Multispectral remote sensing inspection and fault diagnosis method for electric power facilities

Abstract

The invention relates to the technical field of electric power facility fault detection, and discloses an electric power facility multispectral remote sensing inspection and fault diagnosis method, and the method comprises the steps: carrying a multispectral remote sensing device through an unmanned plane platform to obtain multispectral image data of an electric power facility, and carrying out the preprocessing to extract spectral features; and constructing a spectral feature analysis model to evaluate the health condition of the electric power facility, constructing a fault diagnosis model in combination with historical fault data, and determining the fault type and position. In order to optimize the fault diagnosis process, fault diagnosis is modeled as a Markov decision process, and the accuracy and efficiency of fault diagnosis are improved by defining a state space, an action space, a reward function and an output strategy and continuously optimizing decision actions by using a reinforcement learning algorithm. According to the invention, intelligent and automatic inspection and fault diagnosis of the electric power facility are realized, powerful technical support is provided for safe and stable operation of the electric power facility, and the method has wide application prospect and important practical value.

Description

Technical Field

[0001] The present invention relates to the field of power facility fault detection, and in particular to a multi-spectral remote sensing inspection and fault diagnosis method for power facilities. Background Art

[0002] With the rapid development of the power industry and the continuous expansion of the power grid, the safe and stable operation of power facilities is particularly important. With the rapid development of remote sensing technology and drone technology, the use of drones equipped with remote sensing equipment to inspect power facilities has gradually become a research hotspot. Traditional remote sensing inspection methods are mostly based on visible light images, which are difficult to effectively detect some subtle defects or potential faults in power facilities. As an advanced remote sensing technology, multispectral remote sensing technology can obtain spectral information of the target in different bands and provide richer and more detailed ground features. Therefore, it has broad application prospects in power facility inspections.

[0003] There are still some problems with the current multispectral remote sensing inspection technology. The processing and analysis methods of multispectral image data are not yet perfect. How to effectively extract the spectral characteristics of power facilities from massive data and conduct accurate health status assessment is a problem that needs to be solved urgently. Although multispectral remote sensing technology can provide rich feature information, how to combine these features with fault diagnosis of power facilities to achieve intelligent and automated fault diagnosis is also a difficulty in current research.

[0004] The fault diagnosis process of power facilities often involves multiple factors, including the operating status of the facility, historical fault data, maintenance resources, etc. The interaction between these factors makes fault diagnosis more complicated. Traditional fault diagnosis methods are often based on experience or simple rules, which are difficult to adapt to the diversity and complexity of power facility faults. Summary of the invention

[0005] The object of the present invention is to provide a multi-spectral remote sensing inspection and fault diagnosis method for electric power facilities to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a multi-spectral remote sensing inspection and fault diagnosis method for electric power facilities, the method comprising: Step 1: Use the multispectral remote sensing equipment on the UAV platform to inspect the power facilities and obtain multispectral image data of the target area; Step 2: Preprocess the acquired multispectral image data, including correction, denoising and segmentation, to extract the spectral characteristic data of the power facilities; Step 3: Construct a spectral feature analysis model, the input of which is the preprocessed spectral feature data, and the output is the health status assessment value of the power facility; Step 4: Construct a fault diagnosis model. This model is based on the output of the spectral feature analysis model and combines the historical fault data of the power facilities to diagnose the faults of the power facilities and output the fault type and fault location. Step 5: Construct a fault diagnosis decision model, and model the fault diagnosis process of power facilities as a Markov decision process. The specific definitions include: ① State space: defined as the combination of real-time spectral feature data and health status assessment values ​​of power facilities; ② Action space: defined as a set of executable inspection and maintenance operations; ③ Reward function: designed according to the stable operation and fault reduction goals of power facilities, including reward mechanisms for state stability, fault prevention and maintenance efficiency; at the same time, combined with the health status assessment value output by the spectral feature analysis model, additional rewards are given to operations that improve health status, and additional penalties are given to operations that reduce health status; ④ Output strategy: defined as a method for selecting the optimal action under the current state, and the decision actions generated are continuously optimized through the reinforcement learning algorithm; Step 6: Perform real-time fault diagnosis based on the trained model, collect spectral feature data of power facilities in real time as model input, and select the optimal decision action based on the current status and strategy.

[0007] Preferably, the executable inspection and maintenance operation set includes: adjusting inspection frequency, changing spectrum acquisition parameters, arranging on-site inspections, performing preventive maintenance, replacing aged components and adjusting equipment operating parameters.

[0008] Preferably, the spectral feature analysis model is constructed using a convolutional neural network (CNN) algorithm, and its network structure specifically includes: Input layer: used to receive preprocessed spectral feature data; Hidden layer: contains multiple convolutional layers and pooling layers to extract high-level representations of spectral features; Output layer: The high-level features are mapped to the health status assessment value of the power facility through the fully connected layer.

[0009] Preferably, the step of training the spectral feature analysis model comprises: F101: Initialize the weights and biases of the CNN network; F102: Calculate the mean squared error loss as the loss function to measure the difference between the health status assessment value predicted by the model and the true value; F103: Use the back propagation algorithm and gradient descent optimization algorithm to calculate the gradient according to the loss function, update the weights and bias terms of the CNN network, and iteratively train the network until the loss function converges; F104: Save the trained CNN network model as a spectral feature analysis model in the multi-spectral remote sensing inspection and fault diagnosis method of power facilities.

[0010] Preferably, the fault diagnosis model is constructed using a support vector machine (SVM) algorithm, and its training steps include: F201: Collect spectral data of power facilities and corresponding health status labels to build a training dataset; F202: The support vector machine (SVM) algorithm is used to build a fault diagnosis model. The model formula is: in, is the input spectral feature vector, is the support vector, is the corresponding Lagrange multiplier, is the kernel function, is the bias term; F203: Use the training data set to train the constructed SVM model and solve the model parameters through the sequential minimum optimization (SMO) algorithm; F204: After the training is completed, the model is verified using the validation data set to evaluate the prediction performance of the model. If the requirements are not met, return to step F203 to adjust the model parameters until the model performance meets the requirements.

[0011] Preferably, the reward function R is composed of multiple sub-reward items, and the total reward value is obtained by weighted summation. The specific formula is: R = w1 * R_stab + w2 * R_prev + w3 * R_eff + w4 * R_model; wherein R_stab represents the reward item for state stability, which is calculated according to the fluctuation degree of the operating state of the power facility; R_prev represents the reward item for fault prevention, which is calculated according to the frequency and severity of faults; R_eff represents the reward item for maintenance efficiency, which is calculated according to the success rate and time consumption of maintenance operations; R_model represents the model evaluation reward item, which gives rewards or penalties according to the increase or decrease of the health status assessment value output by the spectral feature analysis model; w1, w2, w3 and w4 are the weight coefficients of each sub-reward item, respectively, and w1+w2+w3+w4=1.

[0012] Preferably, a deep deterministic policy gradient DDPG algorithm is used to train the fault diagnosis decision model.

[0013] Preferably, the step of training the fault diagnosis decision model includes: S1: Initialize the actor network and the critic network. The actor network is used to generate actions, and the critic network is used to evaluate the value of actions. S2: Initialize the experience playback buffer D, which is used to store the state transition experience generated during the inspection and fault diagnosis of power facilities; S3: For each training round: S301: Initialize the state s at the beginning of the round; S302: Before the round ends, the following steps are executed in a loop: i) the actor network selects action a according to the current state s; ii) executes action a, observes the reward r, the next state s' and the termination flag done; iii) stores the experience (s, a, r, s', done) in the experience replay buffer D; iv) randomly samples a batch of experience from the experience replay buffer D to update the network parameters; v) uses the sampled experience to update the parameters of the actor network and the critic network; vi) updates the current state s to the next state s'; S303: When the round ends, reset the state s to the new round start state; S4: Repeat step S3 until a satisfactory performance level is reached.

[0014] Preferably, step v in S302 is specifically: using the sampled experience to update the parameters of the critic network through a temporal difference learning method to minimize the prediction error of the value function.

[0015] Preferably, the multispectral image data includes spectral information of multiple bands of visible light, infrared light and ultraviolet light.

[0016] Compared with the prior art, the present invention has the following beneficial effects: By using multispectral remote sensing equipment on drone platforms, efficient and comprehensive inspections of power facilities are achieved, greatly reducing the time and labor of manual inspections, while overcoming the limitations of external factors such as the environment and weather. Multispectral remote sensing technology can obtain spectral information of power facilities in different bands, providing richer and more detailed feature data. Compared with traditional visible light image inspections, it can more effectively detect subtle defects or potential faults in power facilities.

[0017] The constructed spectral feature analysis model and fault diagnosis model can accurately assess the health status of power facilities based on the preprocessed spectral feature data, perform fault diagnosis in combination with historical fault data, and output specific fault types and locations, greatly improving the accuracy and efficiency of fault diagnosis.

[0018] Through Markov decision process modeling, the fault diagnosis process is transformed into a decision-making problem, and the reinforcement learning algorithm is used to continuously optimize the decision-making actions, making fault diagnosis more intelligent and automated, and adapting to the needs of the diversity and complexity of power facility faults.

[0019] The fault diagnosis decision model not only considers the real-time health status of power facilities, but also combines the actual situation of maintenance resources. Through the reward function and penalty mechanism, it optimizes the decision of maintenance operation and ensures the rational allocation and efficient use of maintenance resources. The spectral characteristic data of power facilities are collected in real time, and the optimal decision action is selected according to the current status and strategy, realizing dynamic monitoring and timely maintenance of power facilities, effectively preventing the occurrence of faults and reducing maintenance costs.

[0020] Through accurate health status assessment and fault diagnosis, the present invention can timely discover and handle potential faults of power facilities, avoid safety accidents such as power outages and equipment damage caused by faults, and improve the operational safety and reliability of power facilities. The continuous optimization of reinforcement learning algorithms makes fault diagnosis and maintenance decisions more accurate and efficient, providing a strong guarantee for the long-term stable operation of power facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A processing step diagram of a multi-spectral remote sensing inspection and fault diagnosis method for electric power facilities according to the present invention; Figure 2 This is the training flow chart of the spectrum feature analysis model; Figure 3 Flowchart for training a fault diagnosis decision model using the Deep Deterministic Policy Gradient (DDPG) algorithm. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] See also Figure 1-3 The present invention provides a technical solution: a multi-spectral remote sensing inspection and fault diagnosis method for electric power facilities, the method comprising: Step 1: Data acquisition: Use a drone platform equipped with multispectral remote sensing equipment to inspect power facilities. The drone flies according to the preset flight route and altitude to ensure that the multispectral remote sensing equipment can fully cover the power facilities in the target area. During the flight, the multispectral remote sensing equipment simultaneously captures spectral image data of multiple bands such as visible light, infrared, and ultraviolet, and stores it in the onboard storage device, and the data is transmitted and processed after the flight.

[0024] Step 2: Data preprocessing: Preprocess the acquired multispectral image data, including image correction to eliminate image distortion caused by factors such as lens distortion and uneven illumination; use denoising algorithms to reduce noise interference in the image; use image segmentation technology to separate power facilities from the background and extract spectral feature data of the region of interest (ROI). The preprocessed spectral feature data is stored in a standard format for subsequent analysis and processing.

[0025] Step 3: Construct a spectral feature analysis model: Construct a spectral feature analysis model based on machine learning algorithms (such as support vector machines, random forests, etc.). The model input is the preprocessed spectral feature data, and the output is the health status assessment value of the power facility. In order to train the model, it is necessary to collect a large amount of spectral feature data of power facilities with known health conditions as training samples, and optimize the model parameters through methods such as cross-validation to ensure the accuracy and generalization ability of the model.

[0026] Step 4: Construct a fault diagnosis model: Combine the historical fault data of the power facility to construct a fault diagnosis model. This model diagnoses faults of the power facility based on the output of the spectral feature analysis model, as well as the operating status of the power facility, historical maintenance records and other information. Classification algorithms (such as decision trees, neural networks, etc.) are used to identify the fault type, and the fault location is determined through positioning algorithms. The fault diagnosis model also needs to be learned and optimized through training samples.

[0027] Step 5: Construct a fault diagnosis decision model and model the fault diagnosis process of power facilities as a Markov decision process. The specific definition is as follows: State space: It consists of real-time spectral characteristic data and health status assessment values ​​of power facilities, indicating the current status of power facilities.

[0028] Action space: includes a set of executable operations such as inspection and maintenance, which represent the possible measures that may be taken on power facilities.

[0029] Reward function: designed according to the stable operation and fault reduction goals of power facilities, including state stability rewards, fault prevention rewards, maintenance efficiency rewards, etc. At the same time, combined with the health status assessment value output by the spectral feature analysis model, additional rewards are given to operations that improve health status, and additional penalties are given to operations that reduce health status.

[0030] Output strategy: Use reinforcement learning algorithm to learn the optimal strategy, that is, the method of selecting the best action under the current state. Through continuous iterative learning, the generated decision actions are optimized.

[0031] Step 6. Real-time fault diagnosis: Real-time fault diagnosis is performed based on the trained model. The spectral feature data of the power facilities are collected in real time through drone inspections as model input. The spectral feature analysis model first evaluates the health status of the power facilities, and then the fault diagnosis model performs fault diagnosis based on the health status assessment value and other relevant information. Finally, the fault diagnosis decision model selects the optimal decision action based on the current status and strategy, such as arranging maintenance, adjusting the inspection frequency, etc., to ensure the safe and stable operation of the power facilities.

[0032] The present invention will be further described below in conjunction with Examples 1 to 3: Embodiment 1: This embodiment describes in detail an efficient and intelligent multispectral remote sensing inspection and fault diagnosis method for power facilities. The specific implementation method is as follows: A. Data acquisition and processing: Use high-precision multispectral remote sensing equipment carried by the UAV platform to conduct regular or on-demand inspections of power facilities according to the preset flight plan and parameter settings. The collected multispectral image data undergoes pre-processing steps such as image correction, denoising and segmentation to extract clear and accurate spectral feature data of power facilities.

[0033] B. A set of executable inspection and maintenance operations. Based on the actual operation and maintenance requirements of power facilities, a series of executable inspection and maintenance operations are defined, including: Adjust inspection frequency: Dynamically adjust the inspection frequency based on the historical fault data, operating environment and current health status assessment of the power facilities to ensure that key facilities are monitored more intensively.

[0034] Change spectral acquisition parameters: According to weather conditions, lighting changes or the detection requirements of specific fault types, adjust the parameters of spectral acquisition (such as band selection, exposure time, etc.) to optimize data acquisition quality.

[0035] Arrange on-site inspection: When the spectral signature analysis model or fault diagnosis model indicates a potential fault, arrange for professionals to conduct an on-site inspection to confirm the fault and take appropriate measures.

[0036] Perform preventive maintenance: Based on model predictions and historical data, preventive maintenance or replacement of parts that may fail can be performed to reduce the occurrence of sudden failures.

[0037] Replace aging components: Based on the service life and health status of the components, replace aging key components in a timely manner to ensure the stable operation of power facilities.

[0038] Adjust equipment operating parameters: According to the actual operating conditions and needs of the power facilities, adjust the operating parameters of the equipment (such as voltage, current, etc.) to optimize equipment performance and extend its service life.

[0039] C. Constructing a spectral feature analysis model. The convolutional neural network (CNN) algorithm is used to construct a spectral feature analysis model. The specific design of its network structure is as follows: Input layer: Receives preprocessed spectral feature data, which are represented in the form of a multidimensional array, with each dimension corresponding to a different spectral band and spatial position.

[0040] Hidden layer: Contains multiple convolutional layers and pooling layers. The convolutional layer extracts high-level representations of spectral features, such as edges, textures, and shapes, through convolution operations; the pooling layer is used to reduce the dimension of feature data, reduce the amount of calculation, and retain important features. These layers are arranged alternately to form a deep network structure to fully mine the information in the spectral features.

[0041] Output layer: The high-level features extracted by the hidden layer are mapped to the health status assessment value of the power facility through the fully connected layer. This assessment value can be a continuous value, indicating the overall health status of the power facility; or it can be a classification label, indicating whether the power facility is in a normal, warning or fault state.

[0042] During the model training process, a large amount of labeled data (i.e., spectral feature data of power facilities with known health conditions) is used to train the CNN model, and the model parameters are optimized through the back-propagation algorithm so that the model can accurately predict the health conditions of power facilities.

[0043] D. Construct fault diagnosis model and decision model: Based on the spectral feature analysis model, the fault diagnosis model is constructed in combination with the historical fault data and operation and maintenance knowledge of the power facilities. The model uses classification or regression algorithms to diagnose faults of power facilities and outputs the fault type and fault location. The fault diagnosis process of power facilities is modeled as a Markov decision process, and the state space, action space, reward function and output strategy are defined. The generated decision actions are continuously optimized through the reinforcement learning algorithm, so that the model can select the optimal inspection and maintenance operations according to the current status and operation and maintenance requirements of the power facilities.

[0044] E. Real-time fault diagnosis and decision-making: Based on the trained model, the spectral feature data of the power facilities are collected in real time and input into the spectral feature analysis model and the fault diagnosis model. According to the output of the model, combined with the decision-making strategy of the Markov decision process, the optimal inspection and maintenance operations are selected. By continuously monitoring the status of power facilities and updating the model, real-time and accurate fault diagnosis and intelligent decision-making of power facilities are achieved to ensure the safe and stable operation of power facilities.

[0045] Embodiment 2: In order to implement the training of a spectral feature analysis model and a fault diagnosis model in a multi-spectral remote sensing inspection and fault diagnosis method for electric power facilities, this embodiment provides the following detailed steps: The convolutional neural network (CNN) algorithm is used to build a spectral feature analysis model. The specific training steps are as follows: F101. Initialize the weights and biases of the CNN network: Use a random initialization method (such as Gaussian distribution or uniform distribution) to assign initial values ​​to the weights and biases of each layer of the CNN network. These initial values ​​will serve as the starting point for network training.

[0046] F102. Calculate the loss function: Select Mean Squared Error (MSE) as the loss function to measure the difference between the health assessment value predicted by the model and the true value. The loss function formula is: in is the real health status assessment value, is the health status assessment value predicted by the model, and N is the number of samples.

[0047] F103. Network training and optimization: Use backpropagation and gradient descent optimization algorithms (such as stochastic gradient descent SGD, mini-batch gradient descent or Adam) for network training; calculate the gradient according to the loss function, that is, the partial derivative of the loss function with respect to the network weights and bias terms; use the gradient descent optimization algorithm to update the network weights and bias terms so that the loss function gradually decreases; iteratively train the network until the loss function converges or the preset number of training rounds is reached.

[0048] F104, model saving: After the network training is completed, save the trained CNN network model, including parameters such as network structure, weights and bias terms; use the saved model as a spectral feature analysis model in the multi-spectral remote sensing inspection and fault diagnosis method of power facilities for subsequent spectral feature extraction and health status assessment.

[0049] The fault diagnosis model is trained by using the support vector machine (SVM) algorithm. The specific training steps are as follows: F201. Construct training data set: Collect spectral data of power facilities and corresponding health status labels (such as normal, warning, fault, etc.). Preprocess the spectral data and extract spectral feature vectors. Combine the spectral feature vectors and health status labels into a training data set.

[0050] F202, construct SVM model: Use SVM algorithm to construct fault diagnosis model, the model formula is: Where x is the input spectral feature vector, is the support vector, is the corresponding Lagrange multiplier, is the kernel function (such as linear kernel, Gaussian kernel, etc.), and b is the bias term.

[0051] F203, Model training: Use the training data set to train the constructed SVM model. Use the Sequential Minimal Optimization (SMO) algorithm to solve the model parameters, that is, optimize the Lagrange multiplier And bias term b. Through iterative optimization, the model achieves better classification performance on the training data set.

[0052] F204, model verification and tuning: After training is completed, use the verification data set to verify the model and evaluate the model's prediction performance (such as accuracy, recall, F1 score, etc.). If the model performance does not meet the requirements, return to step F203, adjust the model parameters (such as kernel function type, penalty parameter C, etc.), and retrain the model. Repeat the verification and tuning process until the model performance meets the requirements.

[0053] Embodiment 3: This embodiment is used to describe a method for building a fault diagnosis decision model, and uses a deep deterministic policy gradient (DDPG) algorithm for training. The following are detailed implementation steps: Design reward function: The reward function R is a key part of the reinforcement learning algorithm. It is used to evaluate the contribution of the actions taken to the operation and maintenance effect of power facilities. The reward function consists of multiple sub-reward items, and the total reward value is obtained by weighted summation. The specific formula is: R = w1 * R_stab + w2 * R_prev + w3 * R_eff + w4 * R_model Where: R_stab represents the reward for state stability, which is calculated based on the degree of fluctuation of the operating state of the power facility. If the state is stable, a positive reward is given; if the state fluctuates greatly, a negative reward is given.

[0054] R_prev represents the reward for fault prevention, which is calculated based on the frequency and severity of the fault. If the fault can be effectively prevented, a positive reward will be given; if the fault occurs frequently or is of high severity, a negative reward will be given.

[0055] R_eff represents the reward item for maintenance efficiency, which is calculated based on the success rate and time of the maintenance operation. If the maintenance operation is successful and takes a short time, a positive reward is given; if the maintenance operation fails or takes a long time, a negative reward is given.

[0056] R_model represents the model evaluation reward item, which rewards or punishes according to the increase or decrease of the health assessment value output by the spectral feature analysis model. If the health assessment value increases, a positive reward is given; if the health assessment value decreases, a negative reward is given. w1, w2, w3 and w4 are the weight coefficients of each sub-reward item, and they satisfy: w1 + w2 + w3 +w4 = 1. These weight coefficients can be adjusted according to actual conditions to reflect the importance of different sub-reward items in the total reward.

[0057] Training of fault diagnosis decision model: The deep deterministic policy gradient (DDPG) algorithm is used to train the fault diagnosis decision model. The specific steps are as follows: S1. Network initialization: Initialize the actor network and the critic network. The actor network is used to generate actions, that is, to select the optimal inspection and maintenance operations based on the current state; the critic network is used to evaluate the value of actions, that is, to evaluate the contribution of the actions taken to the operation and maintenance effect of power facilities.

[0058] S2. Initialization of experience replay buffer: Initialize the experience replay buffer D, which is used to store the state transition experience generated during the inspection and fault diagnosis of power facilities, including information such as state, action, reward, next state, and whether to terminate.

[0059] S3, training round loop, for each training round, perform the following steps: S301 State initialization: Initialize the state s at the beginning of the round, which may include information such as spectrum characteristics, historical operating status, and maintenance records of the power facility.

[0060] S302 In-round step loop: Before the round ends, the following steps are looped: i) Action selection: The actor network selects action a according to the current state s, i.e., selects the optimal inspection and maintenance operation.

[0061] ii) Action execution and observation: Execute action a and observe the reward r, next state s′ and whether the end flag is done. The reward r is calculated according to the reward function, the next state s′ is updated according to the actual operating status of the power facility, and the end flag done is used to determine whether the round is over.

[0062] iii) Experience storage: The experience (s, a, r, s′, done) is stored in the experience replay buffer D for subsequent use in updating network parameters.

[0063] iv) Experience sampling and network update: A batch of experiences are randomly sampled from the experience replay buffer D to update the parameters of the actor network and the critic network. An optimization algorithm such as gradient descent is used to calculate the loss function based on the sampled experience and update the network parameters.

[0064] v) State update: Update the current state s to the next state s′ to select and execute the next action.

[0065] S303 Round termination processing: When the round ends (that is, done is true), reset the state s to the new round start state and prepare to start the next round of training.

[0066] S4, training termination condition: repeat step S3 until a satisfactory performance level is achieved. The performance level can be determined by evaluating the performance of the model on the validation dataset, such as when the accuracy, recall, F1 score and other indicators reach the preset threshold, or when the loss function during training converges to a smaller value and no longer changes significantly.

[0067] Step v in S302 is specifically: using the sampled experience to update the parameters of the critic network through a temporal difference learning method to minimize the value function prediction error.

[0068] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0069] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-spectral remote sensing inspection and fault diagnosis method for electric power facilities, characterized in that: The method comprises: Step 1: Use the multispectral remote sensing equipment on the drone platform to inspect the power facilities and obtain multispectral image data of the target area; Step 2: preprocessing the acquired multispectral image data, including correction, denoising and segmentation, to extract the spectral characteristic data of the power facilities; Step 3: Construct a spectral feature analysis model, the input of which is the preprocessed spectral feature data, and the output is the health status assessment value of the power facility; Step 4: Construct a fault diagnosis model. This model is based on the output of the spectral feature analysis model and combines the historical fault data of the power facilities to diagnose the faults of the power facilities and output the fault type and fault location. Step 5: Construct a fault diagnosis decision model, and model the fault diagnosis process of power facilities as a Markov decision process. The specific definitions include: ① State space: defined as the combination of real-time spectral feature data and health status assessment values ​​of power facilities; ② Action space: defined as a set of executable inspection and maintenance operations; ③ Reward function: designed according to the stable operation and fault reduction goals of power facilities, including reward mechanisms for state stability, fault prevention and maintenance efficiency; at the same time, combined with the health status assessment value output by the spectral feature analysis model, additional rewards are given to operations that improve health status, and additional penalties are given to operations that reduce health status; ④ Output strategy: defined as a method for selecting the optimal action under the current state, and the decision actions generated are continuously optimized through the reinforcement learning algorithm; Step 6: Perform real-time fault diagnosis based on the trained model, collect spectral feature data of power facilities in real time as model input, and select the optimal decision action based on the current status and strategy.

2. A method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 1, characterized in that: The executable inspection and maintenance operation set includes: adjusting inspection frequency, changing spectrum acquisition parameters, arranging on-site inspections, performing preventive maintenance, replacing aging parts, and adjusting equipment operating parameters.

3. A method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 1, characterized in that: The spectral feature analysis model is constructed using a convolutional neural network (CNN) algorithm, and its network structure specifically includes: Input layer: used to receive preprocessed spectral feature data; Hidden layer: contains multiple convolutional layers and pooling layers to extract high-level representations of spectral features; Output layer: The high-level features are mapped to the health status assessment value of the power facility through the fully connected layer.

4. A method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 3, characterized in that: The steps to train the spectral feature analysis model include: F101: Initialize the weights and biases of the CNN network; F102: Calculate the mean squared error loss as the loss function to measure the difference between the health assessment value predicted by the model and the true value; F103: Use the back propagation algorithm and gradient descent optimization algorithm to calculate the gradient according to the loss function, update the weights and bias terms of the CNN network, and iteratively train the network until the loss function converges; F104: Save the trained CNN network model as a spectral feature analysis model in the multi-spectral remote sensing inspection and fault diagnosis method of power facilities.

5. The method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 1, characterized in that: The fault diagnosis model is constructed using a support vector machine (SVM) algorithm, and its training steps include: F201: Collect spectral data of power facilities and corresponding health status labels to build a training dataset; F202: The support vector machine (SVM) algorithm is used to build a fault diagnosis model. The model formula is: in, is the input spectral feature vector, is the support vector, is the corresponding Lagrange multiplier, is the kernel function, is the bias term; F203: Use the training data set to train the constructed SVM model and solve the model parameters through the sequential minimum optimization (SMO) algorithm; F204: After the training is completed, the model is verified using the validation data set to evaluate the prediction performance of the model. If the requirements are not met, return to step F203 to adjust the model parameters until the model performance meets the requirements.

6. A method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 1, characterized in that: The reward function R is composed of multiple sub-reward items, and the total reward value is obtained by weighted summation. The specific formula is: R = w1 * R_stab + w2 * R_prev + w3 * R_eff + w4 * R_model; wherein R_stab represents the reward item for state stability, which is calculated according to the degree of fluctuation of the operating state of the power facility; R_prev represents the reward item for fault prevention, which is calculated according to the frequency and severity of faults; R_eff represents the reward item for maintenance efficiency, which is calculated according to the success rate and time consumption of maintenance operations; R_model represents the model evaluation reward item, which gives rewards or penalties according to the increase or decrease of the health status assessment value output by the spectral feature analysis model; w1, w2, w3 and w4 are the weight coefficients of each sub-reward item, and w1+w2+w3+w4=1.

7. A method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 6, characterized in that: The deep deterministic policy gradient DDPG algorithm is used to train the fault diagnosis decision model.

8. A method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 7, characterized in that: The steps to train a fault diagnosis decision model include: S1: Initialize the actor network and the critic network. The actor network is used to generate actions, and the critic network is used to evaluate the value of actions. S2: Initialize the experience playback buffer D, which is used to store the state transition experience generated during the inspection and fault diagnosis of power facilities; S3: For each training round: S301: Initialize the state s at the beginning of the round; S302: Before the round ends, the following steps are executed in a loop: i) the actor network selects action a according to the current state s; ii) executes action a, observes the reward r, the next state s' and the termination flag done; iii) stores the experience (s, a, r, s', done) in the experience replay buffer D; iv) randomly samples a batch of experience from the experience replay buffer D to update the network parameters; v) uses the sampled experience to update the parameters of the actor network and the critic network; vi) updates the current state s to the next state s'; S303: When the round ends, reset the state s to the new round start state; S4: Repeat step S3 until a satisfactory performance level is reached.

9. A method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 8, characterized in that: Step v in S302 is specifically: using the sampled experience to update the parameters of the critic network through a temporal difference learning method to minimize the value function prediction error.

10. The method for multi-spectral remote sensing inspection and fault diagnosis of electric power facilities according to claim 1, characterized in that: The multispectral image data includes spectral information of multiple bands of visible light, infrared light and ultraviolet light.

Images