Power transmission line short-term icing prediction method and system, storage medium and electronic equipment

Abstract

The invention relates to the technical field of icing prediction of power transmission lines, and discloses a short-term icing prediction method and system for a power transmission line, a storage medium and electronic equipment. The method comprises the following steps: acquiring micrometeorological data and equivalent icing thickness data of a power transmission line, and constructing an input and output sample set; performing feature vector screening on the sample set to obtain a combined sample set; initializing a whale optimization algorithm population, and obtaining an improved whale algorithm based on an adaptive weight factor and Tent chaotic mapping; optimizing through an improved whale algorithm to obtain an optimal solution of a penalty factor and a kernel function parameter, and taking the optimal solution as a model parameter of a support vector regression model; and training and testing the support vector regression model by using the combined sample set to obtain an icing prediction model, and carrying out short-term icing pre-treatment on the power transmission line based on the icing prediction model. According to the invention, the model prediction efficiency and generalization ability are improved; the problem that an original algorithm is prone to falling into a local optimal solution is solved, and the application range of the model is widened.

Description

Technical Field

[0001] The present invention relates to the technical field of icing prediction of power transmission lines, and in particular to a method, system, storage medium and electronic equipment for predicting short-term icing of power transmission lines. Background Art

[0002] In recent years, with the frequent occurrence of severe weather such as freezing disasters due to climate warming and other reasons, the power grid infrastructure has been damaged, and the resistance of transmission lines to ice and snow disasters has become one of the research hotspots in the electrical field. Severe icing of transmission lines can cause transmission line overload, conductor dancing, tower collapse, and insulator string flashover, which seriously endangers people's normal life and the safe operation of the power system. Therefore, the research on transmission line icing prediction has important engineering practical value.

[0003] At present, the existing transmission line icing prediction schemes generally include icing prediction based on physical models and icing prediction based on neural network models. The icing prediction method based on physical models is generally complex in calculation, difficult, and has high construction costs, and is difficult to apply to the actual icing thickness prediction. The icing prediction based on the BP neural network model requires a large amount of historical icing data for training. Before this, the transmission line icing detector has not been fully perfected, and the icing time in most areas is very short. The micro-meteorological data accumulated over the years and its equivalent icing thickness are not enough to support the completion of the entire training process, resulting in low model prediction accuracy. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method, system, storage medium and electronic device for predicting short-term icing of a transmission line in view of the problems existing in the prior art.

[0005] In a first aspect, the present application provides a method for predicting short-term icing of a transmission line, comprising the following steps:

[0006] Obtain micro-meteorological data and equivalent ice thickness data of local transmission lines and construct input and output sample sets;

[0007] The micro-meteorological data in the input and output sample sets are screened for feature vectors using the random forest algorithm, and the obtained feature vectors are combined with the equivalent ice thickness data to obtain a combined sample set.

[0008] Initialize the whale optimization algorithm population, and obtain the improved whale algorithm based on the adaptive weight factor and Tent chaotic mapping;

[0009] By improving the whale algorithm to find the optimal solution of the penalty factor and kernel function parameters, the optimal solution is used as the model parameters of the support vector regression model;

[0010] The support vector regression model is trained and tested using the combined sample set to obtain an icing prediction model. Short-term icing prediction of transmission lines is performed based on the icing prediction model.

[0011] In a second aspect, the present application provides a short-term icing prediction system for a power transmission line, comprising:

[0012] The sample set construction module is used to obtain the micro-meteorological data and equivalent ice thickness data of the local transmission lines and construct the input and output sample sets;

[0013] The sample set screening module is used to screen the feature vectors of the micrometeorological data in the input and output sample sets through the random forest algorithm, and combine the obtained feature vectors with the equivalent ice thickness data to obtain a combined sample set;

[0014] Whale algorithm improvement module, used to initialize the whale optimization algorithm population, and obtain the improved whale algorithm based on adaptive weight factor and Tent chaotic mapping;

[0015] The optimal parameter solution module is used to find the optimal solution of the penalty factor and kernel function parameters by improving the whale algorithm, and use the optimal solution as the model parameter of the support vector regression model;

[0016] The model training and application module is used to train and test the support vector regression model using the combined sample set to obtain an icing prediction model, and perform short-term icing prediction on the transmission line based on the icing prediction model.

[0017] In a third aspect, the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a computer, enable the computer to execute the method for predicting short-term icing of transmission lines as provided in the above technical solution.

[0018] In a fourth aspect, the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for predicting short-term icing of transmission lines as provided in the above technical solution is implemented.

[0019] In a fifth aspect, the present application further provides a computer program product or a computer program, the computer program product or the computer program including computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the short-term icing prediction method for a transmission line provided by the above solution.

[0020] The beneficial effects of the present invention are as follows: the weights of various meteorological factors are analyzed by the random forest algorithm, and the cross-overlap between micro-meteorological information can be reduced through feature selection, the sample space dimension can be reduced, and the model prediction efficiency and generalization ability can be improved; by introducing the Tent mapping, the whale group is evenly distributed, and the chaos has randomness, ergodicity and initial value sensitivity, which can make the algorithm converge faster, and to a certain extent solve the problem that the original algorithm is prone to fall into the local optimal solution; by training the support vector regression model SVR, the regression problem is transformed into the problem of solving a set of support vectors, which can handle nonlinear regression problems, and by selecting an appropriate kernel function to map data to a high-dimensional space, the adaptability of the model is improved.

[0021] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description or learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of a method for predicting short-term icing of a transmission line according to an exemplary embodiment of the present application;

[0023] Figure 2 for Figure 1 A flowchart of step S3 in the illustrated embodiment in an exemplary embodiment;

[0024] Figure 3 for Figure 1 A flowchart of step S4 in the illustrated embodiment in an exemplary embodiment;

[0025] Figure 4 A schematic diagram of a random forest structure shown as an exemplary embodiment of the present application;

[0026] Figure 5 This is a flowchart of an improved whale algorithm optimization process shown in an exemplary embodiment of the present application;

[0027] Figure 6 This is a schematic diagram of the screening results of micro-meteorological factors using a random forest algorithm, shown as an exemplary embodiment of the present application;

[0028] Figure 7 This is a schematic diagram of a test set prediction value change curve for an exemplary embodiment of the present application;

[0029] Figure 8 A prediction error comparison diagram shown in an exemplary embodiment of the present application;

[0030] Fig. 9 A block diagram of a short-term icing prediction system for power transmission lines shown as an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0031] The following describes the embodiments of the present disclosure through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0032] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0033] In recent years, with the frequent occurrence of severe weather such as freezing disasters due to climate warming and other reasons, the power grid infrastructure has been damaged, and the resistance of transmission lines to ice and snow disasters has become one of the research hotspots in the electrical field. Severe icing of transmission lines can cause accidents such as overload of transmission lines, dancing of conductors, tower collapse and tower breakage, and flashover of insulator strings, which seriously endanger people's normal life and the safe operation of the power system. This paper summarizes and analyzes icing accidents of transmission lines at home and abroad, the types and properties of icing, and the factors affecting icing. Theoretical calculations are combined with condensed liquid and environmental factors during the icing process. Prediction modeling is carried out based on factors such as ice thickness and conductor angle. Finally, experiments are conducted to verify the accuracy and effectiveness of the prediction, which will help to further improve the transmission line icing prediction system. Therefore, the research on icing prediction of transmission lines has important engineering practical value.

[0034] At present, domestic and foreign scholars have conducted many studies on the prediction of equivalent ice thickness on transmission lines, mainly including physical models and statistical models. In terms of physical models, starting from the mechanical perspective, a physical model is proposed to predict the equivalent ice thickness on the line by measuring the inclination angle and tension sensor of the transmission line from the torque aspect; an ice parameter prediction model based on the improved differential evolution algorithm is established through the conductor shape. However, the common point of the above literature is that they are all based on physical structures. The ice prediction methods based on physical models are generally complex in calculation, difficult, and have high construction costs, and are difficult to apply to actual ice thickness prediction. Therefore, starting from the statistical regression model, it was proposed to use a multivariate grey prediction model to screen multiple input feature variables and build a model for small data samples to obtain the equivalent ice thickness; the typical ice shapes that will appear on the lines under different icing conditions were summarized, and the icing process was divided into light icing and heavy icing, and icing prediction was performed based on the ice shapes. This method has low accuracy, slow iteration rate, and takes too long, and requires a lot of initial accumulation time, and the initial investment is too time-consuming; an improved BP neural network is used to construct an icing model prediction, but the construction based on the neural network requires a large amount of historical icing data for training. Before this, the transmission line icing detector has not been fully perfected, and the icing time in most areas is very short. The micrometeorological data accumulated over the years and its equivalent ice thickness are not enough to support the completion of the entire training process.

[0035] The ice growth process itself is non-static and non-sequential. When using medium- and long-term data modeling and calculation, the prediction results will have large errors, reducing the reliability of the cable's anti-icing ability, and it is difficult to apply it to actual overhead transmission lines. Compared with medium- and long-term icing, short-term icing lacks a large amount of sample data, and the calculation results of its statistical analysis model have large errors compared to the actual values. At present, the main method for predicting short-term icing on transmission lines is through BP neural networks. Traditional BP neural networks require a large amount of historical icing data to predict transmission line icing, and the prediction accuracy is low.

[0036] Therefore, in order to ensure the safe operation of transmission lines in icing-prone areas, the solution to the line icing process is complex and requires a large number of samples. It is urgent to propose a new prediction method with high accuracy that can be used for short-term prediction of the thickness of ice covering on transmission lines.

[0037] To solve the above technical problems, the embodiments of the present invention provide a method, system, storage medium and electronic device for predicting short-term icing of power transmission lines. The solution provided by the embodiments of the present invention can better solve the prediction problems under conditions such as insufficient icing data samples and large random fluctuations, and can provide guidance and basis for icing prediction.

[0038] The short-term icing prediction method for transmission lines provided in the embodiment of the present invention can be executed by any electronic device, for example, a terminal device, or jointly executed by a terminal device and a server. The server can be an independent server, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms, etc., and is not limited here.

[0039] Figure 1 The flowchart of a method for predicting short-term icing of a transmission line is shown as an exemplary embodiment of the present application. Figure 1 As shown, in an exemplary embodiment, the short-term icing prediction method for a transmission line may include steps S1 to S5, which are described in detail as follows:

[0040] S1, obtain the micro-meteorological data and equivalent ice thickness data of the local transmission lines, and construct the input and output sample sets.

[0041] In an embodiment of the present invention, micro-meteorological data and equivalent ice thickness data of the local power transmission line are obtained, the data is pre-processed, the training set and the test set are divided, and the input and output sample sets are constructed. Specifically, micro-meteorological data refers to the meteorological data in a small place or a small area, such as various precise meteorological data of a tower or a hilltop, rather than general meteorological data. In an embodiment of the present invention, the micro-meteorological data may include temperature, humidity, wind speed, wind direction, atmospheric pressure, rainfall, etc., and the equivalent ice thickness refers to the ice thickness when the weather remains stable at a certain moment.

[0042] The calculation formula of equivalent ice thickness is as follows:

[0043]

[0044] Where b is the equivalent ice thickness, q ice is the ice weight, q ice =ω 1 -ω 2 ,ω 1 is the sum of the self-weight per unit length of the overhead line and the weight of ice, ω 2 is the self-weight per unit length of the overhead line, D is the conductor diameter, and ρ is the standard ice density designed for the line.

[0045] S2, the micrometeorological data in the input and output sample sets are screened for feature vectors using the random forest algorithm, and the obtained feature vectors are combined with the equivalent ice thickness data to obtain a combined sample set.

[0046] Figure 4FIG. 1 is a schematic diagram of a random forest structure shown in an exemplary embodiment of the present application. Figure 4 As shown in Figure 1, input: randomly select a part of samples from the original input and output training set to form a sub-sample set. This allows each decision tree to be trained on a different sample set, thereby increasing the diversity of the model.

[0047] Estimator: For each node of each decision tree, only a randomly selected portion of features is considered when selecting the best split feature. This prevents certain features from having too much influence on the entire model, thereby improving the robustness of the model.

[0048] Combiner: Use a decision tree algorithm (such as CART algorithm) to build a decision tree on each sub-sample set. In the process of growing a decision tree, the best partitioning features are usually selected recursively to divide the data set into subsets with the least impurity.

[0049] Output: For new input samples, the prediction results of multiple decision trees are averaged or weighted averaged to obtain the final regression result.

[0050] The embodiment of the present invention analyzes the weights of various meteorological factors through the random forest algorithm, and can reduce the cross-overlap between micro-meteorological information through feature selection, reduce the sample space dimension, and improve the model prediction efficiency and generalization ability.

[0051] S3, initialize the whale optimization algorithm population, and obtain the improved whale algorithm based on the adaptive weight factor and Tent chaotic mapping.

[0052] The embodiment of the present invention introduces Tent mapping to make the whale group evenly distributed. Chaos has randomness, ergodicity and initial value sensitivity, which can make the algorithm converge faster and solve the problem that the original algorithm is prone to fall into local optimal solutions to a certain extent.

[0053] S4, by optimizing the improved whale algorithm, the optimal solution of the penalty factor and the kernel function parameters is obtained, and the optimal solution is used as the model parameters of the support vector regression model.

[0054] S5, using the combined sample set to train and test the support vector regression model to obtain an icing prediction model, and based on the icing prediction model, short-term icing prediction of the transmission line is performed.

[0055] The embodiment of the present invention trains the support vector regression model SVR to transform the regression problem into the problem of solving a set of support vectors, can handle nonlinear regression problems, and maps data to a high-dimensional space by selecting an appropriate kernel function, thereby improving the adaptability of the model.

[0056] Optionally, Figure 1In an exemplary embodiment, step S2 in the illustrated embodiment further includes normalizing the data in the input and output sample sets before performing feature vector screening on the micrometeorological data in the input and output sample sets by using the random forest algorithm.

[0057] The normalization formula is as follows:

[0058]

[0059] In the formula, y is the normalized data; x is the original data, x min and x max They are the maximum and minimum values ​​in the original data, respectively, and the normalized range is [0,1].

[0060] Optionally, Figure 1 In step S3 of the illustrated embodiment, in an exemplary embodiment, a whale optimization algorithm population is initialized, and an improved whale algorithm is obtained based on an adaptive weight factor and a Tent chaotic map.

[0061] It should be noted that the method for running the whale optimization algorithm in the prior art is specifically as follows:

[0062] 1) Encirclement and predation: When a whale is in the best position X D When the other whales move towards this point, the distance formula (3) between any whale when it moves towards the whale with the best position during the hunting process is:

[0063]

[0064] Where t is the time when the whale group moves, is the coordinates of the best whale at time t; is the coordinate of the whale currently traveling at time t. The expression (4) is:

[0065]

[0066] Each time the whale swims to the optimal position, the position formula (5) after the change is:

[0067]

[0068] in, is the coordinate of the whale currently traveling at time t+1, is the best whale coordinate at time t, and is the intermediate calculation parameter, is a random number between [0,1], is the convergence factor, q is the current iteration number, and Q is the maximum iteration number.

[0069] 2) Spiral Bubble Predation: This process considers that when the whale is located below the prey, it will release a large amount of air from the lungs to form a rotating airflow, which in turn generates a vortex to attract smaller fish and shrimp. The mathematical model formula (6) of this behavior is as follows:

[0070]

[0071] In the formula, is the best whale coordinate at time t+1, is the best whale coordinate at time t, is the coordinate of the whale at time t, k is the whale's rotation constant, and l is a random value in the range of (-1,1).

[0072] 3) Global search and predation: Whales will constantly adjust their direction of travel as the surrounding environment changes during the predation process, which means that the whale's movement trajectory is not moving towards a single coordinate, but constantly changing its target point in the vast ocean. This algorithm needs to correct the convergence factor In order to simulate the movement of whales more accurately, it is also possible to obtain their optimal coordinates. When , it means that the whale's action status at this time is that it has found other targets while moving towards the original destination. Its movement range increases, the search space expands, and it can often find the best hunting object. When the corresponding optimization is achieved, it is the global optimal. And the prey position formula (7) is updated as follows:

[0073]

[0074] Where D rand and is the intermediate calculation parameter, is a random number between [0,1], X rand (t) is the position of a whale randomly selected from the group at time t, is the coordinate of the whale currently traveling at time t, is the coordinate of the whale currently traveling at time t+1, is the intermediate calculation parameter, is a random number between [0,1], is the convergence factor, q is the current iteration number, and Q is the maximum iteration number.

[0075] 4) Iterative operation: Repeat steps 1) to 3) to update the prey position until the preset condition is met or the number of iterations reaches the maximum number of iterations.

[0076] However, since the whale algorithm may fall into a local optimal solution when searching for the best solution, the embodiment of the invention proposes to use an improved whale algorithm. By introducing the Tent mapping, the whale group is evenly distributed. Chaos has randomness, ergodicity, and initial value sensitivity, which can make the algorithm converge faster. To a certain extent, it improves the problem that the original algorithm is prone to fall into a local optimal solution.

[0077] See also Figure 2 , Figure 2 for Figure 1 Step S3 in the illustrated embodiment is a flow chart in an exemplary embodiment. Figure 2 As shown, step S3 may specifically include steps S31 to S32. The improved whale algorithm is obtained through the above steps, which is described in detail as follows:

[0078] S31, in the stage of initializing the whale population, a set of random kernel parameters and penalty factors are generated as candidate solutions, each solution represents a possible parameter combination;

[0079] S32, the population position is initialized by Tent chaotic mapping, and the improved whale algorithm is obtained by adding the whale optimization algorithm with adaptive weight factors.

[0080] Specifically, the population position is initialized by the Tent chaotic map, and the formula is as follows:

[0081]

[0082] Among them, X m Represents the mth mapping function value, where m represents the number of mappings, and X 0 is a random value in [0, 1].

[0083] The improved whale algorithm is obtained by adding the whale optimization algorithm with adaptive weight factor. The formula is as follows:

[0084]

[0085]

[0086]

[0087] In the formula, is the best whale coordinate at time t+1, ω is an adaptive weight factor randomly selected in [0.4, 0.9], X * (t) is the coordinate of the best whale position at time t, and D rand is the intermediate calculation parameter, is a random number between [0,1], is the convergence factor, q is the current number of iterations, Q is the maximum number of iterations, is the coordinate of the whale currently traveling at time t, X rand (t) is the position of a whale randomly selected from the group at time t, k is the whale rotation constant, l is a random value in the range (-1,1), and p is a random value in the range [0,1].

[0088] That is to say, in the improved whale algorithm in the embodiment of the present invention, the encirclement predation is implemented by formulas (4) and (5), the spiral bubble predation is implemented by formula (11), and the global search predation is implemented by formulas (9) and (10).

[0089] In the embodiment of the present invention, the Tent chaotic mapping is used to disrupt the input sample values, thereby maintaining population diversity. In addition, the Tent chaotic mapping has ergodicity and uniformity, which can improve the global search capability and avoid falling into a local optimal solution.

[0090] See also Figure 3 , Figure 3 for Figure 1 Step S4 in the illustrated embodiment is a flow chart in an exemplary embodiment. Figure 3 As shown, step S4 may specifically include step S41 to step S42. The optimal kernel parameter and penalty factor are obtained through the above steps, which are described in detail as follows:

[0091] S41, through the improved whale algorithm, encirclement predation, spiral bubble predation and global search predation are carried out, and repeated iterations are performed.

[0092] It should be noted that, in the embodiment of the present invention, the improved whale algorithm uses formulas (4) and (5) when performing encirclement predation, uses formula (11) when performing spiral bubble predation, and uses formulas (9) and (10) when using global search predation.

[0093] S42, when the prey position is updated to reach a preset condition or the number of iterations reaches a maximum number of iterations, the optimal kernel parameter and penalty factor are obtained.

[0094] The improved whale algorithm used in the embodiment of the present invention effectively solves the problems of low solution accuracy, slow convergence speed, and easy falling into local optimal solutions of the traditional whale algorithm. The method provided by the embodiment of the present invention has the advantages of high convergence accuracy and good algorithm stability compared with the traditional whale algorithm.

[0095] Figure 5The flowchart of the improved whale algorithm optimization process is shown in an exemplary embodiment of the present application. Figure 5 As shown in Figure 1, in the initialization stage of the whale group, a set of random kernel parameters and penalty factors are generated as candidate solutions, each of which represents a possible parameter combination; then the population position is initialized through the Tent chaotic map, the fitness value of the population individuals is calculated, and the current optimal individual position is recorded. The weight coefficient is brought in to perform the encirclement and predation formulas (4) and (5), thereby updating The intermediate parameter is then used to determine whether p is less than 0.5. If so, the next step is to be screened. Is it less than 1? Put it into formulas (9) and (10) respectively. If p is greater than or equal to 0.5, put it into formula (11). After reaching the maximum number of iterations, output the optimal penalty factor c and kernel function parameter g.

[0096] Since the penalty factor and kernel parameter in the SVR model are difficult to determine, and the penalty factor and kernel parameter play a vital role in the model calculation. Therefore, the role of the improved whale algorithm in the embodiment of the present invention is reflected. The optimal penalty factor and kernel parameter are obtained by iterating the improved whale algorithm and inputting them into the SVR model, so that the accuracy and robustness of the model are improved and the convergence rate is accelerated.

[0097] Optionally, Figure 1 Step S5 in the illustrated embodiment includes steps S51 to S52 in an exemplary embodiment. The ice cover prediction model is obtained through the above steps, which is described in detail as follows:

[0098] S51, dividing the combined sample set into a training set and a test set according to a preset ratio.

[0099] Specifically, the combined sample set can be divided into a training set and a test set in a ratio of 3:1.

[0100] S52, using a nonlinear transformation function to map the training set samples from the low-dimensional space to the high-dimensional space, and constructing an optimal linear regression function.

[0101] Specifically, a nonlinear transformation φ(x) is defined to transform the training set sample vector with N-dimensional input and one-dimensional output:

[0102]

[0103] Map the low-dimensional space to the high-dimensional space and construct the optimal linear regression function, as shown in formula (12):

[0104] y(x)=φ(x)·ω+z(12).

[0105] S53, defining boundary constraints in high-dimensional space;

[0106]

[0107] Where θ is the weight vector, z is the threshold, and ε is the tolerance loss.

[0108] S54, under all given constraints, a kernel function is selected to obtain a trained support regression vector machine.

[0109] The selected kernel function is RBF (Radial Basic function) kernel function K(x,y) formula (14) is as follows:

[0110] K(x,y)=exp(-||xy|| 2 / 2σ 2 )(14)

[0111] Where σ is the scale parameter.

[0112] In the embodiment of the present invention, a training set is used to train the SVR model, and a test set is input into the prediction model for prediction; a final ice cover prediction value is obtained, and a correlation error calculation is performed on the prediction result.

[0113] Calculate the relevant evaluation indicators of the total predicted value of ice thickness; the three evaluation indicators used are mean square error, mean absolute error and determination coefficient;

[0114] The calculation formula of mean square error (15) is as follows:

[0115]

[0116] The calculation formula (16) of the mean absolute error is as follows:

[0117]

[0118] The calculation formula of the coefficient of determination (17) is as follows:

[0119]

[0120] In the formula, n is the number of samples, is the predicted value of the model, y i is the true value of the sample.

[0121] In the embodiment of the present invention, the ice thickness is predicted by the optimized SVR model, which improves the prediction accuracy and has a faster convergence speed. Compared with other models, the SVR model has the advantages of high-precision prediction, strong generalization ability, strong robustness and the ability to operate in a small sample space. And after the penalty factor and kernel parameter obtained by the improved whale algorithm, the accuracy of the model is further improved.

[0122] The present invention is described in detail below with reference to a specific example.

[0123] Based on the above-mentioned RF-MWOA-SVR (random forest algorithm-improved whale optimization algorithm-support vector regression machine) transmission line short-term icing prediction method, the training samples selected actual monitoring data from a certain place in Henan Province. The main monitoring data indicators include meteorological data and ice thickness data of the transmission line.

[0124] According to the real-time meteorological data obtained from the online monitoring of Henan Power Grid, the seven given micro-meteorological factors were analyzed, and the random forest algorithm was used to calculate the influence of temperature, humidity, wind speed, wind direction, rainfall, atmospheric pressure, and light radiation intensity on the growth of ice cover, and the influence of each micro-meteorological factor on the ice thickness. Figure 6 This is a schematic diagram of the screening results of micro-meteorological factors using the random forest algorithm, which is an exemplary embodiment of the present application. Figure 6 It can be seen that the four input characteristic variables of temperature, humidity, wind speed and wind direction all have an influence on ice growth of more than 97.5%. Therefore, the embodiment of the present invention uses these four variables as input of the ice growth model and uses the equivalent ice thickness as the output of the model.

[0125] In the example of the present invention, four characteristic values ​​with high correlation are retained, and the data set includes temperature, humidity, wind speed, wind direction and ice thickness.

[0126] The influencing factors and ice thickness are normalized by a normalization method. In order to improve the accuracy of the model, the improved whale algorithm MWOA is introduced in the embodiment of the invention to optimize the penalty factor c and the kernel function parameter g in the slack variable in the SVR model. The optimization parameter dimension is set to 2, the lower limit of the parameter is [0.01, 1], the upper limit of the parameter is [10, 100], and the maximum number of iterations is 100.

[0127] After optimizing SVR with the MWOA algorithm, the values ​​of parameters c and g are 88.2498 and 0.8709, respectively.

[0128] Table 1 is an evaluation table of the prediction effect of the RF-MWOA-SVR transmission line short-term icing prediction and the traditional SVM model provided by the embodiment of the present invention. Figure 7 This is a schematic diagram of a test set prediction value change curve for an exemplary embodiment of the present application; Figure 8 This is a prediction error comparison diagram shown in an exemplary embodiment of the present application.

[0129] Table 1

[0130]

[0131] According to Table 1, Figure 7 and Figure 8 It can be seen that the average relative error of the present invention is 12.17%, and the average relative error of the traditional SVM model is 23.98%; it can be seen that the RF-MWOA-SVR prediction model proposed in the embodiment of the present invention can further reduce the prediction error, improve the prediction accuracy, and has a faster convergence speed. With the addition of the random forest algorithm and the improved whale algorithm, the model has better robustness and is not easy to fall into the local optimal solution.

[0132] According to the above description, for the short-term memory network model, the improved whale algorithm used in the embodiment of the present invention reduces the probability of falling into the local optimal solution and improves the robustness of the algorithm compared to the original whale algorithm model.

[0133] Different from traditional regression methods, SVR transforms the regression problem into the problem of solving a set of support vectors by using the idea of ​​support vector machine (SVM). It can handle nonlinear regression problems and map data to high-dimensional space by selecting appropriate kernel functions, thereby improving the adaptability of the model.

[0134] By running the improved whale algorithm to optimize the parameters in SVR, and then using the optimized SVR model to predict ice thickness, the prediction accuracy is improved while the model converges faster. With the addition of the random forest algorithm and the improved whale algorithm, the model has better robustness and is not prone to falling into the local optimal solution.

[0135] Fig. 9 FIG. 9 is a block diagram of a short-term icing prediction system 900 for a power transmission line according to an exemplary embodiment of the present application. Fig. 9 As shown, the transmission line short-term icing prediction system 900 includes a sample set construction module 901, a sample set screening module 902, a whale algorithm improvement module 903, an optimal parameter solving module 904 and a model training and application module 905.

[0136] The sample set construction module 901 is used to obtain the micro-meteorological data and equivalent ice thickness data of the local power transmission line and construct the input and output sample sets;

[0137] The sample set screening module 902 is used to screen the feature vectors of the micrometeorological data in the input and output sample sets by using a random forest algorithm, and to combine the obtained feature vectors with the equivalent ice thickness data to obtain a combined sample set;

[0138] The whale algorithm improvement module 903 is used to initialize the whale optimization algorithm population and obtain the improved whale algorithm based on the adaptive weight factor and the Tent chaotic map;

[0139] The optimal parameter solving module 904 is used to search for the optimal solution of the penalty factor and the kernel function parameters by improving the whale algorithm, and use the optimal solution as the model parameter of the support vector regression model;

[0140] The model training and application module 905 is used to train and test the support vector regression model using the combined sample set to obtain an icing prediction model, and perform short-term icing prediction of the transmission line based on the icing prediction model.

[0141] An embodiment of the present invention further provides a computer-readable storage medium, comprising instructions, which, when executed on a computer, enable the computer to execute the method for predicting short-term icing of a transmission line as provided in any of the above embodiments.

[0142] An embodiment of the present invention further provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a method for predicting short-term icing of a transmission line as provided in any of the above embodiments is implemented.

[0143] The embodiment of the present invention also provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the short-term icing prediction method for a transmission line provided in the above-mentioned various optional embodiments.

[0144] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0145] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0146] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0147] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0148] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-On l yMemory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0149] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for predicting short-term icing of a transmission line, characterized in that: The steps include: Obtain micro-meteorological data and equivalent ice thickness data of local transmission lines and construct input and output sample sets; Performing feature vector screening on the micrometeorological data in the input and output sample set by using a random forest algorithm, combining the obtained feature vector with equivalent ice thickness data to obtain a combined sample set; Initialize the whale optimization algorithm population, and obtain the improved whale algorithm based on the adaptive weight factor and Tent chaotic mapping; The improved whale algorithm is used to search for the optimal solution of the penalty factor and the kernel function parameter, and the optimal solution is used as the model parameter of the support vector regression model; The support vector regression model is trained and tested using the combined sample set to obtain an icing prediction model, and short-term icing prediction of the transmission line is performed based on the icing prediction model.

2. The method according to claim 1, characterized in that The initialization of the whale optimization algorithm population, and obtaining an improved whale algorithm based on an adaptive weight factor and a Tent chaotic map, include: In the whale population initialization stage, a set of random kernel parameters and penalty factors are generated as candidate solutions, each of which represents a possible parameter combination; The population position is initialized by Tent chaotic mapping, and the improved whale algorithm is obtained by adding the whale optimization algorithm with adaptive weight factors.

3. The method according to claim 2, characterized in that The population position is initialized by Tent chaotic mapping, and the formula is as follows: Among them, X m Represents the mth mapping function value, where m represents the number of mappings and X0 is a random value in [0, 1].

4. The method according to any one of claims 1 to 3, characterized in that: The method of performing optimization through the improved whale algorithm to obtain the optimal solution of the penalty factor and the kernel function parameter includes: The improved whale algorithm is used to perform encirclement predation, spiral bubble predation and global search predation, and repeated iterations are performed: When the prey position is updated to meet the preset conditions or the number of iterations reaches the maximum number of iterations, the optimal kernel parameters and penalty factors are obtained.

5. The method according to claim 4, characterized in that The encirclement predation formula is as follows: in, is the coordinate of the whale currently traveling at time t+1, is the best whale coordinate at time t, and is the intermediate calculation parameter, is a random number between [0, 1], is the convergence factor, q is the current number of iterations, Q is the maximum number of iterations, is the coordinate of the whale currently traveling at time t.

6. The method according to claim 4, characterized in that The spiral bubble predation and global search predation formulas are as follows: In the formula, is the best whale coordinate at time t+1, ω is an adaptive weight factor randomly selected in [0.4, 0.9], X * (t) is the coordinate of the best whale position at time t, and D rand is the intermediate calculation parameter, is a random number between [0, 1], is the convergence factor, q is the current number of iterations, Q is the maximum number of iterations, is the coordinate of the whale currently traveling at time t, X rand (t) is the position of a whale randomly selected from the group at time t, k is the whale rotation constant, l is a random value in the range of (-1, 1), and p is a random value in the range of [0, 1].

7. The method according to any one of claims 1 to 3, characterized in that: Before the feature vector screening is performed on the micrometeorological data in the input and output sample sets by the random forest algorithm, the method further includes: normalizing the data in the input and output sample sets.

8. A short-term icing prediction system for power transmission lines, characterized in that: include: The sample set construction module is used to obtain the micro-meteorological data and equivalent ice thickness data of the local transmission lines and construct the input and output sample sets; A sample set screening module, used to screen the feature vectors of the micrometeorological data in the input and output sample sets by using a random forest algorithm, and combine the obtained feature vectors with the equivalent ice thickness data to obtain a combined sample set; Whale algorithm improvement module, used to initialize the whale optimization algorithm population, and obtain the improved whale algorithm based on adaptive weight factor and Tent chaotic mapping; An optimal parameter solving module, used to perform optimization through the improved whale algorithm to obtain the optimal solution of the penalty factor and the kernel function parameters, and use the optimal solution as the model parameters of the support vector regression model; The model training and application module is used to train and test the support vector regression model using the combined sample set to obtain an icing prediction model, and perform short-term icing prediction of the transmission line based on the icing prediction model.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a computer, the computer is enabled to execute the method for predicting short-term icing of a power transmission line according to any one of claims 1 to 7.

10. An electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for predicting short-term icing of a transmission line as claimed in any one of claims 1 to 7 when executing the program.

Images