A high-voltage transmission line monitoring method and system with miniaturized complex neural network

Abstract

A method and system for monitoring high-voltage transmission lines using a miniaturized complex neural network is described, characterized by the following steps: setting skip connections within a convolutional neural network architecture and searching for the optimal connection combination based on a reinforcement learning strategy; constructing an intelligent dual-branch decoupling head, which utilizes a cross-attention mechanism and a dynamic weight adjustment module to generate classification and positioning outputs. The cross-attention mechanism performs attention calculations on the differences between different types of information to obtain differential attention results, performs attention calculations on monitoring situation features to obtain situational attention results, and performs cross-calculations on the differential attention and situational attention results; coupling the classification and positioning outputs and inputting them into a backbone network, thereby obtaining monitoring situation results for the high-voltage transmission line through backbone network labeling. The present invention introduces a search strategy based on reinforcement learning to help optimize the model search and design process.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and more specifically, to a high-voltage transmission line monitoring method and system using a miniaturized complex neural network. Background Art

[0002] In the power industry, traditional high-voltage transmission line monitoring methods rely primarily on manual patrols, but this approach presents a range of challenges in terms of efficiency and accuracy. Advances in modern technology have made it possible to leverage computer vision and artificial intelligence for intelligent monitoring and identification. The application of these technologies holds enormous potential and substantial practical value. However, given that high-voltage transmission lines are often located outdoors in complex and changing environments, monitoring tasks present visual challenges such as small target detection and complex background segmentation, necessitating more sophisticated and robust neural network models.

[0003] With their widespread application in fields such as image recognition, autonomous driving, and speech recognition, deep learning models are becoming increasingly important. Improving the performance of deep learning models requires designing increasingly complex architectures, a process that typically requires significant time and practical experience. Therefore, neural architecture search (NAS) has emerged as an important solution, automatically searching for optimal neural network architectures. However, training deep learning models typically requires significant computing resources and time, posing a significant challenge for hardware devices and small businesses. Therefore, introducing search strategies based on reinforcement learning is crucial for optimizing the model search and design process and reducing model complexity.

[0004] Because drone inspection equipment for high-voltage transmission lines is limited in size and computing power, we need to optimize neural network models and minimize their size. This miniaturization is crucial for improving operational efficiency and conserving resources. However, existing technologies lack a miniaturized approach to model search and design for inspection tasks.

[0005] Traditional high-voltage transmission line monitoring often encounters challenges that cannot be solved manually, such as low efficiency and insufficient accuracy. Therefore, it is necessary to seek new solutions, and the use of computer vision and AI technologies for intelligent detection and identification is particularly important. However, the task of inspecting high-voltage transmission lines is quite daunting. The challenge lies in the fact that these lines are often located outdoors and are significantly affected by the environment. This often requires handling extremely challenging visual tasks, such as detecting small objects and segmenting complex backgrounds. Solving these problems requires more sophisticated and robust neural network models.

[0006] The abstraction and self-learning capabilities of deep learning models have made them increasingly important in numerous fields, including image recognition, autonomous driving, and speech recognition. With the continuous advancement of hardware technology, we can design more complex deep learning models to improve performance. However, manually designing an excellent model requires considerable time and experience, and even requires continuous trial and error to achieve optimal results, which undoubtedly increases the design difficulty. To address this issue, Neural Architecture Search (NAS) applies automated search technology to automatically find and determine the optimal neural network architecture, significantly reducing design time and improving efficiency. However, deep learning models still require significant computing resources and long training times, which undoubtedly increases the difficulty for small businesses with limited computing power or scarce resources. Therefore, we need to find a method to optimize the model structure and computing process to minimize training time and resource consumption.

[0007] In response to the above problems, there is an urgent need for a high-voltage transmission line monitoring method and system with a miniaturized complex neural network. Summary of the Invention

[0008] To address the deficiencies in the prior art, the present invention provides a high-voltage transmission line monitoring method and system with a miniaturized complex neural network. This system introduces a search strategy based on reinforcement learning to help optimize the model search and design process. Furthermore, through decoupling head design and multi-level knowledge distillation, the system reduces the complexity of the model while ensuring model performance.

[0009] The present invention adopts the following technical solutions.

[0010] The first aspect of the present invention relates to a high-voltage transmission line monitoring method with a miniaturized complex neural network, the method comprising the following steps: setting jump connections in a convolutional neural network architecture, and searching for an optimal connection combination based on a reinforcement learning strategy; constructing an intelligent dual-branch decoupling head, the decoupling head comprising a classification branch and a positioning branch, the classification branch being used to provide difference features between different types of information in the monitoring information of the high-voltage transmission line, and the positioning branch being used to provide monitoring situation features in the monitoring information of the high-voltage transmission line, the decoupling head generating classification output and positioning output using a cross-attention mechanism and a dynamic weight adjustment module, the cross-attention mechanism performing attention calculations on the differences between different types of information to obtain difference attention results, performing attention calculations on the monitoring situation features to obtain situation attention results, and performing cross operations on the difference attention results and the situation attention results; coupling the classification output and the positioning output into a backbone network, thereby obtaining a monitoring situation result of the high-voltage transmission line through backbone network labeling, the monitoring situation result including the classification and alarming of the current high-voltage transmission line situation using a variety of typical situations and risk warning rules of the high-voltage transmission line.

[0011] Preferably, a skip connection is set in the convolutional neural network architecture, including: using MAE-NAS to construct a backbone convolutional neural network, defining the skip connection matrix as A N×N , N is the number of upsampling and downsampling in the backbone convolutional neural network, and the skip connection matrix A N×N The value of all elements in is 0 or 1; when the element value a ij When it is 0, it indicates that there is no skip connection from the i-th downsampling to the j-th upsampling in the search network; when the element value a ij When it is 1, it indicates that there is a skip connection from the i-th downsampling to the j-th upsampling in the search network.

[0012] Preferably, setting a skip connection in the convolutional neural network architecture includes: for the skip connection a ij , adjust the dimension of the output result of the i-th downsampling by the connection scale factor, and input the adjusted output result of the i-th downsampling into the j-th upsampling; the connection scale factor is:

[0013]

[0014] Where T(·) is the scale of the input feature map for downsampling or upsampling;

[0015] Based on the value of the connection scale factor, the output of the i-th downsampling is connected to the input of the j-th upsampling by using a quick connection upsampling or a quick downsampling.

[0016] Preferably, based on the value of the connection scale factor, quick connection upsampling or quick downsampling is used to connect the output of the i-th downsampling and the input of the j-th upsampling, including: the quick upsampling uses the nearest neighbor method to obtain the value of the supplementary position of the high-voltage transmission line monitoring characteristic map during the quick upsampling calculation process; or, when the connection scale factor meets When the connection scale factor meets the requirement, the 2-fold upsampling of n connections is used as the quick upsampling, where n is an arbitrary integer; When , 2 times downsampling of n connections is used as the quick downsampling.

[0017] Preferably, searching for the optimal connection combination based on the reinforcement learning strategy includes: defining an action B of the reinforcement learning strategy t Modifications made by the agent to the backbone convolutional neural network at time step t include adding skip connections, removing skip connections, adjusting the type of layer between standard convolution kernel depth-separable convolutions, adjusting the size of the convolution kernel in the layer, and adjusting the type of acquisition function in the layer.

[0018] Preferably, searching for the optimal connection combination based on the reinforcement learning strategy includes: defining the reward function of the reinforcement learning strategy as the agent taking action B at time step t tAfter that, the generated new architecture G t+1 After that, the new architecture G t+1 Calculate the model objective function; the model objective function is:

[0019]

[0020] Where α, β, and γ are the weights of model accuracy, inference time, and total number of parameters, respectively;

[0021] mAP(G) is the average precision of architecture G,

[0022] Inference Time (G) is the inference time of a single image of architecture G on specific hardware.

[0023] Parameter Count(G) is the total number of parameters of architecture G,

[0024] Baseline Time and Baseline Params are the baseline quantities for inference time and total number of parameters, respectively.

[0025] Preferably, the search for the optimal connection sequence based on the reinforcement learning strategy includes: defining the reinforcement learning strategy as the probability distribution of all actions selected by the agent in a given state S; constructing a policy network using MLP, and training the policy network using the REINFORCE algorithm with the goal of maximizing the cumulative reward; using the trained policy network to output the probability distribution of the agent's selected actions in different states and the immediate R after selecting different actions. t ; To accumulate instant R t The highest goal is to obtain the optimal connection sequence, and based on the optimal connection sequence, the optimal backbone network searched by the reinforcement learning strategy is obtained.

[0026] The second aspect of the present invention relates to a high-voltage transmission line monitoring system with a miniaturized complex neural network, which is implemented using the high-voltage transmission line monitoring method with a miniaturized complex neural network described in the first aspect of the present invention; the system includes a search module, a decoupling module and a monitoring module; the search module is used to set jump connections in the convolutional neural network architecture and search for the optimal connection combination based on the reinforcement learning strategy; the decoupling module is used to construct an intelligent dual-branch decoupling head, the decoupling head includes a classification branch and a positioning branch, the classification branch is used to provide the difference characteristics between different types of information in the monitoring information of the high-voltage transmission line, and the positioning branch is used to provide the monitoring information in the high-voltage transmission line. Situation features, the decoupling head uses a cross-attention mechanism and a dynamic weight adjustment module to generate classification output and positioning output. The cross-attention mechanism performs attention calculations on the differences between different types of information to obtain difference attention results, performs attention calculations on the monitoring situation features to obtain situation attention results, and performs cross operations on the difference attention results and situation attention results. The monitoring module is used to couple the classification output and the positioning output into the backbone network, thereby obtaining the monitoring situation results of the high-voltage transmission line through the backbone network labeling. The monitoring situation results include the use of multiple typical situations and risk warning rules of the high-voltage transmission line to classify and alarm the current high-voltage transmission line situation. The third aspect of the present invention relates to a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the method described in the first aspect of the present invention.

[0027] A fourth aspect of the present invention relates to a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect of the present invention.

[0028] The beneficial effect of the present invention is that, compared with the existing technology, the present invention provides a high-voltage transmission line monitoring method and system with a miniaturized complex neural network, introduces a search strategy based on reinforcement learning, helps optimize the search and design process of the model, and at the same time reduces the complexity of the model while ensuring the model performance through technologies such as decoupling head design and multi-level knowledge distillation. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 Schematic diagram of the search space based on skip connection;

[0030] Figure 2 This is a schematic diagram of the overall network;

[0031] Figure 3 It is the decoupled head diagram;

[0032] Figure 4 is a classification branch diagram;

[0033] Figure 5 is a depth-wise separable convolutional graph;

[0034] Figure 6 To locate the branch diagram;

[0035] Figure 7 This is a diagram of the cross attention mechanism;

[0036] Figure 8 It is a dynamic weight generation mechanism. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of the present invention clearer and more accurate, the technical solutions of the present invention are described in detail below through multiple specific embodiments. The embodiments used in the present invention are only used to explain the present invention and are not intended to limit the content of the present invention.

[0038] This paper introduces a search strategy based on reinforcement learning to help optimize the model search and design process. By using techniques such as decoupled head design and multi-level knowledge distillation, it reduces model complexity while ensuring model performance. This paper provides a more optimized, efficient, and resource-efficient deep learning model search, design, and optimization technology to better address the problems encountered in high-voltage transmission line monitoring.

[0039] A first aspect of the present invention relates to a high-voltage transmission line monitoring method using a miniaturized complex neural network, the method comprising steps 1 to 3.

[0040] Step 1: Set up skip connections in the convolutional neural network architecture and search for the optimal connection combination based on the reinforcement learning strategy.

[0041] This paper builds on the backbone neural network construction of Neural Architecture Search (MAE-NAS) and utilizes a reinforcement learning search strategy to optimize the model. The Neural Architecture Search (NAS) method defines the search space and selects the optimal model. Based on the search space of the UNet convolutional neural architecture design model, the optimal model is defined.

[0042] A key factor for neural architecture search (NAS) is that the potential search space needs to be large and diverse enough to encompass optimal designs. Therefore, the method uses UNet, a typical convolutional neural architecture, as the basis for model search because its skip connection pattern between the encoder and decoder can provide ample model structure alternatives. In addition, the configuration of the upsampling module (such as kernel size, activation function) can also provide countless possibilities when combined. The method expands the search space of the model from two aspects, namely the upsampling module and the skip connection pattern.

[0043] Setting up skip connections in the convolutional neural network architecture includes: using MAE-NAS to construct the backbone convolutional neural network, defining the skip connection matrix as A N×N , N is the number of upsampling and downsampling in the backbone convolutional neural network, and the skip connection matrix A N×N The value of all elements in is 0 or 1; when the element value a ij When it is 0, it indicates that there is no skip connection from the i-th downsampling to the j-th upsampling in the search network; when the element value a ij When it is 1, it indicates that there is a skip connection from the i-th downsampling to the j-th upsampling in the search network.

[0044] Figure 1 is a schematic diagram of the search space based on jump connections. Figure 1 As shown, assume that there are t levels of encoders and decoders, and denote them as e1, e2, ..., e N and d1,d2,…,d N Consider different skip connection modes between the encoder and decoder. To represent each mode, define a skip connection matrix A∈{0,1} N×N , is used as a mask to determine whether there will be a new residual connection between the encoder and decoder. More specifically, a ij =1 means there is i to d j The skip connection, a ij =0 means there is no such skip connection.

[0045] Setting up skip connections in the convolutional neural network architecture includes: ij , adjust the dimension of the output result of the i-th downsampling by the connection scale factor, and input the adjusted output result of the i-th downsampling into the j-th upsampling; the connection scale factor is:

[0046]

[0047] Where T(·) is the scale of the input feature map for downsampling or upsampling;

[0048] Based on the value of the connection scale factor, the output of the i-th downsampling is connected to the input of the j-th upsampling by using a quick connection upsampling or a quick downsampling.

[0049] Based on the value of the connection scale factor, the output of the i-th downsampling is connected to the input of the j-th upsampling by using quick connection upsampling or quick downsampling, including: the quick upsampling uses the nearest neighbor method to obtain the value of the supplementary position of the high-voltage transmission line monitoring characteristic map during the quick upsampling calculation process; or, when the connection scale factor meets When the connection scale factor meets the requirement, the 2-fold upsampling of n connections is used as the quick upsampling, where n is an arbitrary integer; When , 2 times downsampling of n connections is used as the quick downsampling.

[0050] Since the feature map sizes between different layers of the encoder and decoder may be different, for example, the output of a layer of the encoder is 136×136, and the input of a layer of the decoder is 200×200, it is necessary to adjust the size of the feature map through the connection scale factor to make them match.

[0051] The specific acquisition method first defines possible resizing operations, such as a series of upsampling operations such as 2x bilinear interpolation, bicubic interpolation, nearest neighbor interpolation, or downsampling operations such as convolution with shared weights. These resizing operations are combined as needed to achieve matching feature map sizes from the encoder layer to the decoder layer.

[0052] All possible scale factors are combinations of operations that can achieve matching feature map sizes between the encoder and decoder layers. For example, resizing from 64×64 to 256×256 can be achieved by upsampling twice by a factor of 2: 64→128→256, or other combinations. Each valid resizing scheme corresponds to a specific connection scale factor. Each specific resizing scheme has a unique connection scale factor. These factors are defined as options in the search space, and NAS selects the most appropriate scale factor during the search process to optimize network performance.

[0053] To structure the search process for reinforcement learning agents, the upsampling operation is decomposed into five key components: shortcut upsampling, feature transformation, activation function, kernel size, and dilation rate. Each component is assigned a range of possible options, providing choices for the search. When deciding on shortcut upsampling, choose from commonly used upsampling interpolation techniques such as bilinear interpolation, bicubic interpolation, and nearest neighbor interpolation. For feature transformation, choose from classic convolution techniques such as 2D convolution, separable convolution, and depthwise convolution. For activation functions, choose from ReLU, LeakyReLU, PReLU, and others. More choices are provided for kernel size and dilation rate, such as kernel sizes of 1×1, 3×3, or 5×5, and dilation rates of 1, 3, or 5. The combination of these flexible components increases the diversity of the upsampling module.

[0054] The skip connection matrix A is obtained through the NAS search process. The reinforcement learning RL agent explores different skip connection patterns, i.e., different matrices A, in the search space to find the optimal network structure.

[0055] Searching for the optimal connection combination based on the reinforcement learning strategy, including: defining the action B of the reinforcement learning strategy tModifications made by the agent to the backbone convolutional neural network at time step t include adding skip connections, removing skip connections, adjusting the type of layer between standard convolution kernel depth-separable convolutions, adjusting the size of the convolution kernel in the layer, and adjusting the type of acquisition function in the layer.

[0056] A reinforcement learning-based search strategy is defined. In each episode, the RL agent selects actions in the state space based on the current strategy, changing and optimizing the network architecture's upsampling modules or connectivity patterns. The resulting new architecture is used for training and evaluated on a validation set. The model's performance is fed back to the RL agent as rewards. Based on this feedback, reinforcement learning methods are used to optimize the RL agent's strategy. This continuous optimization process ends when the search time reaches a preset limit or the strategy converges, resulting in the optimal neural network architecture.

[0057] At each search step, the agent chooses to add or remove a skip connection according to the current strategy, which is equivalent to adding the corresponding a in the matrix A. ij The generated new matrix A corresponds to a specific network structure, which is then trained and evaluated on the validation set. Based on the evaluation results, i.e. rewards, the agent’s strategy is updated to prefer skip connection patterns with better performance. NAS uses reinforcement learning agents to systematically explore and evaluate different skip connection combinations in the search space. ij Setting it to 1 can lead to better model performance, thus determining the final skip connection configuration.

[0058] The NAS task is modeled as a Markov Decision Process (MDP), whose core elements include: State Space ( ), ActionSpace, ), reward function (RewardFunction, R), strategy (Policy, π).

[0059] state Indicates the configuration of the currently constructed UNet architecture at time step t. t ) includes the number of encoder and decoder layers, and the operation type of each layer (such as convolution type, activation function, etc.).

[0060] Skip connection matrix A t ∈{0,1} N×N represents the skip connection matrix at time step t, where N is the maximum number of layers in UNet. The matrix element a i,jIndicates whether there is a jump connection from the encoder layer i to the decoder layer j.

[0061] S t ={L t ,A t}

[0062] in:

[0063] L t ={l 1,t ,l 2,t ,…,l k,t}, indicating the level information of the t-th time step.

[0064] A t ∈{0,1} N×N , represents the skip connection matrix.

[0065] Action B t Represents the modification operations performed by the agent on the current UNet architecture at time step t. These operations include adding skip connections, that is, choosing to add skip connections from the encoder layer i to the decoder layer j, removing skip connections, that is, choosing to remove skip connections from the encoder layer i to the decoder layer j, and adjusting layer operations, such as changing the operation type of a specific layer, such as switching from standard convolution to depth-separable convolution, or adjusting the parameters of a specific layer, such as kernel size, activation function type, etc. Action Space Defined as the set of all possible UNet architecture modification operations:

[0066]

[0067] Among them, each B i Corresponding to the above specific operations.

[0068] Searching for the optimal connection combination based on the reinforcement learning strategy includes: defining the reward function of the reinforcement learning strategy as the agent taking action B at time step t t After that, the generated new architecture G t+1 After that, the new architecture G t+1 Calculate the model objective function; the model objective function is:

[0069]

[0070] Where α, β, and γ are the weights of model accuracy, inference time, and total number of parameters, respectively; mAP(G) is the average precision of architecture G, Inference Time(G) is the single-image inference time of architecture G on specific hardware, Parameter Count(G) is the total number of parameters of architecture G, and Baseline Time and Baseline Params are the baseline quantities for inference time and total number of parameters, respectively.

[0071] Reward R t is the action B taken by the agent at time step t t After that, the generated new architecture G t+1 The reward function comprehensively considers the detection accuracy and resource consumption of the model to guide the agent to search for a high-performance and efficient architecture.

[0072] As training progresses, the policy network is optimized using policy gradient methods (e.g., PolicyGradient), gradually learning which choices lead to higher rewards (i.e., better model performance). The initial stage may contain more exploratory choices (similar to random selection), but as the policy is optimized, the choices will tend to utilize known good choices. The generated structure must be within the predefined search space, that is, using the defined upsampling module and skip connection pattern. All skip connections must ensure that the feature map size matches through the connection scale factor. The components of the upsampling module must be selected and configured according to the predefined combination.

[0073] Figure 2 This is a schematic diagram of the overall network. Figure 2 As shown in FIG, when the e9 feature map is connected to the d8 feature map, a89=1, and the upsampled nearest neighbor difference is used to convert the 136x136 feature map into a 200x200 feature map.

[0074] The optimal connection sequence is searched based on the reinforcement learning strategy, including: defining the reinforcement learning strategy as the probability distribution of all actions selected by the agent in a given state S; using MLP to construct a policy network, and using the REINFORCE algorithm to train the policy network with the goal of maximizing the cumulative reward; using the trained policy network to output the probability distribution of the agent's selected actions in different states and the immediate R after selecting different actions t ; To accumulate instant R t The highest goal is to obtain the optimal connection sequence.

[0075] The policy π(B|S; θ) defines the probability distribution of the agent choosing action B given a state S. The policy is represented by a parameterized neural network (e.g., a multilayer perceptron, MLP) with parameters θ. Assume the policy network is an MLP whose input is the state vector S and whose output is the probability π(B|S; θ) of each action.

[0076] π(B|S; θ)=Softmax(f(S; θ))

[0077] Here, f(S;θ) is a linear transformation defined by a neural network.

[0078] The core idea of ​​the REINFORCE algorithm is to sample actions and adjust the strategy based on the rewards obtained, so that actions with high rewards are selected more frequently and actions with low rewards are selected less frequently. t Next, select Action B t The probability is π θ (B t |S t ), set the objective function to maximize the expected cumulative reward J(θ)

[0079]

[0080] Where: γ is the discount factor used to balance the importance of current rewards and future rewards, R t is the immediate reward obtained at time step (t), and T is the termination time step of the task.

[0081] In order to maximize J(θ), we need to calculate its gradient with respect to the parameter θ and use the gradient ascent method to update the parameter:

[0082]

[0083] Among them, α is the learning rate, which controls the step size of parameter update.

[0084] In each sampling process, the policy parameter θ is fixed until the trajectory sampling of the current batch is completed and the parameter is updated. At each time step t, the agent moves according to the current state S t and strategy π θ Select Action B t ,Right now:

[0085] B t ~π θ (·|S t )

[0086] Action B t It is sampled based on the probability distribution output by the policy network. Execute action B t After that, the environment returns to the next state St+1 and instant reward R t Record states, actions, and rewards to form a complete trajectory:

[0087] τ=(S0,B0,R0,S1,B1,R1,…,S T )

[0088] where T is the terminal time step of the trajectory, e.g., reaching the goal or reaching the maximum number of steps.

[0089] Repeat the above process to generate M independent trajectories, each of which is based on the current fixed strategy π θ generated.

[0090] The REINFORCE algorithm estimates the gradient by sampling multiple trajectories (i.e., multiple complete state-action sequences):

[0091]

[0092] in:

[0093] M is the number of sampled trajectories.

[0094] is the cumulative reward starting from time step t, defined as:

[0095]

[0096] Among them, the gradient part is the gradient of the policy log probability, which reflects the sensitivity of the policy parameter θ to the probability of action selection. Through this gradient, we can understand how to adjust the parameter to increase or decrease the probability of selecting a specific action. t It is the discounted sum of all future rewards starting from time step t, representing the total reward obtained after time step t. It acts as a weight to indicate the impact of the current action on future rewards. Indicates how to adjust the policy parameters based on the cumulative reward brought by the current action. t If G is high, it means that the current action brings high rewards, then the corresponding gradient will increase the probability of the action. t If ρ is low or even negative, the gradient will push the probability of the action to decrease.

[0097] Implement the REINFORCE algorithm, first according to the current strategy π θ Sample multiple trajectories from the environment, each trajectory consists of a series of states, actions, and rewards. The trajectory sampling process is usually completed through interaction with the environment, that is, at each time step t, according to the current state S t and strategy π θ Select Action Bt , then observe the next state S t+1 and reward R t . Secondly, calculate the cumulative reward G t , calculate for each time step t:

[0098] The cumulative reward (G t ),Right now:

[0099]

[0100] This cumulative reward reflects the total reward that the policy can bring when choosing actions after time step t.

[0101] The policy gradient is calculated using the gradient estimation formula. For each time step (t) in each trajectory, the gradient estimate is calculated:

[0102]

[0103] This step integrates the gradient information from all trajectories to obtain the overall adjustment direction of the policy parameters.

[0104] Update the policy parameters θ by gradient ascent. Using the calculated gradient Update the parameters according to the following rules:

[0105]

[0106] Among them, α is the learning rate, which controls the step size of each update.

[0107] By continuously updating the policy function, the policy network gradually learns the strategy of selecting the optimal action under different states, thereby effectively searching for a high-performance and efficient UNet architecture suitable for abnormal object recognition tasks in high-voltage transmission scenarios.

[0108] Step 2: Construct a smart dual-branch decoupling head. The decoupling head includes a classification branch and a positioning branch. The classification branch is used to provide the difference characteristics between different types of information in the monitoring information of the high-voltage transmission line. The positioning branch is used to provide the monitoring situation characteristics in the monitoring information of the high-voltage transmission line. The decoupling head uses the cross-attention mechanism and the dynamic weight adjustment module to generate classification output and positioning output. The cross-attention mechanism performs attention calculation on the differences between different types of information to obtain the difference attention results, performs attention calculation on the monitoring situation characteristics to obtain the situation attention results, and performs cross operations on the difference attention results and the situation attention results.

[0109] Figure 3 The connection between IFTHead and the network body and its components is as follows: Figure 3As shown in the figure, after building the main network, a decoupling head (Intelligent Fusion Two-Branch Decoupled Head, IFTHead) is constructed for the network. IFTHead consists of two main branches: the classification branch and the localization branch. In addition, it integrates a cross-attention mechanism and a dynamic weight adjustment module.

[0110] Figure 4 is a classification branch diagram. Figure 4 As shown in the figure, the classification branch receives fused feature maps output by the fusion feature extraction module in the main network. These fused feature maps integrate features from different levels in the main UNet, achieving multi-scale information fusion and providing rich semantic information and fine-grained feature representation. The classification branch uses depthwise separable convolution to effectively reduce the number of model parameters and computational complexity while maintaining efficient feature extraction capabilities. After processing by the convolutional layer, the features are input to the fully connected layer to generate a class probability distribution, ultimately outputting a probability prediction for the target class.

[0111] Figure 5 is a depth-wise separable convolutional graph. Figure 5 As shown in the figure, the classification branch receives the fused feature map from the fusion feature extraction module, processes it through a depthwise separable convolutional layer and a fully connected layer, and outputs the classification result. The features of the classification branch then interact with the localization branch through a cross-attention mechanism, achieving information sharing and complementarity.

[0112] Figure 6 is the positioning branch diagram. Figure 6 As shown in the figure, the localization branch also receives the fused feature maps output by the fused feature extraction module of the main network. These feature maps contain multi-scale information and are suitable for accurate bounding box regression. The localization branch uses grouped convolution to improve computational efficiency and model expressiveness while reducing the number of parameters. After processing by the grouped convolution layer, the features are input to the regression layer, which predicts the four parameters of the bounding box (center coordinates \(x_c\), \(y_c\), width \(w\), height \(h\)), and finally outputs the bounding box coordinate prediction.

[0113] The localization branch receives the fused feature map from the fusion feature extraction module, processes it through the grouped convolutional layer and the regression layer, and outputs the localization result. The features of the localization branch then interact with the classification branch through a cross-attention mechanism, achieving information sharing and complementarity.

[0114] Figure 7 is a diagram of the cross attention mechanism. Figure 7As shown in the figure, the cross-attention mechanism introduces an attention mechanism between the classification branch and the localization branch, allowing the two branches to focus on each other's important features, thereby enhancing information sharing and complementarity. First, the feature maps of the classification branch and the localization branch are linearly transformed to generate query, key, and value vectors, respectively. The attention weights are obtained by calculating the similarity between the query and the key and normalizing them using the Softmax function. The attention weights are applied to the value vector to generate a weighted feature representation. The classification and localization features are weightedly fused using the attention weights to generate a fused feature representation, which is further optimized through residual connections and layer normalization.

[0115] When enhancing the classification features, a linear transformation is performed to generate a query vector (Q), which represents the important features that require attention in the classification task. The feature map of the localization branch is linearly transformed to generate a key vector (K) and a value vector (V), respectively, which represent the key feature information in the localization task. By calculating the similarity between the query (Q) and the key (K), the attention weight matrix is ​​generated using the dot product method and normalized using the Softmax function. These attention weights are used to perform a weighted summation of the value (V) to generate a weighted feature representation. The weighted feature representation is re-assigned to the classification branch, strengthening the classification branch's attention to the important features of the localization branch. At the same time, the localization branch can also use its own features as Q and the features of the classification branch as K and V in a similar manner to achieve two-way information interaction and feature enhancement.

[0116] Through the cross-attention mechanism, the classification and localization branches can dynamically capture and enhance each other's important features, improving the detection performance of the overall model. This mechanism effectively promotes the flow of information between the two branches, enabling the classification and localization tasks to be optimized and supported in a coordinated manner.

[0117] Figure 8 It is a dynamic weight generation mechanism. The dynamic weight adjustment module dynamically generates weight coefficients for classification and positioning tasks based on the input fusion feature map to adaptively balance the loss function in multi-task learning and optimize the model training process. This module receives the feature map after fusion from the classification and positioning branches. Through the global average pooling layer, global feature extraction is performed on the fused feature map to generate a feature vector of fixed dimension. The feature vector is processed by several layers of fully connected layers (including ReLU activation function) to learn the high-level representation of the features. Two weight coefficients are output through a fully connected layer, corresponding to the classification loss weight w respectively. cls,i and the positioning loss weight w loc,i . Use the Sigmoid activation function to ensure that the generated weight coefficients are between (0,1).

[0118] The dynamic weight adjustment module uses a lightweight, fully connected network to adaptively generate appropriate weight coefficients based on input features, ensuring that classification and localization tasks receive optimal weights during training. This mechanism improves the model's balance and robustness in multi-task learning, avoiding the problem of a single task dominating the training process.

[0119] The fused classification and localization feature maps are passed through a dynamic weight adjustment module to calculate dynamic weights for classification and localization losses. These weights are used to balance the classification and localization loss functions during training, achieving adaptive optimization.

[0120] The output layer is responsible for integrating the outputs of the classification branch and the localization branch to generate the final detection result. When calculating the head loss, the weight coefficient generated by the dynamic weight adjustment module is used to weight the classification loss and the localization loss, which is formulated as follows:

[0121]

[0122] in:

[0123] B is the batch size.

[0124] w cls,i and w loc,i are the classification and positioning weight coefficients of the i-th sample respectively.

[0125] l cls,i is the classification loss of the i-th sample, and the cross entropy loss function is used to measure the difference between the predicted distribution and the true distribution.

[0126] l loc,i For the positioning loss of the i-th sample, the smooth L1 loss (SmoothL1Loss) is adopted, which combines the advantages of L1 loss and L2 loss. L2 loss is used for small errors and L1 loss is used for large errors, thereby reducing the sensitivity to outliers.

[0127] Loss function details:

[0128]

[0129] in:

[0130] C is the number of categories.

[0131] y i,c is the true label of sample i corresponding to category c, usually a hot single encoding.

[0132] p i,c is the probability that the model predicts that sample i belongs to category c.

[0133] The positioning loss function uses smooth L1 loss, which combines the advantages of L1 loss and L2 loss. For each bounding box parameter (center coordinate x c 、y c , the difference z between the predicted value and the true value of width w and height h i Calculate the loss:

[0134]

[0135] In target detection, four parameters are usually used to represent a bounding box, namely the x-coordinate x of the center of the bounding box c , the y-coordinate of the center of the bounding box c , the width of the bounding box, the height of the bounding box.

[0136] Therefore, for each predicted bounding box and the corresponding ground-truth bounding box, the difference between these four parameters is calculated and a smooth L1 loss is applied. Finally, the losses of the four parameters are aggregated, usually by averaging or summing them.

[0137]

[0138] The differences are calculated for each bounding box’s four parameters separately and a smooth L1 loss is applied. Finally, the losses of the four parameters are averaged or summed to obtain the total localization loss.

[0139] By dividing the classification loss and the localization loss by the dynamic weight coefficient w cls,i and w loc,i By weighting the multi-task loss, IFTHead achieves adaptive balanced optimization of multi-task losses during training. This design not only improves the synergy between classification and localization tasks, but also enhances the overall detection performance and robustness of the model.

[0140] The decoupling head achieves efficient collaboration and optimization between classification and localization tasks. It utilizes multi-layer feature fusion within the UNet main body to generate rich semantic and fine-grained features, providing high-quality input for the classification and localization branches. Deeply separable convolution and grouped convolution are used to achieve efficient feature extraction and regression, allowing the classification and localization branches to focus on each other's important features, enhancing information sharing and complementarity, thereby improving overall detection performance. Based on the fused features, weight coefficients for classification and localization losses are dynamically generated, achieving adaptive balanced optimization of multi-task losses and improving the model's training effectiveness and generalization capabilities. The head integrates classification and localization outputs, generating the final detection result through weighted fusion, and applying dynamic weight adjustment in loss calculation to achieve adaptive multi-task learning. Through this design, the intelligent dual-branch decoupling head, IFTHead, not only achieves efficient collaboration between classification and localization tasks, but also significantly improves the overall performance and robustness of the model through dynamic weight adjustment and cross-attention mechanisms, making it suitable for a variety of complex object detection scenarios.

[0141] Step 3: Couple the classification output and positioning output and input them into the backbone network, thereby obtaining the monitoring status results of the high-voltage transmission line through the backbone network marking. The monitoring status results include the classification and alarm of the current high-voltage transmission line status using multiple typical status and risk warning rules of the high-voltage transmission line.

[0142] The backbone neural network is constructed using the Neural Architecture Search (NAS) method. NAS automatically designs efficient and sophisticated neural network structures, fully exploiting the potential of the search space. Through exploration and search in complex environments, it can achieve a more optimal model structure. A cross-attention module enables information sharing between the classification and localization branches, enhancing the complementarity of feature representation. A lightweight network dynamically adjusts the weights of the classification and localization losses to achieve adaptive loss balance and improve the model's generalization capabilities. Techniques such as depthwise separable convolution and group convolution reduce model parameters and computational complexity, ensuring efficient real-time detection.

[0143] The second aspect of the present invention relates to a high-voltage transmission line monitoring system with a miniaturized complex neural network, which is implemented using a high-voltage transmission line monitoring method with a miniaturized complex neural network described in the first aspect of the present invention; the system includes a search module, a decoupling module and a monitoring module; the search module is used to set jump connections in the convolutional neural network architecture and search for the optimal connection combination based on the reinforcement learning strategy; the decoupling module is used to construct an intelligent dual-branch decoupling head, the decoupling head includes a classification branch and a positioning branch, the classification branch is used to provide the difference characteristics between different types of information in the monitoring information of the high-voltage transmission line, and the positioning branch is used to provide the monitoring information in the high-voltage transmission line. The decoupling head uses a cross-attention mechanism and a dynamic weight adjustment module to generate classification output and positioning output. The cross-attention mechanism performs attention calculation on the differences between different types of information to obtain difference attention results, performs attention calculation on the monitoring situation features to obtain situation attention results, and performs cross operations on the difference attention results and the situation attention results; the monitoring module is used to couple the classification output and the positioning output into the backbone network, thereby obtaining the monitoring situation results of the high-voltage transmission line through the backbone network labeling. The monitoring situation results include the classification and alarm of the current high-voltage transmission line situation using a variety of typical situations and risk warning rules of the high-voltage transmission line.

[0144] A third aspect of the present invention relates to a terminal, comprising a processor and a storage medium; the storage medium is used to store instructions; and the processor is used to operate according to the instructions to execute the steps of the method described in the first aspect of the present invention.

[0145] A fourth aspect of the present invention relates to a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect of the present invention.

[0146] Finally, it should be noted that the above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art will appreciate that the technical solutions of the present invention still include modifications or equivalent substitutions that may be made to the specific embodiments of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention are intended to be covered by the claims of the present invention.

Claims

1. A high-voltage transmission line monitoring method using a miniaturized complex neural network, characterized in that: The method comprises the following steps: Set up skip connections in the convolutional neural network architecture and search for the optimal connection combination based on reinforcement learning strategy; The step of setting a skip connection in a convolutional neural network architecture includes: For skip connection a ij , adjust the dimension of the output result of the i-th downsampling by the connection scale factor, and input the adjusted output result of the i-th downsampling into the j-th upsampling; The connection scale factor is: Where T(·) is the scale of the input feature map for downsampling or upsampling; Based on the value of the connection scale factor, using quick connection upsampling or quick downsampling to connect the output of the i-th downsampling with the input of the j-th upsampling; Construct an intelligent dual-branch decoupling head. The decoupling head includes a classification branch and a positioning branch. The classification branch is used to provide difference characteristics between different types of information in the monitoring information of high-voltage transmission lines. The positioning branch is used to provide monitoring situation characteristics in the monitoring information of high-voltage transmission lines. The decoupling head uses a cross-attention mechanism and a dynamic weight adjustment module to generate classification output and positioning output. The cross-attention mechanism performs attention calculations on the differences between different types of information to obtain difference attention results, performs attention calculations on the monitoring situation characteristics to obtain situation attention results, and performs cross operations on the difference attention results and situation attention results. The classification output and positioning output are coupled and input into the backbone network, so that the monitoring situation results of the high-voltage transmission line are obtained through the backbone network marking. The monitoring situation results include the classification and alarm of the current high-voltage transmission line situation using multiple typical situations and risk warning rules of the high-voltage transmission line.

2. The high-voltage transmission line monitoring method using a miniaturized complex neural network according to claim 1, characterized in that: The step of setting a skip connection in a convolutional neural network architecture includes: MAE-NAS is used to construct the backbone convolutional neural network, and the skip connection matrix is ​​defined as A N×N , N is the number of upsampling and downsampling in the backbone convolutional neural network, and the skip connection matrix A N×N The value of all elements in is 0 or 1; When the element value a ij When it is 0, it indicates that there is no skip connection from the i-th downsampling to the j-th upsampling in the search network; When the element value a ij When it is 1, it indicates that there is a skip connection from the i-th downsampling to the j-th upsampling in the search network.

3. The high-voltage transmission line monitoring method using a miniaturized complex neural network according to claim 2, characterized in that: The step of connecting the output of the i-th downsampling and the input of the j-th upsampling by using quick upsampling or quick downsampling based on the value of the connection scale factor includes: The quick upsampling adopts the nearest neighbor method to obtain the value of the supplementary position of the high-voltage transmission line monitoring characteristic map during the quick upsampling calculation process; or, When the connection scale factor meets When , n connections of 2 times upsampling are used as the quick upsampling, where n is an arbitrary integer; When the connection scale factor meets When , 2 times downsampling of n connections is used as the quick downsampling.

4. The high-voltage transmission line monitoring method using a miniaturized complex neural network according to claim 3, characterized in that: The search for the optimal connection combination based on the reinforcement learning strategy includes: Define the action B of the reinforcement learning policy t is the modification made by the agent to the backbone convolutional neural network at time step t; The modifications include adding skip connections, removing skip connections, adjusting the type of layer between standard convolution kernel depth-separable convolution, adjusting the size of the convolution kernel in the layer, and adjusting the type of acquisition function in the layer.

5. The high-voltage transmission line monitoring method using a miniaturized complex neural network according to claim 4, characterized in that: The search for the optimal connection combination based on the reinforcement learning strategy includes: Define the reward function of the reinforcement learning policy as the agent takes action B at time step t t After that, the generated new architecture G t+1 After that, the new architecture G t+1 Calculate the model objective function; The objective function of the model is: Where α, β, and γ are the weights of model accuracy, inference time, and total number of parameters, respectively; mAP(G) is the average precision of architecture G, Inference Time (G) is the single image inference time of architecture G on hardware. Parameter Count(G) is the total number of parameters of architecture G, Baseline Time and Baseline Params are the baseline quantities for inference time and total number of parameters, respectively.

6. The high-voltage transmission line monitoring method using a miniaturized complex neural network according to claim 5, characterized in that: The search for the optimal connection sequence based on the reinforcement learning strategy includes: Define the reinforcement learning policy as the probability distribution of all actions chosen by the agent in a given state S; Use MLP to construct a policy network, and use the REINFORCE algorithm to train the policy network with the goal of maximizing cumulative rewards; The trained policy network is used to output the probability distribution of the agent's selected actions under different states and the immediate R after selecting different actions. t ; To accumulate instant R t The highest goal is to obtain the optimal connection sequence, and based on the optimal connection sequence, the optimal backbone network searched by the reinforcement learning strategy is obtained.

7. A high-voltage transmission line monitoring system with a miniaturized complex neural network, characterized by: The method is implemented by using a high-voltage transmission line monitoring method with a miniaturized complex neural network as described in any one of claims 1 to 6; The system includes a search module, a decoupling module and a monitoring module; The search module is used to set skip connections in the convolutional neural network architecture and search for the optimal connection combination based on the reinforcement learning strategy; The search module sets up skip connections in the convolutional neural network architecture, including: For skip connection a ij , adjust the dimension of the output result of the i-th downsampling by the connection scale factor, and input the adjusted output result of the i-th downsampling into the j-th upsampling; The connection scale factor is: Where T(·) is the scale of the input feature map for downsampling or upsampling; Based on the value of the connection scale factor, using quick connection upsampling or quick downsampling to connect the output of the i-th downsampling with the input of the j-th upsampling; The decoupling module is used to construct a smart dual-branch decoupling head, which includes a classification branch and a positioning branch. The classification branch is used to provide difference features between different types of information in the monitoring information of the high-voltage transmission line, and the positioning branch is used to provide monitoring situation features in the monitoring information of the high-voltage transmission line. The decoupling head uses a cross-attention mechanism and a dynamic weight adjustment module to generate classification output and positioning output. The cross-attention mechanism performs attention calculation on the differences between different types of information to obtain difference attention results, performs attention calculation on the monitoring situation features to obtain situation attention results, and performs cross operations on the difference attention results and the situation attention results; The monitoring module is used to couple the classification output and the positioning output into the backbone network, thereby obtaining the monitoring situation results of the high-voltage transmission line through the backbone network marking. The monitoring situation results include the classification and alarm of the current high-voltage transmission line situation using multiple typical situations and risk warning rules of the high-voltage transmission line.

8. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images