific steps of step S2 are as follows: S201, filtering and denoising the zero-sequence current fundamental wave, the zero-sequence current fifth harmonic and the zero-sequence current active component signal x(t) to obtain a noise-filtered signal s(t); S202, performing Fourier transform on the signal s(t) to obtain the frequency of any mode and the amplitude of the corresponding Fourier spectrum; S203, determining whether the mode has a modal aliasing phenomenon, if so, performing frequency deviation processing on the signal s(t); if not, performing no processing; S204, restore the intrinsic mode function, and then perform Hilbert transformation on the dominant intrinsic mode function component to obtain its instantaneous amplitude and instantaneous phase angle information.

4. The single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics according to claim 3 is characterized in that: In step S203, the specific calculation of the frequency deviation processing is as follows: There is a signal f(t) containing two oscillation modes x1 and x2, whose frequency components are f1 and f2 respectively. After the signal s(t) is transformed by Hilbert, the signal f is obtained. H (t); The frequency component of s(t) is deviated by the carrier signal with a carrier frequency of F = 2f1-f2, and the single-sideband modulated signal is obtained as f LSSB (t)=Re[(f(t)+jf H (t))e -j2πFt ] (1) f USSB (t)=Re[(f(t)+jf H (t))e j2πFt ] (2) Among them, f LSSB (t) is the lower sideband signal expression of the single sideband modulation signal, f USSB (t) is the upper sideband signal expression of the single sideband modulation signal, j is the imaginary part calculation unit; After the frequency deviation, the frequency of the oscillation mode x1 is f 11 =F-f1=f1-f2 (3) The frequency of the oscillation mode x2 is <h2 style=";text-align:left;direction:ltr">f<h2 style=";text-align:left;direction:ltr"> 12 <h2 style=";text-align:left;direction:ltr"> (F-f2) = (f1-f2) = 2f<h2 style=";text-align:left;direction:ltr"> 11 <h2 style=";text-align:left;direction:ltr"> (4) 5. The single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics according to claim 1 is characterized in that: The specific steps of step S3 are as follows: S301, assuming that the mathematical model sequence of the first half-wave signal of the fault is In the formula, ω i =2πf i , k is the number of harmonics and A i 、f i and θ i Respectively represent the amplitude, frequency and phase of the ith harmonic, and x(n) represents the noise signal in the first half-wave signal of the fault S302, the signal y(n) is subjected to noise reduction processing, and the sampling signal sequence y'(n) is reconstructed. The mathematical model of the first half-wave signal of the fault after noise reduction processing is expressed as follows: In the formula, A i 、f i and θ i represent amplitude, frequency and phase respectively, * Indicates taking the conjugate of this quantity, m i * Represented as m i The conjugate of i * Represented as v i The conjugation of the quantity, j is the unit for calculating the imaginary part; S303, define the sample function as Wherein, i and j represent the i-th row and j-th column of the matrix respectively, y*(x) represents the conjugate of the function y(x), k represents the number of input measurement signal estimation values, n=k represents the summation starts from n equals k, and N-1 represents the summation stops when n=N-1; According to the signal after filtering and the effective order L (2k=L) of the obtained signal, a sample matrix is ​​constructed using the sample function, and a normal equation is constructed through the sample matrix; S304, according to the constructed normal equation, solve the homogeneous linear equation system with constant coefficients, and the solution of the equation system is the coefficient make and S305, construct characteristic polynomial: In the formula, v i is the unit modulus root, a i is the coefficient, and a i =a 2k-i , i=0,1,…k; Solve the characteristic polynomial (8) with the obtained coefficients and obtain the conjugate root pair (v i ,v i * ), i = 1, 2, ... k, and then substitute the value of the conjugate root pair into equation (6) to obtain the parameter m i (i=1,2,…k); S306, the amplitude and phase information of the first half-wave signal of the fault are calculated by the following formula: In the formula, A i is the amplitude of the signal, θ i is the phase of the signal, i=1,2,…k.

6. The single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics according to claim 5 is characterized in that: In step S303, the normal equation is as follows: In the formula, r(i,j) is the sample function, is the coefficient.

7. The single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics according to claim 1 is characterized in that: The specific steps of step S4 are as follows: S401, establish a deep neural network with three hidden layers, the input value of the input layer is the m fault feature values ​​extracted, recorded as column vector {x m }; S402, using the three steady-state characteristic quantities of step S2: zero-sequence current fundamental amplitude and phase information, zero-sequence current fifth harmonic amplitude and phase information, and zero-sequence current active component amplitude and phase information, and the transient characteristic quantity of step S3: fault first half-wave amplitude and phase information, as input values ​​of the deep network to train the deep neural network; S403, after the training is completed, the best model training parameters are saved, and the verification set is input into the single-phase grounding fault diagnosis model of the best model parameters to complete the single-phase grounding fault diagnosis.

8. The single-phase grounding fault diagnosis method integrating steady-state characteristics and transient characteristics according to claim 7 is characterized in that: In step S401, the input value of each activation function is the sum of the product of the input value of each neuron in the previous layer and the corresponding weight value w plus the error value b, and the output value of the activation function is the input value of the next layer of neurons.

Images