Control method and system for preventing bird droppings flashover of power transmission line tower

Abstract

The embodiment of the invention relates to the technical field of power transmission line maintenance, in particular to a control method and system for preventing bird droppings flashover of a power transmission line tower, and the method comprises the steps: building an intelligent risk prediction model, carrying out the fusion and quantitative analysis of multi-dimensional complex information, such as bird activities, environmental factors, tower structures, etc. The design of the anti-bird thorn is established on the scientific basis of calculation and reproducibility for the first time, the design mode that a traditional anti-bird thorn depends on experience of old people, general rules or simple table look-up is changed, and key environment parameters such as altitude and air density serve as direct input of a model; and the non-linear relationship between the parameters and the flashover risk is learned by using the pre-training model, so that the design scheme can automatically adapt to different geographical climate conditions such as plains and plateaus, and the problem of poor protection effect of a traditional fixed parameter design in a high-altitude area is solved.

Description

Technical Field

[0001] This application relates to the field of power transmission line maintenance technology, and in particular to a method and system for preventing flashover of bird droppings on power transmission line towers. Background Technology

[0002] With the continuous expansion of high-voltage transmission lines, the impact of bird activity on the safe operation of these lines is becoming increasingly prominent. One of the most dangerous and typical problems is the flashover hazard caused by bird droppings. Numerous studies and operational statistics show that birds often perch, nest, or roost near tower crossarms and insulator strings, and their droppings often fall freely in long, thin strands. When bird droppings adhere to the surface of insulators or bridge the gaps between conductors and fittings, they significantly alter the local electric field distribution and reduce the pollution creepage distance, thereby inducing flashover and even causing line tripping accidents. Especially in mid- to high-altitude areas, the reduced air density lowers the withstand voltage of insulators, making the probability of bird droppings flashover significantly higher than in plains areas. These non-fault-related power outages caused by bird activity have become a problem that transmission line operation and maintenance systems urgently need to solve.

[0003] Currently, the main measures used domestically and internationally to suppress bird droppings flashover include: installing bird spikes to prevent birds from landing on critical areas; setting up bird barriers to guide birds to land away from insulation areas; using bird deterrents or audio-visual equipment to reduce bird approach; and improving the pollution resistance level by modifying the insulator structure or coating it with hydrophobic materials. Among these, bird spikes are widely used in transmission lines due to their simple structure, low cost, and easy installation. Bird spikes are usually made of metal or polymer materials and are arranged radially. The spikes create a deterrent effect, making it difficult for birds to stand stably on the crossarm or fitting surface, thereby reducing the probability of bird droppings falling into the insulation gaps by gravity.

[0004] Existing research on the electrical properties of bird guano, a unique pollutant, and its flashover distance at different altitudes is limited. The morphology of bird guano exhibits significant uncertainty; its length, water content, and conductivity are influenced by various factors, including bird species, dietary habits, excretion posture, and meteorological conditions. At increasing altitude, air density decreases and electric field distortion intensifies, making it easier for strings of bird guano of the same size to induce discharge channels. Current designs do not systematically calibrate the flashover distance of bird guano based on altitude, thus lacking a targeted model for calculating the protection radius against bird spikes, resulting in inadequate protection in high-altitude transmission lines.

[0005] Traditional bird spike design parameters (such as spike length and spacing) are mostly based on general experience or simple rules, failing to establish a quantitative relationship with the physical mechanism of bird droppings flashover (such as the trajectory of bird droppings and electric field distortion) and specific environmental conditions (such as altitude and wind speed). The risk of bird droppings flashover is the result of nonlinear coupling of multiple factors such as bird behavior, tower structure, and meteorological environment, which is difficult to predict accurately using traditional methods. Summary of the Invention

[0006] To address this, the present invention provides a method and system for preventing flashover of bird droppings on transmission line towers. This method overcomes the limitations of existing technologies where the design parameters of traditional bird-proof spikes (such as spike length and spacing) are mostly based on general experience or simple rules, failing to establish a quantitative relationship with the physical mechanisms of bird droppings flashover (such as the trajectory of bird droppings and electric field distortion) and specific environmental conditions (such as altitude and wind speed). The risk of bird droppings flashover is the result of nonlinear coupling of multiple factors such as bird behavior, tower structure, and meteorological environment, which traditional methods cannot accurately predict.

[0007] To achieve the above objectives, in a first aspect, the present invention provides a method for preventing flashover of bird droppings on transmission line towers. This method includes: Step S1: Obtain the environmental parameters, structural parameters, and historical bird droppings observation data of the target tower. Step S2: Based on the environmental parameters, tower structural parameters, and historical bird droppings landing point observation data, calculate the probability distribution of bird droppings landing points on the target tower and the basic flashover risk; Step S3: Input the landing point probability distribution, basic flashover risk, environmental parameters and tower structure parameters into the pre-trained intelligent risk prediction model to obtain the flashover risk prediction value for the target tower and the recommended initial parameters for bird spikes. The initial parameters for bird spikes include spike length, spike spacing and installation position. Step S4: Perform risk fusion processing based on the predicted flashover risk value and the basic flashover risk to obtain the final flashover risk value, and determine whether the final flashover risk value exceeds a preset risk threshold. Step S5: If the final flashover risk value exceeds the preset risk threshold, then based on the final flashover risk value, the initial parameters of the bird spikes are iteratively optimized using an optimization algorithm to generate optimized bird spike parameters. If the final flashover risk value does not exceed the preset risk threshold, the initial parameters of the bird spikes will be output as optimized parameters for the bird spikes. Step S6: Based on the bird spike optimization parameters, configure a bird spike device on the target tower.

[0008] Furthermore, in step S2, the landing point of bird droppings is calculated using the following formula. :

[0009]

[0010]

[0011] in, The starting point for bird droppings. It is the acceleration due to gravity. For flight time, V is the throwing angle for birds. b V represents the flight speed of birds. w Wind speed; The droplet landing point is corrected for air density and droplet characteristics based on the correction factor to obtain the corrected droplet landing point. .

[0012] Further, in step S2, the basic flashover risk R(Sj) is calculated using the following formula: ; in, Let the landing point probability density be... For sensitive areas, weights These are the coordinates of the sensitive area.

[0013] Furthermore, in step S3, the intelligent risk prediction model is a deep reinforcement learning model. The state vector st of the deep reinforcement learning model includes: the basic flashover risk R(Sj) of each sensitive area, wind speed Vw, wind direction θw, altitude H, tower height Ht, and conductor spacing Dc; the action vector at includes: spike length Ls, spike spacing Ds, and installation position Ps; the reward function rt of the deep reinforcement learning model is used to balance the reduction of flashover risk with the cost of protective devices.

[0014] Furthermore, the deep reinforcement learning model is trained using a deep deterministic policy gradient algorithm, which outputs action vectors through a policy network, evaluates rewards through a value network, and iteratively updates network parameters.

[0015] Further, in step S4, the risk fusion process calculates the final flashover risk value using the following formula:

[0016]

[0017] ; in, This is the final flashover risk value. This is the predicted value for flashover risk. Basic flashover risk, and These are adaptive weights.

[0018] Furthermore, in step S5, the optimization algorithm performs closed-loop iterative optimization of the initial parameters of the bird spike protection based on the final flashover risk value until the final flashover risk value is lower than the preset risk threshold.

[0019] Secondly, embodiments of this application provide a prevention system for bird droppings flashover on transmission line towers, comprising: The data acquisition module is used to obtain environmental parameters, structural parameters, and historical bird droppings observation data of the target tower. The physical landing point verification module, which is connected to the data acquisition module, is used to calculate the probability distribution of bird droppings landing on the target tower and the basic flashover risk based on the environmental parameters, tower structure parameters and historical bird droppings landing point observation data. The intelligent risk prediction module is connected to the data acquisition module and the physical landing point verification module respectively. It has a pre-trained intelligent risk prediction model built in, which is used to input the landing point probability distribution, basic flashover risk, environmental parameters and tower structure parameters into the intelligent risk prediction model to obtain the flashover risk prediction value for the target tower and the recommended initial parameters of the bird spikes. The initial parameters of the bird spikes include spike length, spike spacing and installation position. The risk fusion judgment module is connected to the physical landing point verification module and the intelligent risk prediction module respectively. It is used to perform risk fusion processing based on the flashover risk prediction value and the basic flashover risk to obtain the final flashover risk value, and to determine whether the final flashover risk value exceeds the preset risk threshold. An adaptive bird spike protection optimization module, connected to the physical impact point verification module, the intelligent risk prediction module, and the risk fusion judgment module, is used to iteratively optimize the initial parameters of the bird spike protection based on the final flashover risk value when the final flashover risk value exceeds a preset risk threshold, thereby generating optimized bird spike protection parameters. And when the final flashover risk value does not exceed the preset risk threshold, the initial parameters of the bird spikes are output as optimized parameters for the bird spikes. The protection configuration output module is connected to the adaptive bird spike optimization module to generate a scheme for configuring bird spike devices on the target tower based on the bird spike optimization parameters.

[0020] Furthermore, the intelligent risk prediction module employs a lightweight few-shot machine learning model, specifically a deep reinforcement learning model.

[0021] Furthermore, the lightweight small-sample machine learning model is jointly trained using historical data from low altitudes and small-sample monitoring data from high altitudes to predict the probability distribution of bird droppings landing points and the flashover risk index under different environmental conditions.

[0022] Compared with existing technologies, the beneficial effects of this invention are as follows: By constructing an intelligent risk prediction model, this invention integrates and quantifies multi-dimensional complex information such as bird activity, environmental factors, and tower structure, enabling bird spike design to be established on a calculable and reproducible scientific basis for the first time. This changes the traditional design mode of bird spikes that relies on experienced workers' experience, general rules, or simple table lookups. By using key environmental parameters such as altitude and air density as direct inputs to the model, and utilizing a pre-trained model to learn the nonlinear relationship between parameters and flashover risk, the design scheme can automatically adapt to different geographical and climatic conditions such as plains and plateaus. This solves the problem of poor protection effect of traditional fixed parameter designs in high-altitude areas. Through closed-loop iteration of optimization algorithms, the goal is to achieve the lowest final flashover risk value. At the same time, the scale of bird spikes is constrained by a cost function, automatically searching for the most economical and material-saving parameter scheme that meets the safety threshold, avoiding insufficient or excessive protection.

[0023] Furthermore, this invention introduces a projectile motion model that includes wind speed, wind direction, bird flight speed, and throwing angle, accurately simulating the parabolic trajectory of bird droppings in the real world. It abandons the traditional static and fixed protection range assumption, significantly improving the accuracy and realism of landing point prediction. By correcting air density and bird droppings characteristics using correction coefficients, it solves the core problem of changes in the trajectory and dispersion range of bird droppings due to thin air at high altitudes, giving the model universality and adaptability across altitudes. This invention also correlates the calculated physical landing point probability with the weak points and importance of electrical insulation through a basic flashover risk formula, providing a reliable, physically based objective function for subsequent intelligent optimization.

[0024] Furthermore, this invention takes into account deep reinforcement learning models, especially deep deterministic policy gradient algorithms, which have function approximation capabilities. Using the state vector as input, it directly learns the extremely complex and nonlinear mapping relationship between the state vector and the optimal bird spike parameter action vector, avoiding the problem that traditional empirical formulas or simple regression models cannot capture. This achieves high precision and intelligence in generating protection schemes in variable environments.

[0025] Furthermore, this invention creatively combines the basic flashover risk R(Sj) calculated based on physical laws with the flashover risk prediction value Rpred(Sj) based on data-driven prediction, effectively avoiding the limitations or biases that may exist in a single method, and significantly improving the robustness and reliability of the final risk assessment results. On the other hand, the weights w1 and w2 are not fixed values, but are dynamically allocated according to the relative magnitudes of R(Sj) and Rpred(Sj), enabling the system to intelligently judge the reliability of the information source.

[0026] Furthermore, this invention takes into account that traditional designs aim to meet basic safety thresholds, while this step, through closed-loop iteration, drives the bird spike protection parameters to continuously evolve until the optimal solution with comprehensive performance under the condition of meeting the preset risk threshold is found. This ensures the safety and efficiency of the protection scheme. By setting a clear preset risk threshold as the iteration termination condition, the reliability of the protection scheme is improved, ensuring that the final output optimized bird spike protection parameters meet safety requirements at the theoretical calculation level. This avoids the possibility of protection failure due to improper initial design and provides quantifiable protection for the safety of transmission lines.

[0027] Furthermore, this invention decomposes the complex bird spike design task into a series of standardized modules, including data acquisition, physical verification, intelligent prediction, risk fusion, parameter optimization, and solution output. The modular architecture achieves functional decoupling and professional division of labor. Each module performs its own function and collaborates closely through clear interfaces, making the system structure clear, easy to maintain and upgrade, far exceeding traditional single and chaotic design tools. Through the connection relationship between modules, the system intelligently weights the output results of deterministic calculations based on physical laws and probabilistic predictions based on historical data in the risk fusion judgment module, and finally forms a closed-loop feedback through the adaptive bird spike optimization module, constructing a decision system that combines theoretical rigor and data adaptability and continuously self-optimizes.

[0028] Furthermore, this invention unifies the conflicting objectives of safety and economy through a fitness function formula. Guided by this formula, the genetic algorithm automatically searches in the parameter space to find the optimal solution between risk and cost without human intervention, outputting the best overall performance solution. The output results include flashover risk values ​​and environmental adaptability indicators for each sensitive area, making the reliability and safety margin of the solution completely transparent, traceable, and verifiable, greatly enhancing confidence in engineering applications. Attached Figure Description

[0029] Figure 1 This is a flowchart of a method for preventing flashover of bird droppings on transmission line towers provided in this application embodiment; Figure 2 This is a flowchart of the intelligent risk prediction algorithm provided in the embodiments of this application; Detailed Implementation

[0030] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0031] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0032] Please see Figure 1-2 As shown, Figure 1 This is a flowchart of a method for preventing flashover of bird droppings on transmission line towers provided in an embodiment of this application. Figure 2 This is a flowchart of the intelligent risk prediction algorithm provided in the embodiments of this application.

[0033] The technical solution provided in this application includes the following steps: Step S1: Obtain the environmental parameters, structural parameters, and historical bird droppings observation data of the target tower. Step S2: Based on the environmental parameters, tower structural parameters, and historical bird droppings landing point observation data, calculate the probability distribution of bird droppings landing points on the target tower and the basic flashover risk; Step S3: Input the landing point probability distribution, basic flashover risk, environmental parameters and tower structure parameters into the pre-trained intelligent risk prediction model to obtain the flashover risk prediction value for the target tower and the recommended initial parameters for bird spikes. The initial parameters for bird spikes include spike length, spike spacing and installation position. Step S4: Perform risk fusion processing based on the predicted flashover risk value and the basic flashover risk to obtain the final flashover risk value, and determine whether the final flashover risk value exceeds a preset risk threshold. Step S5: If the final flashover risk value exceeds the preset risk threshold, then based on the final flashover risk value, the initial parameters of the bird spikes are iteratively optimized using an optimization algorithm to generate optimized bird spike parameters. If the final flashover risk value does not exceed the preset risk threshold, the initial parameters of the bird spikes will be output as optimized parameters for the bird spikes. Step S6: Based on the bird spike optimization parameters, configure a bird spike device on the target tower.

[0034] Specifically, in this embodiment, the optimization algorithm in step S5 can be a genetic algorithm or a particle swarm optimization algorithm.

[0035] Specifically, in this embodiment, the pre-training method for the intelligent risk prediction model is as follows: Step T1 involves constructing and preprocessing a historical dataset, specifically including: historical environmental parameter data obtained from weather stations and tower monitoring devices, including altitude H, wind speed Vw, wind direction θw, ambient temperature T, relative humidity RH, and air density ρa; design drawings and operational data for different types of towers, including tower height Ht, spatial coordinates of conductors and insulator strings (Xi, Yi, Zi), predefined flashover sensitive area coordinates Sj (such as at the insulator cap, near conductor fittings, etc.), and conductor spacing Dc; historical data on bird activity and landing parameters recorded through manual inspection, image monitoring, or drone patrols, including but not limited to: bird flight altitude Hb and flight speed Vb; initial position (x0, y0, z0) and throwing angle φ of bird droppings; and physical characteristics of bird droppings, including typical length Lf and weight mf. Step T2: Based on the historical dataset, construct the input features and output labels for model training. The input features specifically include: environmental features (altitude H, wind speed Vw, wind direction θw, temperature T, humidity RH); tower features (tower height Ht, conductor spacing Dc); and physical landing point features, that is, apply the physical landing point verification module in step S2 to each historical scenario to calculate the basic flashover risk value R(Sj) and the key statistics (such as the center and dispersion of the landing point distribution) of the bird droppings probability distribution matrix Pf(xf, yf, zf) in that scenario. The output labels include: for the risk prediction task, the label is whether a flashover or a high-risk state close to flashover actually occurred in the scenario in the historical records; for the parameter recommendation task, the label is the optimal bird spike protection parameters (including spike length Ls, spike spacing Ds, and installation position Ps) that have been verified to be effective in practice or obtained by inversion through advanced optimization algorithms in that scenario. Step T3 involves training a lightweight few-shot machine learning model and selecting a deep reinforcement learning model to model the bird spike protection design problem as a reinforcement learning task: a state vector st (i.e., the input feature vector constructed in T2); an action vector at; and a reward function rt. The training is performed using a deep deterministic policy gradient algorithm, comprising: a policy network μ(s|θμ): input state s, output deterministic action a, i.e., bird spike protection parameters; a value network Q(s, a|θQ) that evaluates the long-term cumulative reward obtained by taking action a in state s; and iteratively updating the network parameters by sampling batch data from the experience replay buffer as follows:

[0036]

[0037]

[0038] in, As a discount factor, This is the learning rate.

[0039] Step T4: Use test set data that was not used for training to evaluate the performance of the trained model. Evaluation metrics include the accuracy of risk prediction, mean squared error, and the closeness of the recommended parameters to the optimal parameters. Cross-validation is used to ensure the generalization ability of the model. The best-performing model and its parameters (network weights θμ and θQ of deep reinforcement learning) are solidified and saved to form a pre-trained intelligent risk prediction model that can be called upon.

[0040] This invention constructs an intelligent risk prediction model that integrates and quantifies multi-dimensional and complex information such as bird activity, environmental factors, and tower structure. This enables bird spike design to be based on a calculable and reproducible scientific foundation for the first time, changing the traditional design mode that relies on experienced workers' experience, general rules, or simple table lookups. By using key environmental parameters such as altitude and air density as direct inputs to the model and utilizing a pre-trained model to learn the nonlinear relationship between parameters and flashover risk, the design scheme can automatically adapt to different geographical and climatic conditions such as plains and plateaus. This solves the problem of poor protection effect of traditional fixed parameter designs in high-altitude areas. Through closed-loop iteration of optimization algorithms, the goal is to achieve the lowest final flashover risk value. At the same time, the scale of bird spikes is constrained by a cost function, and the most economical and material-saving parameter scheme under the safety threshold is automatically searched, avoiding insufficient or excessive protection.

[0041] Specifically, in step S2, the landing point of bird droppings is calculated using the following formula. :

[0042]

[0043]

[0044] in, The starting point for bird droppings. It is the acceleration due to gravity. For flight time, V is the throwing angle for birds. b V represents the flight speed of birds. w Wind speed; The droplet landing point is corrected for air density and droplet characteristics based on the correction factor to obtain the corrected droplet landing point. .

[0045] Specifically, in this embodiment, Vb is the flight speed of a bird in level flight.

[0046] Specifically, in step S2, the basic flashover risk R(Sj) is calculated using the following formula: ; in, Let the landing point probability density be... Weights are assigned to sensitive areas (such as insulator tips or fittings).

[0047] This invention introduces a projectile motion model that incorporates wind speed, wind direction, bird flight speed, and throwing angle to accurately simulate the parabolic trajectory of bird droppings in the real world. It abandons the traditional static and fixed protection range assumption, significantly improving the accuracy and realism of landing point prediction. By correcting air density and bird droppings characteristics with correction coefficients, it solves the core problem of changes in the trajectory and dispersion range of bird droppings due to thin air at high altitudes, giving the model universality and adaptability across altitudes. Furthermore, this invention correlates the calculated physical landing point probability with the weak points and importance of electrical insulation through a basic flashover risk formula, providing a reliable, physics-based objective function for subsequent intelligent optimization.

[0048] Specifically, in step S3, the intelligent risk prediction model is a deep reinforcement learning model. The state vector st of the deep reinforcement learning model includes: the basic flashover risk R(Sj) of each sensitive area, wind speed Vw, wind direction θw, altitude H, tower height Ht, and conductor spacing Dc; the action vector at includes: spike length Ls, spike spacing Ds, and installation position Ps; the reward function rt of the deep reinforcement learning model is used to balance the reduction of flashover risk with the cost of protective devices.

[0049] Specifically, the deep reinforcement learning model is trained using a deep deterministic policy gradient algorithm, outputs action vectors through a policy network, evaluates rewards through a value network, and iteratively updates network parameters.

[0050] This invention takes into account deep reinforcement learning models, especially deep deterministic policy gradient algorithms, which have function approximation capabilities. Using the state vector as input, it directly learns the extremely complex and nonlinear mapping relationship between the state vector and the optimal bird spike parameter action vector, avoiding the problem that traditional empirical formulas or simple regression models cannot capture. Thus, it achieves high precision and intelligence in generating protection schemes in variable environments.

[0051] Specifically, in step S4, the risk fusion process calculates the final flashover risk value using the following formula:

[0052]

[0053] ; in, This is the final flashover risk value. This is the predicted value for flashover risk. Basic flashover risk, and These are adaptive weights.

[0054] This invention creatively combines the basic flashover risk R(Sj) calculated based on physical laws with the flashover risk prediction value Rpred(Sj) based on data-driven prediction, effectively avoiding the limitations or biases that may exist in a single method, and significantly improving the robustness and reliability of the final risk assessment results. On the other hand, the weights w1 and w2 are not fixed values, but are dynamically allocated according to the relative magnitudes of R(Sj) and Rpred(Sj), enabling the system to intelligently judge the reliability of the information source.

[0055] Specifically, in step S5, the optimization algorithm performs closed-loop iterative optimization of the initial parameters of the bird spike protection based on the final flashover risk value until the final flashover risk value is lower than the preset risk threshold.

[0056] This invention takes into account that traditional designs aim to meet basic safety thresholds. However, this step uses closed-loop iteration to drive the bird spike protection parameters to continuously evolve until the optimal solution with comprehensive performance under the condition of meeting the preset risk threshold is found. This ensures the safety and efficiency of the protection scheme. By setting a clear preset risk threshold as the iteration termination condition, the reliability of the protection scheme is improved. This ensures that the final output bird spike protection optimization parameters meet the safety requirements at the theoretical calculation level, avoiding the possibility of protection failure due to improper initial design, and providing quantifiable protection for the safety of transmission lines.

[0057] This application provides a prevention system for bird droppings flashover on transmission line towers, comprising: The data acquisition module is used to obtain environmental parameters, structural parameters, and historical bird droppings observation data of the target tower. The physical landing point verification module, which is connected to the data acquisition module, is used to calculate the probability distribution of bird droppings landing on the target tower and the basic flashover risk based on the environmental parameters, tower structure parameters and historical bird droppings landing point observation data. The intelligent risk prediction module is connected to the data acquisition module and the physical landing point verification module respectively. It has a pre-trained intelligent risk prediction model built in, which is used to input the landing point probability distribution, basic flashover risk, environmental parameters and tower structure parameters into the intelligent risk prediction model to obtain the flashover risk prediction value for the target tower and the recommended initial parameters of the bird spikes. The initial parameters of the bird spikes include spike length, spike spacing and installation position. The risk fusion judgment module is connected to the physical landing point verification module and the intelligent risk prediction module respectively. It is used to perform risk fusion processing based on the flashover risk prediction value and the basic flashover risk to obtain the final flashover risk value, and to determine whether the final flashover risk value exceeds the preset risk threshold. An adaptive bird spike protection optimization module, connected to the physical impact point verification module, the intelligent risk prediction module, and the risk fusion judgment module, is used to iteratively optimize the initial parameters of the bird spike protection based on the final flashover risk value when the final flashover risk value exceeds a preset risk threshold, thereby generating optimized bird spike protection parameters. And when the final flashover risk value does not exceed the preset risk threshold, the initial parameters of the bird spikes are output as optimized parameters for the bird spikes. The protection configuration output module is connected to the adaptive bird spike optimization module to generate a scheme for configuring bird spike devices on the target tower based on the bird spike optimization parameters.

[0058] Specifically, the intelligent risk prediction module uses a lightweight few-shot machine learning model, specifically a deep reinforcement learning model.

[0059] Specifically, the lightweight small-sample machine learning model is jointly trained using historical data from low altitudes and small-sample monitoring data from high altitudes to predict the probability distribution of bird droppings landing points and the flashover risk index under different environmental conditions.

[0060] This invention decomposes the complex bird spike design task into a series of standardized modules, including data acquisition, physical verification, intelligent prediction, risk fusion, parameter optimization, and solution output. The modular architecture achieves functional decoupling and professional division of labor. Each module performs its own function and works closely together through clear interfaces, making the system structure clear, easy to maintain and upgrade, far exceeding traditional single and chaotic design tools. Through the connection relationship between modules, the system intelligently weights the output results of deterministic calculations based on physical laws and probabilistic predictions based on historical data in the risk fusion judgment module, and finally forms a closed-loop feedback through the adaptive bird spike optimization module, constructing a decision system that combines theoretical rigor and data adaptability and continuously self-optimizes.

[0061] Specifically, the adaptive bird spike optimization module also performs the following closed-loop iterative operation: Step 1: Based on the current bird spike parameters, trigger the physical landing point verification module to recalculate the landing point probability distribution; Step 2: Input the recalculated landing point probability distribution into the intelligent risk prediction module to update the flashover risk prediction value; Step 3: Based on the updated flashover risk prediction value, the bird spike protection parameters are optimized again through the adaptive bird spike protection optimization module; Step 4: Repeat steps 1-3 until the final flashover risk value is lower than the preset risk threshold.

[0062] Specifically, in this embodiment, the closed-loop iteration process of the bird spikes includes: initializing the bird spike parameters. ; Calculate the landing point matrix Calculate risk reward Reinforcement learning updates action vectors ; Calculate the final flashover risk value If the final flashover risk value is greater than or equal to the preset risk threshold, return to the iteration; otherwise, output the optimal bird spike protection parameters. The optimal bird spike protection parameters include: the final bird spike protection parameters. Flashover risk values ​​in each sensitive area Environmental adaptability indicators (safety margins under wind speed and humidity).

[0063] Specifically, the adaptive bird spike optimization module uses a genetic algorithm, and its fitness function is configured as follows: F in, The sum of the final flashover risks for all sensitive areas. The protection cost function is related to the length and spacing of the bird spikes. This is the cost weighting coefficient.

[0064] This invention unifies the conflicting goals of safety and economy through a fitness function formula. Guided by this formula, the genetic algorithm automatically searches in the parameter space to find the optimal solution between risk and cost without human intervention, and outputs the best overall performance solution. The output results include flashover risk values ​​and environmental adaptability indicators for each sensitive area, making the reliability and safety margin of the solution completely transparent, traceable, and verifiable, greatly enhancing confidence in engineering applications.

[0065] The above embodiments are merely illustrative examples and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this application.

Claims

1. A method for preventing flashover from bird droppings on transmission line towers, characterized in that, include: Step S1: Obtain the environmental parameters, structural parameters, and historical bird droppings observation data of the target tower. Step S2: Based on the environmental parameters, tower structural parameters, and historical bird droppings landing point observation data, calculate the probability distribution of bird droppings landing points on the target tower and the basic flashover risk; Step S3: Input the landing point probability distribution, basic flashover risk, environmental parameters and tower structure parameters into the pre-trained intelligent risk prediction model to obtain the flashover risk prediction value for the target tower and the recommended initial parameters for bird spikes. The initial parameters for bird spikes include spike length, spike spacing and installation position. Step S4: Perform risk fusion processing based on the predicted flashover risk value and the basic flashover risk to obtain the final flashover risk value, and determine whether the final flashover risk value exceeds a preset risk threshold. Step S5: If the final flashover risk value exceeds the preset risk threshold, then based on the final flashover risk value, the initial parameters of the bird spikes are iteratively optimized using an optimization algorithm to generate optimized bird spike parameters. If the final flashover risk value does not exceed the preset risk threshold, the initial parameters of the bird spikes will be output as optimized parameters for the bird spikes. Step S6: Based on the bird spike optimization parameters, configure a bird spike device on the target tower.

2. The method according to claim 1, characterized in that, In step S2, the landing point of bird droppings is calculated using the following formula. : ； ； ； in, The starting point for bird droppings. It is the acceleration due to gravity. For flight time, V is the throwing angle for birds. b V represents the flight speed of birds. w Wind speed; The droplet landing point is corrected for air density and droplet characteristics based on the correction factor to obtain the corrected droplet landing point. .

3. The method according to claim 1, characterized in that, In step S2, the basic flashover risk R(Sj) is calculated using the following formula: ； in, Let the landing point probability density be... For sensitive areas, weights These are the coordinates of the sensitive area.

4. The method according to claim 1, characterized in that, In step S3, the intelligent risk prediction model is a deep reinforcement learning model. The state vector st of the deep reinforcement learning model includes: the basic flashover risk R(Sj) of each sensitive area, wind speed Vw, wind direction θw, altitude H, tower height Ht, and conductor spacing Dc; the action vector at includes: spike length Ls, spike spacing Ds, and installation position Ps; the reward function rt of the deep reinforcement learning model is used to balance the reduction of flashover risk with the cost of protective devices.

5. The method according to claim 4, characterized in that, The deep reinforcement learning model is trained using a deep deterministic policy gradient algorithm. It outputs action vectors through a policy network, evaluates rewards through a value network, and iteratively updates network parameters.

6. The method according to claim 4, characterized in that, In step S4, the risk fusion process calculates the final flashover risk value using the following formula: ； ； ； in, This is the final flashover risk value. This is the predicted value for flashover risk. Basic flashover risk, and These are adaptive weights.

7. The method according to claim 1, characterized in that, In step S5, the optimization algorithm performs closed-loop iterative optimization of the initial parameters of the bird spike protection based on the final flashover risk value until the final flashover risk value is lower than the preset risk threshold.

8. A system for preventing flashover from bird droppings on transmission line towers, characterized in that, include: The data acquisition module is used to obtain environmental parameters, structural parameters, and historical bird droppings observation data of the target tower. The physical landing point verification module, which is connected to the data acquisition module, is used to calculate the probability distribution of bird droppings landing on the target tower and the basic flashover risk based on the environmental parameters, tower structure parameters and historical bird droppings landing point observation data. The intelligent risk prediction module is connected to the data acquisition module and the physical landing point verification module respectively. It has a pre-trained intelligent risk prediction model built in, which is used to input the landing point probability distribution, basic flashover risk, environmental parameters and tower structure parameters into the intelligent risk prediction model to obtain the flashover risk prediction value for the target tower and the recommended initial parameters of the bird spikes. The initial parameters of the bird spikes include spike length, spike spacing and installation position. The risk fusion judgment module is connected to the physical landing point verification module and the intelligent risk prediction module respectively. It is used to perform risk fusion processing based on the flashover risk prediction value and the basic flashover risk to obtain the final flashover risk value, and to determine whether the final flashover risk value exceeds the preset risk threshold. An adaptive bird spike protection optimization module, connected to the physical impact point verification module, the intelligent risk prediction module, and the risk fusion judgment module, is used to iteratively optimize the initial parameters of the bird spike protection based on the final flashover risk value when the final flashover risk value exceeds a preset risk threshold, thereby generating optimized bird spike protection parameters. And when the final flashover risk value does not exceed the preset risk threshold, the initial parameters of the bird spikes are output as optimized parameters for the bird spikes. The protection configuration output module is connected to the adaptive bird spike optimization module to generate a scheme for configuring bird spike devices on the target tower based on the bird spike optimization parameters.

9. The system according to claim 8, characterized in that, The intelligent risk prediction module uses a lightweight few-shot machine learning model, specifically a deep reinforcement learning model.

10. The system according to claim 9, characterized in that, The lightweight, small-sample machine learning model is jointly trained using historical data from low altitudes and small-sample monitoring data from high altitudes to predict the probability distribution of bird droppings landing points and the flashover risk index under different environmental conditions.

Images