Power transmission line partial discharge detection method and system for unmanned aerial vehicle

Abstract

The invention provides a power transmission line partial discharge detection method and system for an unmanned aerial vehicle, and the method comprises the steps: obtaining and analyzing an environment noise spectrum in real time, and constructing a mixed noise database containing wind noise and corona noise features; performing weighted sparse decomposition on the acquired environmental noise spectrum through a preset composite dictionary representing discharge signals, and synchronously separating discharge pulses and background noise to obtain corresponding characteristic signals; inputting the processed characteristic signal into a pre-trained depth separable convolutional neural network, identifying a corresponding typical discharge mode and quantifying an energy level; and three-dimensional space positioning is carried out according to the typical discharge mode and the quantitative energy level, the risk level is evaluated, and graded early warning is triggered. According to the invention, all-weather and intelligent technical support is provided for power equipment state maintenance, and the timeliness and reliability of power grid hidden danger early warning are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of partial discharge detection of transmission lines, and particularly to a method and system for partial discharge detection of transmission lines for unmanned aerial vehicles (UAVs). Background Art

[0002] Existing UAV inspections mostly rely on single-band sensors to collect original signals, and use fixed-threshold filtering or Fourier transform for noise suppression, making it difficult to cope with dynamic interference coupling problems in complex electromagnetic environments. For example: non-stationary noises such as wind noise and corona discharge highly overlap with real partial discharge signals in the time-frequency domain; fixed weight matrices cannot adapt to noise energy fluctuations under different meteorological conditions; conventional sparse dictionaries (such as a single Gabor basis) have insufficient characterization ability for impulse-type discharge characteristics. Summary of the Invention

[0003] The object of the present invention is to provide a method and system for partial discharge detection of transmission lines for UAVs, so as to solve the technical problem of how to improve the ability to cope with dynamic interference coupling in complex electromagnetic environments.

[0004] On the one hand, a method for partial discharge detection of transmission lines for UAVs is provided, including:

[0005] Obtaining and analyzing the environmental noise spectrum in real time, and constructing a mixed noise database containing wind noise and corona noise characteristics;

[0006] Performing weighted sparse decomposition on the collected environmental noise spectrum through a preset composite dictionary representing discharge signals, synchronously separating discharge pulses and background noise to obtain corresponding characteristic signals; and inputting the processed characteristic signals into a pre-trained depthwise separable convolutional neural network to identify corresponding typical discharge patterns and quantify the energy level; wherein, the depthwise separable convolutional neural network is used to identify corresponding typical discharge patterns and energy levels according to the input characteristic signals;

[0007] Performing three-dimensional space positioning according to the typical discharge patterns and quantified energy levels, and evaluating the risk level and triggering a graded warning.

[0008] Preferably, it further includes establishing an air-ground collaborative positioning system, and performing rough positioning of the discharge source through a temperature compensation positioning algorithm by pre-deployed sensors carried by the UAV cluster.

[0009] Preferably, the air-ground collaborative positioning system includes,

[0010]

[0011] wherein, P discharge represents the three-dimensional coordinates of the discharge point; N represents the number of effective sensors; c represents the electromagnetic wave propagation speed; Δt idenote the time delay of the i-th sensor signal; a denotes the temperature compensation coefficient; T i denote the ambient temperature of the i-th sensor.

[0012] Preferably, the weighted sparse decomposition of the collected ambient noise spectrum includes performing dual-tree complex wavelet baseline drift cancellation according to the following formula:

[0013]

[0014] where s clean [n] denotes the denoised signal; s denotes the original noisy signal; denotes the dual-tree complex wavelet basis function, the wavelet atom at the j-th scale and the k-th translation; <·,·> denotes the inner product operation; K denotes the number of retained coefficients.

[0015] Preferably, the weighted sparse decomposition of the collected ambient noise spectrum includes updating the composite dictionary according to the following formula:

[0016]

[0017] where Φ opt denotes the optimal composite dictionary; M denotes the number of training samples; x i denotes the i-th training sample: the vector of the denoised partial discharge signal; ai denotes the sparse coding coefficient, the representation coefficient of the sample xi under the dictionary Φ; μ denotes the sparsity weight.

[0018] Preferably, it further includes synchronously separating the discharge pulse and the background noise according to the following formula:

[0019]

[0020] where W denotes the dynamic weight matrix; ⊙ denotes the Hadamard product; y denotes the original observed signal; x denotes the target discharge signal; e denotes the noise component; Φ denotes the composite dictionary matrix; λ1,λ2 denote the regularization coefficients.

[0021] Preferably, it further includes performing three-dimensional space positioning according to the following formula:

[0022]

[0023] where denotes the velocity of the i-th particle at the k-th iteration; w denotes the inertia weight; c1,c2 denote the learning factors; r1,r2 denote random numbers; pbest i denotes the historical optimal position of the i-th particle; gbest denotes the group historical optimal position; denotes the position of the i-th particle at the k-th iteration; k denotes the current iteration number.

[0024] Preferably, it further includes an adaptive adjustment strategy for determining the inertia weight according to the following formula:

[0025] w = w min +(w max -w min )×e -10k / K

[0026] where K represents the maximum number of iterations; w min , w max represent the inertia weight boundaries.

[0027] On the other hand, a partial discharge detection system for a transmission line for a drone is also provided, which is used to implement the partial discharge detection method for a transmission line for a drone, including

[0028] a data acquisition module, which is used to acquire and analyze the environmental noise spectrum in real time and construct a mixed noise database containing wind noise and corona noise characteristics;

[0029] a data processing module, which is used to perform weighted sparse decomposition on the acquired environmental noise spectrum through a preset composite dictionary representing the discharge signal, synchronously separate the discharge pulse and the background noise to obtain the corresponding characteristic signal; and input the processed characteristic signal into a pre-trained depthwise separable convolutional neural network to identify the corresponding typical discharge pattern and quantify the energy level; wherein, the depthwise separable convolutional neural network is used to identify the corresponding typical discharge pattern and energy level according to the input characteristic signal;

[0030] an evaluation module, which is used to perform three-dimensional spatial positioning according to the typical discharge pattern and the quantified energy level, and evaluate the risk level and trigger a hierarchical warning.

[0031] Preferably, it further includes a positioning module, which is used to establish an air-ground collaborative positioning system, and perform rough positioning of the discharge source through a temperature compensation positioning algorithm by means of pre-deployed sensors and a drone cluster.

[0032] In summary, implementing the embodiments of the present invention has the following beneficial effects:

[0033] The power transmission line partial discharge detection method and system for unmanned aerial vehicles provided by the present invention achieve precise separation and feature extraction of partial discharge signals in a complex electromagnetic environment by integrating dynamic noise benchmark modeling and improved sparse representation. The synergistic effect of the adaptive weight matrix and the composite dictionary can effectively suppress strong interference sources such as wind noise and corona discharge, and improve the capture ability of weak discharge pulses. The dual-mode analysis architecture combining the depthwise separable convolutional network and the expert system realizes multi-dimensional intelligent identification of discharge patterns and real-time dynamic assessment of risk levels. The positioning strategy based on spatio-temporal joint optimization breaks through the dependence of traditional methods on the density of sensor layout, and while ensuring high-precision spatial positioning, constructs a full-process autonomous detection system from signal acquisition, feature analysis to decision response, providing all-weather and intelligent technical support for the condition-based maintenance of power equipment, and improving the timeliness and reliability of power grid hidden danger warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention, and for those of ordinary skill in the art, obtaining other drawings without creative efforts still belongs to the scope of the present invention.

[0035] Figure 1 It is a schematic diagram of the main process of a power transmission line partial discharge detection method for unmanned aerial vehicles in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings.

[0037] As Figure 1 shown, it is a schematic diagram of an embodiment of a power transmission line partial discharge detection method for unmanned aerial vehicles provided by the present invention. In this embodiment, the method includes the following steps:

[0038] Step S0, establish an air-ground collaborative positioning system, and perform rough positioning of the discharge source through the pre-deployed sensors and the unmanned aerial vehicle cluster, and use the temperature compensation positioning algorithm; construct an unmanned aerial vehicle collaborative inspection system, deploy a quadrotor unmanned aerial vehicle cluster, and carry a dual-band ultra-high frequency sensor, an omnidirectional microwave radiometer, and a three-axis magnetic field sensor; establish an air-ground collaborative positioning system.

[0039] Among them, the air-ground collaborative positioning system includes

[0040]

[0041] Among them, P dischargeIndicates the three-dimensional coordinates of the discharge point: The calculation result is a spatial coordinate vector (x, y, z), unit: meter (m); N represents the number of effective sensors: the number of UAV nodes participating in the positioning calculation (≥3); c represents the electromagnetic wave propagation speed: approximately the speed of light 3×10 8 m / s in air, and the speed factor (0.6 - 0.9) needs to be multiplied in the wire; Δt i Indicates the signal time delay of the i-th sensor: the time difference between the discharge pulse reaching the i-th UAV and the reference node, unit: second (s); a represents the temperature compensation coefficient: the empirical value is 0.02 / °C, which is used to correct the wave speed change caused by the thermal expansion of the wire; T i Indicates the ambient temperature of the i-th sensor: obtained through the infrared temperature measurement module carried by the UAV, unit: degree Celsius (°C).

[0042] Step S1, obtain and analyze the ambient noise spectrum in real time, and construct a mixed noise database containing wind noise and corona noise characteristics; for dynamic noise benchmark modeling, collect data in real time through the above-mentioned air-ground collaborative positioning system, and construct a flight environment noise database according to the collected data.

[0043] Step S2, perform weighted sparse decomposition on the collected ambient noise spectrum through a preset composite dictionary representing the discharge signal, synchronously separate the discharge pulse and the background noise, and obtain the corresponding characteristic signal; and input the processed characteristic signal into a pre-trained depthwise separable convolutional neural network to identify the corresponding typical discharge pattern and quantify the energy level; wherein, the depthwise separable convolutional neural network is used to identify the corresponding typical discharge pattern and energy level according to the input characteristic signal; by improving the sparse representation signal separation and updating the composite dictionary online, apply a weighted sparse optimization model (Sparse-ADT) that fuses adaptive noise estimation, and iteratively solve the objective function through the ADMM algorithm.

[0044] In one embodiment, the weighted sparse decomposition of the collected ambient noise spectrum includes performing dual-tree complex wavelet to eliminate baseline drift according to the following formula:

[0045]

[0046] where s clean [n] represents the denoised signal: the reconstructed signal value at the n-th sampling point; s represents the original noisy signal: the signal sequence before wavelet decomposition input. Represents the dual-tree complex wavelet basis function: the wavelet atom at the j-th scale and the k-th translation, with approximately translation invariance; <·, ·> represents the inner product operation: calculating the projection coefficient of the signal and the wavelet basis; K represents the number of retained coefficients: the number of significant coefficients selected through threshold processing.

[0047] In a specific embodiment, update the composite dictionary according to the following formula:

[0048]

[0049] Among them, Φ opt represents the optimal composite dictionary: a set of basis functions that can most sparsely represent the discharge signal learned from the training data; M represents the number of training samples: the number of historical discharge signal segments; xi represents the i-th training sample: the denoised partial discharge signal vector, dimension = signal length × 1; a i represents the sparse coding coefficient: the representation coefficient of the sample xi under the dictionary Φ; μ represents the sparsity weight: a hyperparameter that balances the signal reconstruction error and the coefficient sparsity.

[0050] Synchronously separate the discharge pulse and background noise according to the following formula:

[0051]

[0052] Among them, W represents the dynamic weight matrix: a two-dimensional matrix calculated according to the real-time noise spectrum, and the element value ∈ [0, 1], and the weight in the high-frequency noise area approaches 0; ⊙ represents the Hadamard product: multiplying the corresponding elements of the matrix (non-standard matrix multiplication); y represents the original observed signal: the time-domain signal vector collected by the UAV sensor, dimension = number of sampling points × 1; x represents the target discharge signal: the partial discharge component to be solved, including pulse amplitude and phase information; e represents the noise component: a mixed signal including environmental interference (wind noise, corona, etc.) and sensor background noise; Φ represents the composite dictionary matrix: an overcomplete dictionary composed of Gabor bases (time-frequency atoms) and impulse response bases (transient atoms), dimension = signal length × number of atoms; λ1, λ2 represent the regularization coefficients: hyperparameters that control the signal sparsity and noise suppression intensity, usually determined by cross-validation.

[0053] Step S3, perform three-dimensional space positioning according to the typical discharge pattern and quantization energy level, and evaluate the risk level and trigger a hierarchical warning.

[0054] In one embodiment, perform three-dimensional space positioning according to the following formula:

[0055]

[0056] Among them, represents the velocity of the i-th particle in the k-th iteration: determines the search direction and step size of the particle in the solution space, dimension = problem dimension (such as 3D coordinates); w represents the inertia weight: balances the global exploration and local development capabilities, and is adaptively adjusted according to the number of iterations; c1, c2 represent the learning factors: control the weights of individual experience and group experience; r1, r2 represent random numbers: random variables uniformly distributed in [0, 1], increasing the search randomness; pbest iDenote the historical optimal position of the $i$-th particle: the best solution found by this particle so far; gbest denotes the historical optimal position of the population: the globally optimal solution among all particles; Denote the position of the $i$-th particle at the $k$-th iteration: the current candidate solution, dimension = problem dimension; $k$ denotes the current iteration number: the number of iteration rounds that the algorithm has executed.

[0057] Among them, the adaptive adjustment strategy of the inertia weight is determined according to the following formula:

[0058] $w = w$ min $+(w$ max $- w$ min )$\times e$ -10k / K

[0059] Among them, $K$ denotes the maximum number of iterations: the preset termination condition, usually set according to convergence; $w$ min , $w$ max denotes the inertia weight boundary: controlling the weight decay range.

[0060] An embodiment of the present invention further provides a partial discharge detection system for a transmission line for a drone, used to implement the above-mentioned partial discharge detection method for a transmission line for a drone, including,

[0061] A data acquisition module, used to acquire and analyze the environmental noise spectrum in real time, and construct a mixed noise database containing wind noise and corona noise characteristics;

[0062] A data processing module, used to perform weighted sparse decomposition on the acquired environmental noise spectrum through a preset composite dictionary representing the discharge signal, synchronously separate the discharge pulse and the background noise to obtain the corresponding characteristic signal; and input the processed characteristic signal into a pre-trained depthwise separable convolutional neural network to identify the corresponding typical discharge pattern and quantify the energy level; among them, the depthwise separable convolutional neural network is used to identify and match the corresponding typical discharge pattern and energy level according to the input characteristic signal;

[0063] An evaluation module, used to perform three-dimensional space positioning according to the typical discharge pattern and the quantified energy level, and evaluate the risk level and trigger a graded warning.

[0064] Specifically, it further includes a positioning module, used to establish an air-ground collaborative positioning system, and perform rough positioning of the discharge source through the pre-deployed sensors and the drone cluster and through the temperature compensation positioning algorithm.

[0065] It should be noted that the system described in the above embodiment corresponds to the method described in the above embodiment. Therefore, the parts not detailed in the system described in the above embodiment can be obtained by referring to the content of the method described in the above embodiment, and will not be elaborated here.

[0066] In summary, implementing the embodiments of the present invention has the following beneficial effects:

[0067] The partial discharge detection method and system for power transmission lines provided by the present invention, through the integration of dynamic noise benchmark modeling and improved sparse representation, achieve the precise separation and feature extraction of partial discharge signals in a complex electromagnetic environment; the synergistic effect of the adaptive weight matrix and the composite dictionary can effectively suppress strong interference sources such as wind noise and corona discharge, and improve the capture ability of weak discharge pulses; the dual-mode analysis architecture combining the depthwise separable convolutional network and the expert system realizes the multi-dimensional intelligent identification of discharge patterns and the real-time dynamic assessment of risk levels; the positioning strategy based on spatio-temporal joint optimization breaks through the dependence of traditional methods on the density of sensor layout, while ensuring high-precision spatial positioning, constructs a full-process autonomous detection system from signal acquisition, feature analysis to decision response, provides all-weather and intelligent technical support for the condition maintenance of power equipment, and improves the timeliness and reliability of power grid hidden danger warning.

[0068] The above-disclosed are only the preferred embodiments of the present invention, and of course, the scope of the rights of the present invention cannot be limited thereby. Therefore, equivalent changes made according to the claims of the present invention still fall within the scope covered by the present invention.

Claims

1. A method for detecting partial discharge in transmission lines for drones, characterized in that, Including: Obtain and analyze the environmental noise spectrum in real time, and construct a hybrid noise database containing the characteristics of wind noise and corona noise; Perform weighted sparse decomposition on the collected environmental noise spectrum through a preset composite dictionary representing the discharge signal, synchronously separate the discharge pulse and the background noise, and obtain the corresponding characteristic signal; and input the processed characteristic signal into a pre-trained depthwise separable convolutional neural network to identify the corresponding typical discharge pattern and quantify the energy level; wherein, the depthwise separable convolutional neural network is used to identify and match the corresponding typical discharge pattern and energy level according to the input characteristic signal; Perform three-dimensional space positioning according to the typical discharge pattern and the quantified energy level, and evaluate the risk level and trigger a graded warning.

2. The method according to claim 1, characterized in that, It further includes establishing an air-ground collaborative positioning system, and through pre-deployed sensors and UAV clusters carried, performing rough positioning of the discharge source through a temperature compensation positioning algorithm.

3. The method according to claim 2, characterized in that The air-ground collaborative positioning system includes: Among them, P discharge represents the three-dimensional coordinates of the discharge point; N represents the number of effective sensors; c represents the electromagnetic wave propagation speed; Δt i represents the signal time delay of the i-th sensor; a represents the temperature compensation coefficient; T i represents the ambient temperature of the i-th sensor.

4. The method according to claim 3, wherein The weighted sparse decomposition of the collected environmental noise spectrum includes performing dual-tree complex wavelet to eliminate baseline drift according to the following formula: where s clean [n] represents the denoised signal; s represents the original noisy signal; represents the dual-tree complex wavelet basis function, the wavelet atom at the j-th scale and the k-th translation; <·, ·> represents the inner product operation; K represents the number of coefficients to be retained.

5. The method according to claim 4, characterized in that The weighted sparse decomposition of the collected environmental noise spectrum further includes updating the composite dictionary according to the following formula: Among them, Φ opt represents the optimal composite dictionary; M represents the number of training samples; x i represents the i-th training sample: the denoised partial discharge signal vector; a i represents the sparse coding coefficient, the representation coefficient of sample x i under the dictionary Φ; μ represents the sparsity weight.

6. The method according to claim 5, wherein It further includes synchronously separating the discharge pulse and the background noise according to the following formula: Wherein, W represents the dynamic weight matrix; ⊙ represents the Hadamard product; y represents the original observed signal; x represents the target discharge signal; e represents the noise component; Φ represents the composite dictionary matrix; λ1, λ2 represent the regularization coefficients.

7. The method according to claim 6, wherein It further includes performing three-dimensional space positioning according to the following formula: Among them, represents the velocity of the i-th particle at the k-th iteration; w represents the inertia weight; c1 and c2 represent the learning factors; r1 and r2 represent random numbers; pbest i represents the historical best position of the i-th particle; gbest represents the historical best position of the population; represents the position of the i-th particle at the k-th iteration; k represents the current iteration number.

8. The method according to claim 7, wherein It further includes determining the adaptive adjustment strategy of the inertia weight according to the following formula: w = w min +(w max -w min )×e -10k / K Among them, K represents the maximum number of iterations; w min , w max represents the inertia weight boundary.

9. A partial discharge detection system for transmission lines of a drone, which is used to implement the method according to any one of claims 1-8, characterized in that, Including: A data acquisition module for obtaining and analyzing the environmental noise spectrum in real time, and constructing a hybrid noise database containing the characteristics of wind noise and corona noise; A data processing module for performing weighted sparse decomposition on the collected environmental noise spectrum through a preset composite dictionary representing the discharge signal, synchronously separating the discharge pulse and the background noise, and obtaining the corresponding characteristic signal; and inputting the processed characteristic signal into a pre-trained depthwise separable convolutional neural network to identify the corresponding typical discharge pattern and quantify the energy level; wherein, the depthwise separable convolutional neural network is used to identify and match the corresponding typical discharge pattern and energy level according to the input characteristic signal; An evaluation module for performing three-dimensional space positioning according to the typical discharge pattern and the quantified energy level, and evaluating the risk level and triggering a graded warning.

10. The system according to claim 9, characterized in that, It further includes a positioning module for establishing an air-ground collaborative positioning system, and through pre-deployed sensors and UAV clusters carried, performing rough positioning of the discharge source through a temperature compensation positioning algorithm.

Images