A power transmission tower angle steel bearing capacity prediction method based on an african vulture intelligent algorithm
By optimizing the BP neural network using an improved African vulture intelligent algorithm and combining it with multiple optimization strategies, the problem of predicting the bearing capacity of angle steel in FRP-reinforced transmission towers was solved. This enabled rapid and accurate evaluation of the reinforcement effect, avoided structural damage, and provided theoretical support for reinforcement schemes.
Patent Information
- Application Number
- CN202311425013.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-30
- Publication Date
- 2025-12-26
- Estimated Expiration
- 2043-10-30
AI Technical Summary
Existing technologies cannot effectively predict the load-bearing capacity of FRP-reinforced transmission tower angle steel. Traditional reinforcement methods cause secondary damage to the structure, are complex to construct, and lack standardized specifications.
An improved African vulture intelligent algorithm was used to optimize the BP neural network. Combined with the slenderness ratio of angle steel and FRP material parameters, the load-bearing capacity after FRP reinforcement was predicted. The algorithm was optimized by Sobol sequence, nonlinearization strategy, multi-point Lévy flight strategy and Cauchy random mutation optimization method to establish a reinforcement effect prediction model.
It enables rapid and accurate evaluation of FRP reinforcement effects, avoids structural damage, provides theoretical reference for reinforcement schemes, and improves prediction accuracy and efficiency.
Smart Images
Figure CN117634286B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of power transmission towers, and particularly relates to a power transmission tower angle steel bearing capacity prediction method based on an African vulture intelligent algorithm. BACKGROUND
[0002] With the continuous increase of power load, more and more power transmission towers are put into operation to meet the demand. However, as the operation time of the power transmission tower increases, the structure may have problems such as fatigue, aging and corrosion due to the effects of factors such as icing, wind load and slope deformation, thereby causing the structural strength to decrease and the bearing capacity to weaken, threatening the safety performance of the power transmission tower and bringing challenges to the reliable operation of the power transmission line.
[0003] In view of this situation, it has become an urgent task to reinforce such towers and predict the reinforcement effect. The traditional reinforcement methods are the component parallel method and the welding reinforcement method, which inevitably cause secondary damage to the angle steel during the punching and welding operations, and at the same time, the two methods require a large amount of demolition and modification of the original structure, thereby increasing the construction difficulty and bringing uncertainty to the construction safety.
[0004] However, the use of lightweight high-strength FRP materials for non-destructive reinforcement of angle steel can ensure that the angle steel of the tower obtains the best reinforcement effect while avoiding secondary damage as much as possible. However, because the structure of FRP materials is complex and is easily affected by factors such as the slenderness ratio of the angle steel, the FRP laying direction, the number of layers, and the bonding length, there is no relevant standard for calculating the bearing capacity of the angle steel reinforced by FRP materials at home and abroad, and it is impossible to predict the reinforcement effect of the angle steel before reinforcement, so it is necessary to accurately predict the bearing capacity of the angle steel after FRP reinforcement, which provides an important theoretical reference for engineering implementation.
[0005] Based on this, a power transmission tower angle steel bearing capacity prediction method based on an African vulture intelligent algorithm is researched and developed. SUMMARY
[0006] The application provides a power transmission tower angle steel bearing capacity prediction method based on an African vulture intelligent algorithm, which optimizes the BP neural network based on the improved African vulture optimization algorithm, predicts the bearing capacity of the power transmission tower angle steel reinforced by FRP materials, and comprehensively considers the slenderness ratio of the angle steel, the FRP material parameters and the number of FRP layers, so as to calculate the bearing capacity of the angle steel after FRP reinforcement, conveniently and effectively evaluate the reinforcement effect, provide a reference for the reinforcement of the power transmission tower project, and effectively ensure the structural safety of the tower.
[0007] A power transmission tower angle steel bearing capacity prediction method based on an African vulture intelligent algorithm, comprising the following steps,
[0008] S001, obtain the power data set of the transmission tower angle steel after reinforcement;
[0009] S002, data preprocessing of African vulture intelligent algorithm
[0010] Vulture population initialization, the initial connection weight and threshold value in BP neural network are used as the initial spatial position of African vulture optimization algorithm, the fitness is obtained, the vultures with the best and the second best fitness values are determined as the first and second groups, and one vulture is selected from each group to lead the population to search for food, and the formula of the selection strategy is as follows:
[0011]
[0012] Among them, R BestVulture1 , R BestVulture2 represent the best and the second best vultures, θ1, θ2∈(0, 1) are self-defined parameters, when θ is infinitely close to 0, it means that the population diversity increases, R best is the best vulture;
[0013] In data preprocessing, the initial connection weight and threshold value in BP neural network are used as the initial spatial position of African vulture optimization algorithm, and the mean square error MSE is used as the fitness function, and its mathematical expression is as follows:
[0014]
[0015] Among them, n is the sample number of the data set, y i is the actual value of the sample, represents the training value of the sample.
[0016] S003, four methods of Sobol sequence, nonlinear strategy, multi-point Levy flight strategy and Cauchy random mutation optimization are used to optimize the original African vulture algorithm, and the improved African vulture optimization algorithm is obtained;
[0017] S004, through steps S002 and S003, update the position of each vulture, in the updating process of African vulture population, calculate the fitness value of each vulture and the optimal position corresponding to the current, when the African vulture optimization algorithm runs to the maximum iteration number or the preset precision, stop updating, output the minimum fitness value of the global vulture and the optimal position corresponding to it, and assign the minimum fitness value and the optimal position of the corresponding vulture to the weight and threshold value in BP neural network;
[0018] S005, the optimal weight value, threshold value and initial weight value of the BP neural network are obtained by iterating the African vulture optimization algorithm obtained in step S003, the optimal value is selected, the power tower angle steel reinforcement bearing capacity data set in step S001 is combined with the African vulture optimization algorithm, and a trained improved African vulture algorithm optimized BP neural network power tower angle steel reinforcement bearing capacity prediction model is obtained;
[0019] S006, input the parameter value of the power tower to be predicted into the power tower angle steel reinforcement bearing capacity prediction model, and predict the reinforcement effect.
[0020] Optionally, in step S001, the method for obtaining the power tower angle steel reinforcement bearing capacity data set is to perform a power tower angle steel reinforcement numerical test, change the angle steel slenderness ratio, FRP cloth bonding thickness, FRP cloth bonding length and laying direction, obtain the load-displacement curve of the reinforced angle steel, and the maximum load value under each reinforcement scheme in the reinforcement numerical test is taken as the bearing capacity extreme value of the reinforced angle steel. The data obtained by normalizing the numerical test results is used as the training sample.
[0021] Optionally, in S003, Sobol sequence is introduced to optimize the initial population of African vulture algorithm, and the mathematical expression is:
[0022] x i =x min +Y n ·(x max -x min ) (3)
[0023] Where Y n is a random number between 0 and 1, x i is the i-th Sobol sequence number, X max , X min are the optimal solution value range.
[0024] Optionally, in S003, a nonlinear strategy is introduced to optimize the original African vulture algorithm, which is to transform the original expression index. The data expression of the updated F is:
[0025]
[0026]
[0027] Where F' is the improved starvation index, r1 is a random value in (0, 1), T i_iter is the current iteration number, T max_iter is the total number of iterations, and r zA random number between -2 and 2, h is an adjustment parameter, randomly selected between -2 and 2, tau is an adjustment parameter for jumping out of local optimal solution of the algorithm, gamma is a user-defined parameter.
[0028] Optionally, the original African vulture algorithm is optimized by using a multi-point Levy flight strategy method, and the specific operation is as follows: a parameter P1 is set to represent the food searching strategy selected by the vulture, and the mathematical expression is:
[0029]
[0030] D(i) = |ξR best -X(i)| (7)
[0031] Wherein, X(i+1) is the position of the vulture in the next iteration, ξ is a random distribution number in [0, 2]; r2 is a random value in (0, 1), P1, randp1 are random values in (0, 1), lb, ub represent the upper and lower limits of the space, r3 is a coefficient for increasing randomness, D(i) is the distance from the current iteration to the i-th vulture randomly shaken off by the leader vulture.
[0032] Optionally, if the vulture hunger index |F|<1 is in the zone, the vulture individual has no ability to search for hunting, and then automatically enters the development stage in the current position. The development stage is divided into two sub-stages, namely the development early stage and the development late stage, and the two sub-stages are distinguished by the hunger index |F| = 0.5.
[0033] When the hunger index |F| is in the zone [0.5, 1), the development early stage is entered, and the vulture has more physical strength in this stage, which can be mainly divided into food guarding and alerting stage and circling hunting stage:
[0034] Food guarding and alerting stage: the vulture that is eating does not want to share food, and the hungry vulture will gather to attack and seize food. The position mathematical expression of this behavior is:
[0035] X(i+1) = D(i) x (F + r4) - (R best -X(i))P2 >= randp2 (8)
[0036] Wherein, X(i+1) is the position of the vulture in the next iteration, R best Indicates the best vulture.
[0037] Circling hunting stage: when the vulture searches for a stationary organism, it will circle hunting above it. If the organism is identified as a corpse, the vulture will stop flying and land to eat. The position update expression of this stage is:
[0038] X(i+1) = R best- (M1+M2)P2< randp2 (9)
[0039]
[0040]
[0041] wherein randp2 represents a strategy decision parameter, taking a value within (0, 1); r5, r6 are random parameters in the hunting stage, taking a random value within (0, 1); M1, M2 represent the positions of the vultures in the circling hunting.
[0042] Optionally, when the hunger index |F| ∈ (0, 0.5), the post-development stage is entered, and the very hungry state of the vultures in this stage will lead to the gathering of weak vultures to food and the initiation of an attack to seize the food. This behavior will also lead to the premature convergence and local optimal solution of the African vulture algorithm. The Cauchy random mutation optimization and levy flight strategy are introduced to optimize the easy gathering disadvantage in the post-development stage, and the updated mathematical expression is:
[0043]
[0044] X(i+1) = (M1'+M2') / 2 P3≥randp3 (13)
[0045]
[0046]
[0047] X(i+1) = Cauchy(0, 1) × R best -|(R best -X(i))|×F×Levy(d)P3< randp3 (16)
[0048] wherein P3, randp3 are strategy decision parameters, taking a value within (0, 1); Levy(d) is the Levy flight strategy; X 11 , X 22 are the positions of the best and second-best vultures; M1', M2' represent the updated expression of the positions of the best and second-best vultures; Cauchy(x) represents the Cauchy random mutation function; a is a random variable, taking a value within (0, 1).
[0049] Optionally, in S005, the method for obtaining the trained improved African vulture algorithm optimized BP neural network power transmission tower angle steel reinforcement post-bearing capacity prediction model is,
[0050] S0051, the initial weight and threshold value of the randomly generated BP neural network are taken as the initial space position of the African vulture optimization algorithm, and the fitness value is calculated, the optimal initial weight and threshold value are obtained by iterative updating of the African vulture optimization algorithm, and the optimal fitness value is obtained, so as to establish the neural network model of the African vulture optimization algorithm;
[0051] S0052, the initial weight and threshold value of the neural network model of the African vulture optimization algorithm in S0051 are trained by the power tower angle steel reinforcement after load capacity data set until the training error reaches the maximum iteration number T' max_iter , the optimal neural network connection weight and threshold value are obtained, and the training sample in the power tower angle steel reinforcement after load capacity data set and the test sample in the test set are imported into the trained African vulture optimization algorithm neural network model to predict the load capacity of the power tower angle steel after reinforcement.
[0052] Optionally, in S0051, after the neural network model of the African vulture optimization algorithm is established, the transfer function between the input layer and the hidden layer of the neural network is set as tansig, and the logsig between the hidden layer and the output layer is taken as the training function, and the initial weight and threshold value obtained in the adjustment calculation process are adjusted.
[0053] Optionally, in S006, the parameter values of the power tower to be predicted for reinforcement effect include the slenderness ratio of the angle steel to be reinforced, the FRP cloth bonding thickness, the bonding length, and the laying direction, and the parameter values of the power tower to be predicted for reinforcement effect are taken as the input parameters of the African vulture optimization algorithm neural network model, so that the prediction value of the reinforced angle steel load capacity can be obtained.
[0054] Compared with the prior art, the beneficial effects of the present application are:
[0055] 1) Compared with the component parallel reinforcement method commonly used for power tower angle steel, the FRP material used for reinforcing the angle steel can well avoid a large amount of demolition and modification of the original structure during the reinforcement process, can provide reinforcement effect while not significantly increasing the structure dead weight and damaging the original structure, and compared with the existing method, the problem of being unable to predict the technical effect of the angle steel on the load capacity prediction, four factors of the slenderness ratio of the angle steel, the FRP bonding length, the laying direction, and the bonding thickness are taken as the index for predicting the load capacity of the power tower angle steel after FRP reinforcement, a power tower angle steel reinforcement after load capacity prediction model is established, compared with manual evaluation, the time is greatly reduced, the reinforcement effect of the angle steel can be quickly and efficiently predicted, and the reinforcement scheme designer can better judge the reinforcement effect.
[0056] 2) The technical solution improves the original African vulture optimization algorithm by introducing Sobol sequence, nonlinear strategy, multi-point Levy flight strategy and Cauchy random variation optimization, improves the ability of AVOA algorithm to jump out of local optimal value, and improves the running speed and prediction accuracy of BP neural network after optimizing the weight threshold value of improved IAVOA algorithm, avoiding the algorithm result falling into local optimal value.
[0057] 3) The technical solution establishes an African vulture optimization algorithm neural network model to predict the reinforcement effect of the angle steel of the power transmission tower, which can help the personnel designing the reinforcement scheme to better judge the reinforcement effect, and provides guidance for the selection and optimization of the power transmission tower reinforcement scheme. BRIEF DESCRIPTION OF DRAWINGS
[0058] Figure 1 The method flowchart described in the present application. DETAILED DESCRIPTION
[0059] In order to make the purpose, technical scheme and advantages of the present application clearer and more understandable, the present application will be further described in detail below, and the illustrative embodiments of the present application and their descriptions are only used to explain the present application, and not as a limitation of the present application.
[0060] Example 1:
[0061] As shown in the figure, a power transmission tower angle steel bearing capacity prediction method based on African vulture intelligent algorithm, comprising the following steps, Figure 1
[0062] S001, obtaining the bearing capacity data set of the reinforced power transmission tower angle steel;
[0063] In step S001, the method for obtaining the bearing capacity data set of the reinforced power transmission tower angle steel is to perform a numerical test of the reinforcement of the power transmission tower angle steel, change the angle steel slenderness ratio, FRP cloth adhesion thickness, FRP cloth adhesion length and laying direction, obtain the load-displacement curve of the reinforced angle steel, and the maximum load value under each reinforcement scheme in the reinforcement numerical test is taken as the bearing capacity extreme value of the reinforced angle steel. The data after normalization processing of the numerical test results is taken as the training sample.
[0064] S002, data preprocessing of African vulture intelligent algorithm
[0065] Vulture population initialization, the initial connection weight and threshold value in the BP neural network are taken as the initial spatial position of the African vulture optimization algorithm, the fitness is obtained, the vultures with the best and second best fitness values are determined as the first and second groups, and one vulture is selected from each group to drive the population to hunt for food, and the formula of the selection strategy is as follows:
[0066]
[0067] wherein R BestVulture1 , R BestVulture2 represent the best, the second best vulture respectively, θ1, θ2∈(0, 1) are self-defined parameters, when θ is infinitely close to 0, it represents that the population diversity increases, R best is the best vulture;
[0068] S003, four methods of Sobol sequence, nonlinear strategy, multi-point Levy flight strategy and Cauchy random mutation optimization are used to optimize the original African vulture algorithm to obtain the improved African vulture optimization algorithm;
[0069] In S003, Sobol sequence is introduced to optimize the initial population of African vulture algorithm, and the mathematical expression is:
[0070] x i = x min + Y n ·(x max -x min ) (3)
[0071] wherein Y n is a random number between 0 and 1, x i is the i-th Sobol sequence number, X max , X min are the optimal solution value range.
[0072] In the African vulture algorithm AVOA algorithm, the initial population diversity can effectively expand the search range of the algorithm, thereby improving the optimization precision and convergence speed of the algorithm. Data preprocessing, Sobol sequence is a low-discrepancy sequence with good benchmark measurement properties and statistical performance, so it can generate more uniform and more closely traverse the entire space. The introduction of Sobol sequence can further optimize and improve the original African vulture algorithm.
[0073] In S003, the nonlinear strategy is introduced to optimize the original African vulture algorithm, which is to transform the original expression index, and the data expression of the updated F is:
[0074]
[0075]
[0076] wherein F' is the improved starvation index, r1 is a random value in (0, 1), T i_iter is the current iteration number, T max_iter is the total number of iterations, r za random number between -2 and 2, h is an adjustment parameter, randomly selected between -2 and 2, τ is an adjustment parameter for jumping out of local optimal solution of the algorithm, and γ is a user-defined parameter.
[0077] The mathematical expression of the starvation index in the original African vulture algorithm is a linear algorithm, and such linear algorithms usually require more iterations to reach the optimal solution, resulting in slower convergence speed of the algorithm and increased complexity of the algorithm. Therefore, an exponential transformation is introduced to the original expression. After introducing the exponential transformation, the algorithm can quickly explore the solution space in the initial stage and gradually reduce the search space, so as to find the optimal solution faster. In addition, the exponential transformation algorithm can also avoid the original African vulture algorithm falling into a local optimal solution, thereby improving the global search ability of the African vulture optimization algorithm.
[0078] High-altitude optimization stage of vultures. In the natural environment, the visual ability of vultures is very developed, and they have strong search ability. However, it is very difficult for vultures to find food, which is manifested in that vultures either travel long distances to find food or search for food in random areas. The original African vulture algorithm is optimized by using the multi-point Levy flight strategy method. The specific operation is as follows: a parameter P1 is set to represent the strategy selected by the vulture to find food, and the mathematical expression is:
[0079]
[0080] D(i) = |ξR best - X(i) | (7)
[0081] where X(i+1) is the position of the vulture in the next iteration, ξ is a random number distributed in [0, 2]; r2 is a random value in (0, 1), P1, randp1 are random values in (0, 1), lb, ub represent the upper and lower limits of the space, r3 is a coefficient for increasing randomness, and D(i) is the distance from the i-th vulture in the current iteration to be randomly shaken off by the leader vulture.
[0082] The development stage of the African vulture optimization algorithm is improved. If the vulture starvation index |F| < 1 is in the zone, the vulture individual has no ability to hunt far away, and then automatically enters the development stage at the current position. This stage is divided into two sub-stages, namely the development early stage and the development late stage, which are distinguished by the starvation index |F| = 0.5.
[0083] When the starvation index |F| ∈ [0.5, 1) enters the development early stage, the vulture has more physical strength at this stage, which can be mainly divided into the food protection alert stage and the circling hunting stage.
[0084] Food protection alert stage: the vulture that is eating does not want to share food, and the hungry vulture will gather to attack and seize food. The mathematical expression of the position of this behavior is:
[0085] X(i+1) = D(i) x (F + r4) - (R best - X(i)) P2≥ randp2 (8)
[0086] where X(i+1) is the position of vulture in the next iteration, R best represents the best vulture.
[0087] Circling hunting stage: when the vulture searches for a stationary organism, it will circle over it for hunting. If the organism is identified as a corpse, the vulture will stop flying and land to eat. The position update expression in this stage is:
[0088] X(i+1) = R best - (M1+M2) P2< randp2 (9)
[0089]
[0090]
[0091] where randp2 represents the strategy decision parameter, taking value in (0, 1); r5, r6 are random values in (0, 1).
[0092] When the hunger index |F| ∈ (0, 0.5), the vulture enters the late development stage. In this stage, the vulture is in a very hungry state, which will lead to the gathering of weak vultures to food and the initiation of attack to seize food. This behavior will also lead to the premature convergence and local optimal solution of the African vulture algorithm. The Cauchy random mutation optimization and levy flight strategy are introduced to optimize the gathering disadvantage in the late development stage. The updated mathematical expression is:
[0093]
[0094] X(i+1) = (M1'+M2') / 2 P3≥ randp3 (13)
[0095]
[0096]
[0097] X(i+1) = Cauchy(0, 1) x R best - | (R best - X(i)) | x F x Levy(d) P3< randp3 (16)
[0098] where P3, randp3 are strategy decision parameters, taking value in (0, 1); Levy(d) is the Levy flight strategy; X 11 , X22 M1, M2 represent the optimal and sub-optimal vulture position update expression; Cauchy(x) represents the Cauchy random variation function; a is a random variable, which is in (0, 1).
[0099] S004, by step S002 and step S003, update each vulture position, calculate the fitness value of each vulture and the optimal position corresponding to the current during the updating process of African vulture population, stop updating when the African vulture optimization algorithm runs to the maximum iteration number or the preset precision, output the minimum fitness value of the global vulture and the optimal position corresponding to it, and assign the minimum fitness value and the optimal position of the vulture to the weights and thresholds in the BP neural network;
[0100] S005, select the optimal value from the optimal weights and thresholds obtained by the iteration of the African vulture optimization algorithm in step S003 and the initial weights of the BP neural network, combine the power tower angle steel reinforcement bearing capacity data set in step S001 with the African vulture optimization algorithm to obtain a trained improved African vulture algorithm optimized BP neural network power tower angle steel reinforcement bearing capacity prediction model;
[0101] In S005, the method for obtaining the trained improved African vulture algorithm optimized BP neural network power tower angle steel reinforcement bearing capacity prediction model comprises the following steps:
[0102] S0051, randomly generated BP neural network initial weights and thresholds are used as the initial space position of the African vulture optimization algorithm, and the fitness value is calculated and obtained, the optimal initial weights and thresholds are obtained by iteration update of the African vulture optimization algorithm, and the optimal fitness value is obtained, so as to establish an African vulture optimization algorithm neural network model;
[0103] In the above S0051, after establishing the African vulture optimization algorithm neural network model, the transfer function between the input layer and the hidden layer of the neural network is set as tansig, and the logsig between the hidden layer and the output layer is used as the training function, and the initial weights and thresholds obtained in the adjustment calculation process are adjusted.
[0104] In the data preprocessing, the initial connection weights and thresholds in the BP neural network are used as the initial space position of the African vulture optimization algorithm, and the mean square error MSE is used as the fitness function, and its mathematical expression is as follows:
[0105]
[0106] Wherein, n is the number of data set samples, y i is the actual value of the sample, represents the training value of the sample.
[0107] S0052, the initial weight and threshold of the African vulture optimization algorithm neural network model in S0051 are trained by the power tower angle steel reinforcement bearing capacity data set in S001 until the training error reaches the maximum iteration number T' max_iter , the optimal neural network connection weight and threshold are obtained, and the training samples in the power tower angle steel reinforcement bearing capacity data set and the test samples in the test set are imported into the trained African vulture optimization algorithm neural network model to predict the bearing capacity of the power tower angle steel after reinforcement.
[0108] S006, the parameter values of the power tower to be predicted for reinforcement effect are input into the power tower angle steel reinforcement bearing capacity prediction model to predict the reinforcement effect. In S006, the parameter values of the power tower to be predicted for reinforcement effect include the slenderness ratio of the angle steel to be reinforced, the FRP cloth bonding thickness, the bonding length, and the laying direction. The parameter values of the power tower to be predicted for reinforcement effect are used as the input parameters of the African vulture optimization algorithm neural network model, and the predicted value of the bearing capacity of the reinforced angle steel can be obtained.
[0109] According to the method described in the embodiment, the bearing capacity of the reinforced power tower angle steel is predicted, and the expected prediction effect is compared with the effect verified by numerical test, and the effects are consistent.
[0110] The scheme described in the embodiment improves the original African vulture optimization algorithm by introducing Sobol sequence, nonlinear strategy, multi-point Levy flight strategy and Cauchy random variation optimization, improves the ability of AVOA algorithm to jump out of local optimal value, improves the running speed and prediction accuracy of BP neural network after optimizing the weight threshold of improved IAVOA algorithm BP neural network, and improves the accuracy of BP neural network model in predicting the bearing capacity of the power tower angle steel after reinforcement. It can help the reinforcement scheme personnel to better judge the reinforcement effect and provide guidance for the selection and optimization of the power tower reinforcement scheme.
[0111] The above specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application should be included in the protection scope of the present application.
Claims
1. A power transmission tower angle steel bearing capacity prediction method based on an African vulture intelligent algorithm, characterized by: It comprises the following steps: S001, obtaining the bearing capacity data set of the transmission tower angle steel after reinforcement; S002, data preprocessing of the African vulture intelligent algorithm, vulture population initialization, the initial connection weight and threshold value in the BP neural network are used as the initial spatial position of the African vulture optimization algorithm, the fitness is obtained, the vultures with the best and second best fitness values are determined as the first and second groups, and one vulture from each group is selected to drive the population to hunt for food, and the formula of the selection strategy is as follows: (2) where R BestVulture1 , R BestVulture2 represent the best, second best vulture, is a self-defined parameter, when approaches 0, it means the population diversity increases, is the best vulture; In the data preprocessing, the vulture population initialization, the initial connection weight and threshold value in the BP neural network are used as the initial spatial position of the African vulture optimization algorithm, and the mean square error MSE is used as the fitness function, and the formula is: （1） Wherein, n is the number of data set samples, yi is the actual value of the sample, S003, using Sobol sequence, nonlinear strategy, multi-point Levy flight strategy and Cauchy random variation optimization four methods to optimize the original African vulture algorithm, and obtain the improved African vulture optimization algorithm; S004, through step S002 and step S003, update the position of each vulture, calculate the fitness value of each vulture and the optimal position corresponding to the current in the process of African vulture population update, the African vulture optimization algorithm stops updating when the maximum iteration number or the preset precision is reached, the minimum fitness value of the global vulture and the optimal position corresponding to it are output, and the minimum fitness value and the optimal position of the vulture are assigned to the weights and thresholds in the BP neural network. S005, obtaining the optimal weight and threshold value of the African vulture optimization algorithm obtained in step S003, selecting the optimal value, combining the bearing capacity data set of the transmission tower angle steel after reinforcement in step S001 with the African vulture optimization algorithm, and obtaining the trained improved African vulture algorithm optimized BP neural network transmission tower angle steel bearing capacity prediction model; S006, inputting the parameter value of the transmission tower to be predicted into the transmission tower angle steel bearing capacity prediction model after reinforcement to predict the reinforcement effect; In S003, the Sobol sequence is introduced to optimize the initial population of the African vulture algorithm, and the mathematical expression is: (3) where, is a random number between (0, 1), is the first Sobol sequence number, X max , X min is the optimal solution value range; in S003, the original African vulture algorithm is optimized by introducing a nonlinear strategy, which is to transform the original expression index. The data expression of the updated F is: （4） （5） wherein F' is the improved hunger index, r1 is a random value in (0, 1), T i_iter is the current iteration number, T max_iter is the total number of iterations, r z is a random number between (-1, 1), and h is an adjustment parameter randomly selected between (-2, 2), is an adjustment parameter used to jump out of the local optimal solution of the algorithm, is a user-defined parameter; the original African vulture algorithm is optimized by using the multi-point Levy flight strategy method, and the specific operation is as follows: a parameter P1 is set to represent the food searching strategy selected by the vulture, and the mathematical expression is as follows: （6） (7) where, is the position of the vulture at the next iteration, a random distribution number; a random value in, a random value in, lb, ub represent the lower and upper bounds of the space, is a coefficient used to increase the random nature, is the distance from the i-th vulture in the current iteration to be randomly dropped by the leader vulture.
2. The method according to claim 1, wherein the method is based on an intelligent algorithm of Gyps africanus for predicting the bearing capacity of angle steel of a power transmission tower. In step S001, the method for obtaining the bearing capacity data set of the transmission tower angle steel after reinforcement is to perform a numerical test on the reinforcement of the transmission tower angle steel, change the length-to-thickness ratio of the angle steel, the thickness of the FRP cloth adhesion, the length of the FRP cloth adhesion, and the laying direction, obtain the load-displacement curve of the reinforced angle steel, and the maximum load value under each reinforcement scheme in the numerical test is used as the bearing capacity extreme value of the reinforced angle steel. The normalized data of the numerical test results are used as the training samples.
3. The African vulture intelligent algorithm-based transmission tower angle steel bearing capacity prediction method according to claim 2, characterized in that: If the vulture hunger index The vulture individual has no ability to go far to hunt, and automatically enters the development stage at the current location. This stage is divided into two sub-stages, namely the early development stage and the late development stage. The two sub-stages are distinguished by the hunger index The vulture hunger index When the hunger index The pre-development stage is entered when the hunger index The pre-development stage is entered when the hunger index Feeding alert stage: The vultures that are eating do not want to share food, and the hungry vultures will gather to attack and rob food. The mathematical expression of the position of this behavior is: (8) wherein, is the position of the vulture at the next iteration, denotes the best vulture; The circling hunting stage: when the vulture searches for a stationary organism, it will circle above it for hunting, and if it identifies the organism as a corpse, the vulture will stop flying and land to eat. The position update expression in this stage is: （9） （10） （11） wherein, represents a policy decision parameter, taking a value within (0, 1); is a random parameter in the hunting phase, taking a random value within (0, 1); M1, M2 represent the positions of the vultures when circling and hunting.
4. The African vulture intelligent algorithm-based transmission tower angle steel bearing capacity prediction method according to claim 3, characterized in that: When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index When the hunger index < (12) （13） （14） （15） （16） wherein, is a strategy decision parameter, taking values in (0, 1); is the Levy flight strategy; is the position of the best, second best vulture; M1 ’ , M2 ’ represents the expression after updating the position of the best, second best vulture; Cauchy(x) represents the Cauchy random variation function; a is a random variable, taking values in (0, 1).
5. The method according to claim 4, wherein the method is based on the intelligent algorithm of Gyps africanus for predicting the bearing capacity of the angle steel of the power transmission tower. In S005, the method for obtaining the trained improved African vulture algorithm optimized BP neural network transmission tower angle steel bearing capacity prediction model is as follows: S0051, the randomly generated BP neural network initial weight and threshold value are used as the initial spatial position of the African vulture optimization algorithm, and the fitness value is calculated and obtained, the optimal initial weight and threshold value are obtained through iterative updating of the African vulture optimization algorithm, and the optimal fitness value is obtained, so as to establish the African vulture optimization algorithm neural network model; S0052, the power transmission tower angle steel reinforcement after the bearing capacity data set in step S001 is trained to the initial weight and threshold of the African vulture optimization algorithm neural network model in S0051, until the training error reaches the maximum iteration number T' max_iter , the optimal neural network connection weight and threshold are obtained, and the training samples in the power transmission tower angle steel reinforcement bearing capacity data set and the test samples in the test set are imported into the trained African vulture optimization algorithm neural network model to predict the bearing capacity of the power transmission tower angle steel reinforcement.
6. The method according to claim 5, wherein the method is based on an intelligent algorithm of Gyps africanus for predicting the bearing capacity of angle steel of a power transmission tower. In S0051, after establishing the African vulture optimization algorithm neural network model, the transfer function between the input layer and the hidden layer of the neural network is set as tansig, and the logsig between the hidden layer and the output layer is used as the training function, and the initial weight and threshold value obtained in the adjustment calculation process are adjusted.
7. The method according to claim 6, wherein the method is based on an intelligent algorithm of Gyps africanus for predicting the bearing capacity of angle steel of a power transmission tower. In S006, the parameter values of the power transmission tower to be predicted for reinforcement effect include the slenderness ratio of the angle steel to be reinforced, the FRP cloth bonding thickness, the bonding length, and the laying direction, and the parameter values of the power transmission tower to be predicted for reinforcement effect are taken as the input parameters of the African vulture optimization algorithm neural network model, so that the predicted value of the bearing capacity of the reinforced angle steel can be obtained.