Transformer load rate prediction method based on neural network and related device

Abstract

The invention provides a neural network-based transformer load rate prediction method and a related device, and the method comprises the steps: collecting the parameter information of a transformer, inputting the parameter information into a feature selection model, and carrying out the screening, thereby obtaining a feature subset; features in the screened optimal feature subset are highly associated with the load rate of the transformer, so that the example loss can be effectively reduced, the convergence speed of the prediction model is accelerated, and redundancy in original data is eliminated; constructing a transformer load rate prediction model, optimizing the transformer load rate prediction model based on the Bayesian theory, and introducing probability distribution into a neural network, so that the optimized neural network can take uncertainty among variables influencing the transformer load rate into consideration; training the optimized transformer load rate prediction model by using the feature subset to obtain prediction data; the problems of overfitting and local minimization can be avoided, the convergence rate is improved, the prediction result is more accurate, and the generalization performance is improved.

Description

Technical Field

[0001] The present invention relates to the field of transformer technology, and in particular to the field of transformer load prediction technology, and specifically to a transformer load rate prediction method based on a neural network and a related device. Background Art

[0002] With the rapid development of social economy, the demand for electricity in various industries continues to rise, especially during the peak load period such as summer, the electricity consumption of residents increases sharply, causing many transformers in operation to face severe overload problems. Load rate is an important operating indicator of distribution transformers. Light load or heavy load will lead to increased line loss, and long-term heavy overload will not only affect the life of distribution transformers, but also cause low voltage problems at the end of the line, making the user's electricity experience worse.

[0003] With the development of neural networks in recent years, many researchers have used the adaptive and learning capabilities of neural networks to predict transformer load rates, trying to control the heavy overload problem of transformers from the perspective of prior management. Due to the nonlinear mapping ability of neural networks, they can play a certain role in describing complex relationships such as the relationship between transformer operating data and load rates. However, as the actual operating conditions become more complex and the transformer operating data gradually increases, redundancy problems are likely to occur. When such operating data is used as the basic input feature for transformer load rate prediction, the neural network model will be affected by the redundancy problem, which will reduce the calculation rate and convergence of the model, thereby affecting the final calculation accuracy.

[0004] Therefore, there are technical problems in the prior art that the existing transformer load rate prediction model is prone to redundancy and low calculation accuracy. Summary of the invention

[0005] The main purpose of the present invention is to provide a transformer load rate prediction method and related devices based on neural network to solve the technical problems that the existing transformer load rate prediction model in the prior art is prone to redundancy and low calculation accuracy.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of the present invention, a transformer load rate prediction method based on a neural network is provided, comprising the following steps: collecting parameter information of the transformer, wherein the parameter information includes basic information and operation information of the transformer; constructing a feature selection model, inputting the parameter information into the feature selection model for screening to obtain a feature subset; constructing a transformer load rate prediction model, wherein the prediction model is a back propagation neural network model optimized by a genetic algorithm; optimizing the transformer load rate prediction model based on Bayesian theory; and training the optimized transformer load rate prediction model using a feature subset to obtain prediction data.

[0007] In some embodiments, constructing a feature selection model includes: constructing an XGboost feature selection model; and optimizing parameters of the XGboost feature selection model based on Bayesian theory, wherein the Bayesian theory model is:

[0008] Among them, μ t-1 (x) is the mean, σ t-1 (x) is the standard deviation, x t is the next sampling position, is the weight parameter, and the input position with the maximum sum of the mean and standard deviation is selected as the next sampling point.

[0009] In some embodiments, the transformer load rate prediction model is constructed, and the prediction model is a back propagation neural network model optimized by a genetic algorithm, including: using a back propagation neural network to construct an initial transformer load rate prediction model; using a genetic algorithm to optimize the weights and thresholds of the initial transformer load rate prediction model to obtain a back propagation neural network model.

[0010] In some implementations, the optimizing the transformer load rate prediction model based on the Bayesian theory includes: optimizing the intermediate layer parameters of the transformer load rate prediction model using the Bayesian theory.

[0011] In some embodiments, the optimization of the intermediate layer parameters of the transformer load rate prediction model using Bayesian theory includes: the maximum likelihood function of the network output obeys a Gaussian distribution, whose mean is related to D, determined by determining the trained network, and the variance is Then we have:

[0012]

[0013] The prior probability distribution of the weight w and the prior probability distribution of the bias b are set to Gaussian distribution, in the form of:

[0014]

[0015] For N times independent and identical distribution, the likelihood function is:

[0016] In some embodiments, the XGboost feature selection model is:

[0017]

[0018] In the formula, for y i Prediction value, f k is a single regression tree, D represents the set of all regression trees, f k (xi ) represents the score of the i-th sample in the data set corresponding to the k-th tree; the objective function of the model is:

[0019]

[0020] Among them, l is the error function of the performance model fitting training samples, Ω(f k ) is the penalty function.

[0021] In some embodiments, the penalty function is:

[0022]

[0023] Among them, Υ and λ are model penalty coefficients, T and w represent the number of leaves and corresponding weights of the kth tree respectively.

[0024] According to another aspect of the present invention, the present invention also provides a transformer load rate prediction device based on a neural network, comprising: an acquisition module, used to collect parameter information of the transformer, the parameter information including basic information and operation information of the transformer; an acquisition module, used to construct a feature selection model, and input the parameter information into the feature selection model for screening to obtain a feature subset; a construction module, used to construct a transformer load rate prediction model, the prediction model is a back propagation neural network model optimized by a genetic algorithm; an optimization module, used to optimize the transformer load rate prediction model based on Bayesian theory; a prediction module, used to train the optimized transformer load rate prediction model using a feature subset to obtain prediction data.

[0025] According to another aspect of the present invention, the present invention further provides a storage medium storing a computer program, wherein when the computer program is executed by a processor, the steps of any of the aforementioned neural network-based transformer load rate prediction methods are implemented.

[0026] According to another aspect of the present invention, the present invention also provides an electronic device, comprising at least a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program on the memory, the steps of any of the aforementioned neural network-based transformer load rate prediction methods are implemented.

[0027] The transformer load rate prediction method and related device based on neural network provided by the present invention are applied, and parameter information of the transformer is collected, and the parameter information is input into the feature selection model for screening to obtain a feature subset; the features in the screened optimal feature subset are highly correlated with the transformer load rate, which can effectively reduce the example loss, accelerate the convergence speed of the prediction model, eliminate the redundancy in the original data, and improve the prediction accuracy of the prediction model; a transformer load rate prediction model is constructed, and the transformer load rate prediction model is optimized based on Bayesian theory, and probability distribution is introduced into the neural network, so that the optimized GA-BPNN neural network can take into account the uncertainty between the variables affecting the transformer load rate; the optimized transformer load rate prediction model is trained using the feature subset to obtain prediction data; it can avoid overfitting and local minimization problems, improve the convergence rate, make the prediction result more accurate, and improve the generalization performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The drawings constituting a part of the present application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0029] Figure 1 A flowchart of the steps of the transformer load rate prediction method based on a neural network according to the present invention is shown;

[0030] Figure 2 is a flowchart of XGboost feature selection provided by the present disclosure;

[0031] Figure 3 is a flowchart of the steps of constructing a transformer load rate prediction model provided by the present disclosure;

[0032] Figure 4 It is a schematic diagram of the structure of the back propagation neural network provided by the present disclosure;

[0033] Figure 5 It is a structural block diagram of a transformer load rate prediction device based on a neural network provided by the present disclosure;

[0034] Figure 6 It is a schematic diagram of the structure of the electronic device provided by the present disclosure. DETAILED DESCRIPTION

[0035] The specific embodiments of the present disclosure are described in detail below with reference to the accompanying drawings, but are not intended to limit the present disclosure.

[0036] It should be understood that various modifications may be made to the embodiments disclosed herein. Therefore, the above description should not be considered as limiting, but merely as an example of an embodiment. Other modifications within the scope and spirit of the present disclosure will occur to those skilled in the art.

[0037] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present disclosure and, together with the general description of the present disclosure given above and the detailed description of the embodiments given below, serve to explain the principles of the present disclosure.

[0038] These and other characteristics of the present disclosure will become apparent from the following description of a preferred form of embodiment given as a non-limiting example, with reference to the attached drawings.

[0039] It should also be understood that, although the present disclosure has been described with reference to some specific examples, those skilled in the art will be able to realize many other equivalent forms of the present disclosure that have the features described in the claims and are therefore within the scope of protection defined thereby.

[0040] The above and other aspects, features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.

[0041] Specific embodiments of the present disclosure are described hereinafter with reference to the accompanying drawings; however, it should be understood that the disclosed embodiments are merely examples of the present disclosure, which may be implemented in a variety of ways. Well-known and / or repeated functions and structures are not described in detail to avoid obscuring the present disclosure with unnecessary or redundant details. Therefore, the specific structural and functional details disclosed herein are not intended to be limiting, but merely serve as a basis and representative basis for the claims to teach those skilled in the art to use the present disclosure in a variety of ways with substantially any suitable detailed structure.

[0042] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.

[0043] This specification may use the phrases "in one embodiment," "in another embodiment," "in a further embodiment," or "in other embodiments," all of which may refer to one or more of the same or different embodiments according to the present disclosure.

[0044] The present disclosure is further described below in conjunction with the accompanying drawings and specific embodiments.

[0045] In order to solve the technical problems in the prior art that the existing transformer load rate prediction model is prone to redundancy and low calculation accuracy, the present invention provides a transformer load rate prediction based on a neural network and a related device.

[0046] Figure 1 The flowchart of the step of the transformer load rate prediction method based on neural network provided by the present invention is shown as follows: Figure 1 As shown, the transformer load rate prediction method based on neural network includes the following steps:

[0047] S101, collecting parameter information of a transformer, where the parameter information includes basic information and operation information of the transformer.

[0048] In this step, parameter information of the transformer is collected, and the parameter information includes basic information and operation information of the transformer.

[0049] Specifically, in the transformer load rate prediction process, the transformer parameter information is first collected as initial input data, wherein the parameter information includes basic information and operation information of the transformer. The basic information of the transformer includes the capacity, model, impedance and other information of the transformer, and the operation information of the transformer includes the wiring mode, gear and other information of the transformer.

[0050] The collected transformer parameter information is defined as features, and a data set D is formed, where D is:

[0051] D={(x i ,y i ),i=1,2,3,...,m,x i ∈R M ,y i ∈R};

[0052] Where R is a real number set, m represents the number of samples, and each sample contains n features, that is, the sample contains n independent running data. Each transformer is defined as: i Corresponding to an independent y i Value, y i Represents x under different running data i Description.

[0053] S102, constructing a feature selection model, inputting the parameter information into the feature selection model for screening to obtain a feature subset.

[0054] After completing the above step S101, in this step, a feature selection model is constructed, and the parameter information is input into the feature selection model for screening to obtain a feature subset.

[0055] Specifically, the parameter information data of the transformer is input into a feature selection model for screening to obtain a feature subset, wherein the feature selection model can use XGboost as a feature selection model. Figure 2 The XGboost feature selection flow chart provided by the present disclosure is shown. The XGboost algorithm can automatically perform multi-threaded parallel computing, which greatly improves the speed of computing operation. At the same time, a regularization term consisting of the complexity of the regression tree is added to the model, which effectively improves the generalization ability of the model.

[0056] In the process of XGboost model construction, assuming that the number of regression trees in the model is k, the model is:

[0057]

[0058] in, for y i Prediction value, f k is a single regression tree, D represents the set of all regression trees, f k (x i ) represents the score of the i-th sample in the data set corresponding to the k-th tree.

[0059] The objective function of the model is: Among them, l is the error function of the performance model fitting training samples.

[0060] To avoid overfitting, we set the regularization term Ω(f k ) to penalize complex models, where Ω(f k )for:

[0061]

[0062] Among them, Υ and λ are model penalty coefficients, T and w represent the number of leaves and corresponding weights of the kth tree respectively.

[0063] Since the model is an additive tree model, the training of the objective function can be achieved through a forward step-by-step algorithm.

[0064] Specifically, suppose that after t iterations, the predicted value of sample i is Then the optimization of the objective function can be achieved by introducing f t The optimized objective function is:

[0065]

[0066] in, is the loss function, f i (x i ) is the predicted value of the t-th tree, and Ωf(t) is the model complexity of the t-th tree.

[0067] In order to improve the training speed of the model, the above objective function is expanded by the second order Taylor and the constant term is removed to obtain:

[0068]

[0069] Among them, g i and h i Corresponding to the first and second derivatives in the second-order Taylor expansion, since they are based on the first k-1 models, they are also known when training the kth model and can be regarded as constant terms. At the same time, in the above objective function, It can also be regarded as a constant term, and this term is not multiplied by the variable, so the expanded objective function is simplified again, and the result is:

[0070]

[0071] Among them, I j ={i|q(x i =j)} is the x that is assigned to the jth leaf node i The sequence number set represents the sample set of leaf j.

[0072] At this point, the objective function is converted into a j The system of equations is used to solve the minimum problem. The tree structure q remains unchanged, and the optimal weight corresponding to leaf j is:

[0073] in, They represent the sum of the first-order and second-order derivatives of the loss function, respectively, and λ is the model penalty coefficient.

[0074] The corresponding optimal target value can be obtained:

[0075] in, As a measure of the tree's structural score, in general, the value is inversely proportional to the tree's structural performance.

[0076] It is unrealistic to completely enumerate all the structures of the tree, so the greedy algorithm is combined to divide the subtree of the current node, introduce split points to the existing leaf nodes, and list all feasible split points, among which the split with the smallest function and the largest gain is selected.

[0077] Specifically, the gain formula used is:

[0078] in, I L and I R Respectively represent the sample sets of the left subtree and the right subtree after classification. The classification effect can be scored by using the gain formula. In actual prediction, classification will be based on the leaf node with the largest gain.

[0079] In order to improve the accuracy of the selection results of the XGboost feature selection model, in one embodiment of the present disclosure, the parameters of the XGboost feature selection model are optimized based on the Bayesian theory. In the process of optimizing the parameters of the original XGboost model, the Bayesian theory is introduced to take uncertainty into consideration, which can effectively improve the classification performance of the model; wherein, the Bayesian theory model is:

[0080] Among them, μ t-1 (x) is the mean, σ t-1 (x) is the standard deviation, x t is the next sampling position, is the weight parameter, and the input position with the maximum sum of the mean and standard deviation is selected as the next sampling point.

[0081] A new solution is generated according to the joint distribution function of the Bayesian neural network, and the new solution is substituted into the parameter set to form a new parameter solution. The iteration continues until the manually set termination condition is met. The parameter set output at this time replaces the original XGboost parameter set to complete the XGboost feature selection model after Bayesian optimization.

[0082] S103, constructing a transformer load rate prediction model, wherein the prediction model is a back propagation neural network model optimized by a genetic algorithm.

[0083] After completing the above step S102, in this step, a transformer load rate prediction model is constructed, and the prediction model is a back propagation neural network model optimized by a genetic algorithm.

[0084] Specifically, after obtaining the feature subset, the optimal feature subset obtained after feature selection is input into the neural network for learning. Figure 3 The flowchart of the steps of constructing the transformer load rate prediction model provided by the present disclosure is shown as follows: Figure 3 As shown in FIG. 1 , constructing a transformer load rate prediction model includes the following steps:

[0085] S201, constructing an initial model for transformer load rate prediction using a back propagation neural network.

[0086] In this step, the back propagation neural network is used to build the initial model for transformer load rate prediction.

[0087] Specifically, a back propagation neural network (BPNN) is used to construct a prediction model. The structure of the back propagation neural network is as follows: Figure 4 As shown, where x i is the input data, y n is the output data, w n is the middle layer parameter.

[0088] When the signal strength of the jth neuron in an intermediate layer (et layer) of the neural network is N et,j , its signal strength is: N et,j =∑w ij x i ; where w ij is the weight of the ith neuron acting on the jth neuron, x i is the input signal.

[0089] The threshold of the back propagation neural network is set to θ, when N et,j When >θ, the neuron is activated; the expression of the model is:

[0090] S202, using a genetic algorithm to optimize the weights and thresholds of the initial model for transformer load rate prediction to obtain a back propagation neural network model.

[0091] After completing the above step S201, in this step, a genetic algorithm is used to optimize the weights and thresholds of the initial model for transformer load rate prediction to obtain a back propagation neural network model.

[0092] Specifically, in order to improve the accuracy of model prediction, a genetic algorithm (GA) is used to optimize the weights and thresholds of the back-propagation neural network.

[0093] First, initialize the population and generate a random population X m×n , where the initial weight distribution of the neural network is composed of individual X 1×n Indicates that a single connection weight and threshold are represented by a single value. In this case, the sum of the weight and threshold of the neural network is equal to the individual length:

[0094] n=r×s1+r×s2+s1+s2;

[0095] Among them, n is the individual length, r represents the number of neurons in the input layer, s1 is the number of neurons in the hidden layer, and s2 is the number of neurons in the output layer.

[0096] Combined with the fitness function to evaluate each individual, the input of the back propagation neural network is decoded by all individuals. The fitness function is as follows: Where: E n is the output error value. The individual with the largest fitness calculated will enter the next generation population.

[0097] Among them, the selection of operators is combined with the roulette method. Assuming that the fitness of the i-th individual is f i , then the probability of selection is: Among them, j is the population size.

[0098] The crossover operator is selected using the arithmetic crossover method as follows:

[0099]

[0100] Where: X i (k) and X i+1 (k) are the basis of the i-th and i+1-th individuals at the k-th position, α, β∈[0,1].

[0101] Through the uniform mutation operator, all gene values ​​are mutated at a rate P m Random substitution within the gene domain: X i =X i (p)+r×q+X i (np-1), where q is the threshold width corresponding to the p+1th gene value, and r and n are constants.

[0102] S104, optimizing the transformer load rate prediction model based on Bayesian theory.

[0103] After completing the above step S103, in this step, the transformer load rate prediction model is optimized based on the Bayesian theory.

[0104] Specifically, considering that the transformer load rate prediction problem involves many variables and the nonlinear relationship between the variables is difficult to estimate, the GA-BPNN prediction model cannot take this uncertainty into account. Therefore, the Bayesian theory is introduced to optimize the intermediate layer parameters of the GA-BPNN prediction model, and the determined parameters are converted into probability distribution. The improved Bayesian GA-BPNN neural network can estimate the uncertainty between variables, making the prediction results of the model more realistic. Assume that the maximum likelihood function of the network output obeys a Gaussian distribution, whose mean is related to D, determined by the trained network, and the variance is Then we have:

[0105]

[0106] The prior probability distribution of the weight w and the prior probability distribution of the bias b are set to Gaussian distribution, in the form of:

[0107] For N times independent and identical distribution, the likelihood function is:

[0108]

[0109] By introducing the above-mentioned Bayesian theory into the GA-BPNN prediction model, the improved model can take the uncertainty between each feature into consideration, characterize the uncertainty relationship between each variable, and effectively ensure the accuracy of the prediction model.

[0110] S105, training the optimized transformer load rate prediction model using the feature subset to obtain prediction data.

[0111] After completing the above step S104, in this step, the optimized transformer load rate prediction model is trained using the feature subset to obtain prediction data.

[0112] Specifically, the fitness function is calculated, and after the conditions are met or the maximum number of inheritance times is reached, the optimal individual, that is, the final prediction result, is output.

[0113] The transformer load rate prediction method based on neural network provided by the present invention is applied, and parameter information of the transformer is collected, and the parameter information is input into a feature selection model for screening to obtain a feature subset; the features in the screened optimal feature subset are highly correlated with the transformer load rate, which can effectively reduce the example loss, accelerate the convergence speed of the prediction model, eliminate the redundancy in the original data, and improve the prediction accuracy of the prediction model; a transformer load rate prediction model is constructed, and the transformer load rate prediction model is optimized based on Bayesian theory, and probability distribution is introduced into the neural network, so that the optimized GA-BPNN neural network can take into account the uncertainty between the variables affecting the transformer load rate; the optimized transformer load rate prediction model is trained using the feature subset to obtain prediction data; overfitting and local minimization problems can be avoided, the convergence rate can be improved, the prediction result can be more accurate, and the generalization performance is improved.

[0114] In order to better implement the above method, the second aspect of the present disclosure further provides a transformer load rate prediction device based on a neural network, and the transformer load rate prediction device can be integrated on an electronic device.

[0115] For example, Figure 5 As shown, the prediction device 200 may include: a collection module 210, an acquisition module 220, a construction module 230, an optimization module 240 and a prediction module 250, which are specifically as follows:

[0116] (1) Collection module 210, the collection module 210 is used to collect parameter information of the transformer, the parameter information includes basic information and operation information of the transformer.

[0117] (2) An acquisition module 220, wherein the acquisition module 220 is used to construct a feature selection model and input the parameter information into the feature selection model for screening to obtain a feature subset.

[0118] (3) Construction module 230, the construction module 230 is used to construct a transformer load rate prediction model, and the prediction model is a back propagation neural network model optimized by a genetic algorithm.

[0119] (4) an optimization module 240, the optimization module 240 is used to optimize the transformer load rate prediction model based on Bayesian theory;

[0120] (5) A prediction module 250, which is used to train the optimized transformer load rate prediction model using a feature subset to obtain prediction data.

[0121] Furthermore, the constructing of the feature selection model includes:

[0122] Build the XGboost feature selection model;

[0123] The parameters of the XGboost feature selection model are optimized based on Bayesian theory, where the Bayesian

[0124] The theoretical model is:

[0125] Among them, μ t-1 (x) is the mean, σ t-1 (x) is the standard deviation, x t is the next sampling position, is the weight parameter, and the input position with the maximum sum of the mean and standard deviation is selected as the next sampling point.

[0126] Further, the first determination module 220 includes a first determination unit and a second determination unit, wherein the first determination unit is used to determine the rate of change of the magnetic flux around the seat belt buckle based on the safety parameter information, and the second determination unit is used to determine the state information of the seat belt buckle based on the rate of change of the magnetic flux.

[0127] Furthermore, the construction module 230 includes a construction unit and an optimization unit, wherein the construction unit uses a back-propagation neural network to construct an initial model for transformer load rate prediction; and the optimization unit uses a genetic algorithm to optimize the weights and thresholds of the initial model for transformer load rate prediction to obtain a back-propagation neural network model.

[0128] Furthermore, the optimizing the transformer load rate prediction model based on Bayesian theory includes: optimizing the intermediate layer parameters of the transformer load rate prediction model using Bayesian theory.

[0129] Furthermore, the optimization of the intermediate layer parameters of the transformer load rate prediction model using Bayesian theory includes: the maximum likelihood function of the network output obeys a Gaussian distribution, whose mean is related to D and is determined by the trained network, and the variance is Then we have:

[0130]

[0131] The prior probability distribution of the weight w and the prior probability distribution of the bias b are set to Gaussian distribution, in the form of:

[0132]

[0133] For N times independent and identical distribution, the likelihood function is:

[0134] In some embodiments, the XGboost feature selection model is:

[0135]

[0136] In the formula, for y i Prediction value, f k is a single regression tree, D represents the set of all regression trees, f k (x i ) represents the score of the i-th sample in the data set corresponding to the k-th tree; the objective function of the model is:

[0137]

[0138] Among them, l is the error function of the performance model fitting training samples, Ω(f k ) is the penalty function.

[0139] Furthermore, the penalty function is:

[0140]

[0141] Among them, Υ and λ are model penalty coefficients, T and w represent the number of leaves and corresponding weights of the kth tree respectively.

[0142] The transformer load rate prediction device based on neural network provided by the present invention is applied, and parameter information of the transformer is collected, and the parameter information is input into the feature selection model for screening to obtain a feature subset; the features in the screened optimal feature subset are highly correlated with the transformer load rate, which can effectively reduce the example loss, accelerate the convergence speed of the prediction model, eliminate the redundancy in the original data, and improve the prediction accuracy of the prediction model; a transformer load rate prediction model is constructed, and the transformer load rate prediction model is optimized based on Bayesian theory, and probability distribution is introduced into the neural network, so that the optimized GA-BPNN neural network can take into account the uncertainty between the variables affecting the transformer load rate; the optimized transformer load rate prediction model is trained using the feature subset to obtain prediction data; it can avoid overfitting and local minimization problems, improve the convergence rate, make the prediction result more accurate, and improve the generalization performance.

[0143] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0144] To this end, a third embodiment of the present disclosure provides a storage medium, which is a computer-readable medium and stores a computer program. When the computer program is executed by a processor, the method provided by the embodiment of the present disclosure is implemented, including the following steps S11 to S15:

[0145] S11, collecting parameter information of the transformer, where the parameter information includes basic information and operation information of the transformer;

[0146] S12, constructing a feature selection model, inputting the parameter information into the feature selection model for screening to obtain a feature subset;

[0147] S13, constructing a transformer load rate prediction model, wherein the prediction model is a back propagation neural network model optimized by a genetic algorithm;

[0148] S14, optimizing the transformer load rate prediction model based on Bayesian theory;

[0149] S15, using the feature subset to train the optimized transformer load rate prediction model to obtain prediction data.

[0150] Furthermore, when the computer program is executed by a processor, other methods provided by any of the above embodiments of the present disclosure are implemented.

[0151] The transformer load rate prediction method based on neural network provided by the present invention is applied, and parameter information of the transformer is collected, and the parameter information is input into a feature selection model for screening to obtain a feature subset; the features in the screened optimal feature subset are highly correlated with the transformer load rate, which can effectively reduce the example loss, accelerate the convergence speed of the prediction model, eliminate the redundancy in the original data, and improve the prediction accuracy of the prediction model; a transformer load rate prediction model is constructed, and the transformer load rate prediction model is optimized based on Bayesian theory, and probability distribution is introduced into the neural network, so that the optimized GA-BPNN neural network can take into account the uncertainty between the variables affecting the transformer load rate; the optimized transformer load rate prediction model is trained using the feature subset to obtain prediction data; overfitting and local minimization problems can be avoided, the convergence rate can be improved, the prediction result can be more accurate, and the generalization performance is improved.

[0152] A fourth embodiment of the present disclosure provides an electronic device, such as Figure 6 As shown, the electronic device at least includes a memory 310 and a processor 320. The memory 310 stores a computer program. The processor 320 implements the method provided by any embodiment of the present disclosure when executing the computer program on the memory 310. Exemplarily, the method executed by the electronic device computer program is as follows:

[0153] S21, collecting parameter information of the transformer, where the parameter information includes basic information and operation information of the transformer;

[0154] S22, constructing a feature selection model, inputting the parameter information into the feature selection model for screening to obtain a feature subset;

[0155] S23, constructing a transformer load rate prediction model, wherein the prediction model is a back propagation neural network model optimized by a genetic algorithm;

[0156] S24, optimizing the transformer load rate prediction model based on Bayesian theory;

[0157] S25, using the feature subset to train the optimized transformer load rate prediction model to obtain prediction data.

[0158] In specific implementation, the above-mentioned: acquisition module 210, acquisition module 220, construction module 230, optimization module 240 and prediction module 250 are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to implement corresponding functions.

[0159] The transformer load rate prediction method based on neural network provided by the present invention is applied, and parameter information of the transformer is collected, and the parameter information is input into a feature selection model for screening to obtain a feature subset; the features in the screened optimal feature subset are highly correlated with the transformer load rate, which can effectively reduce the example loss, accelerate the convergence speed of the prediction model, eliminate the redundancy in the original data, and improve the prediction accuracy of the prediction model; a transformer load rate prediction model is constructed, and the transformer load rate prediction model is optimized based on Bayesian theory, and probability distribution is introduced into the neural network, so that the optimized GA-BPNN neural network can take into account the uncertainty between the variables affecting the transformer load rate; the optimized transformer load rate prediction model is trained using the feature subset to obtain prediction data; overfitting and local minimization problems can be avoided, the convergence rate can be improved, the prediction result can be more accurate, and the generalization performance is improved.

[0160] The storage medium may be included in the electronic device, or may exist independently without being installed in the electronic device.

[0161] The storage medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains at least two Internet Protocol addresses; sends a node evaluation request including at least two Internet Protocol addresses to a node evaluation device, wherein the node evaluation device selects an Internet Protocol address from the at least two Internet Protocol addresses and returns it; receives the Internet Protocol address returned by the node evaluation device; wherein the obtained Internet Protocol address indicates an edge node in a content distribution network.

[0162] Alternatively, the storage medium carries one or more programs, and when the one or more programs are executed by the electronic device, the electronic device: receives a node evaluation request including at least two Internet Protocol addresses; selects an Internet Protocol address from the at least two Internet Protocol addresses; and returns the selected Internet Protocol address; wherein the received Internet Protocol address indicates an edge node in a content distribution network.

[0163] Computer program code for performing operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the passenger computer, partially on the passenger computer, as a separate software package, partially on the passenger computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the passenger computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0164] It should be noted that the storage medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any storage medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0165] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0166] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware, wherein the name of a unit does not, in some cases, limit the unit itself.

[0167] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0168] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0169] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0170] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0171] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.

[0172] Multiple embodiments of the present disclosure are described in detail above, but the present disclosure is not limited to these specific embodiments. Those skilled in the art can make various variations and modifications to the embodiments based on the concepts of the present disclosure, and these variations and modifications should all fall within the scope of protection required by the present disclosure.

Claims

1. A transformer load rate prediction method based on neural network, characterized in that: The steps include: Collecting parameter information of the transformer, wherein the parameter information includes basic information and operation information of the transformer; Constructing a feature selection model, inputting the parameter information into the feature selection model for screening to obtain a feature subset; Constructing a transformer load rate prediction model, wherein the prediction model is a back propagation neural network model optimized by a genetic algorithm; Optimizing the transformer load rate prediction model based on Bayesian theory; The optimized transformer load rate prediction model is trained using the feature subset to obtain prediction data.

2. The transformer load rate prediction method based on neural network according to claim 1 is characterized in that: The constructing of the feature selection model comprises: Build the XGboost feature selection model; The parameters of the XGboost feature selection model are optimized based on the Bayesian theory, where the Bayesian theory model is: Among them, μ t-1 (x) is the mean, σ t-1 (x) is the standard deviation, x t is the next sampling position, is the weight parameter, and the input position with the maximum sum of the mean and standard deviation is selected as the next sampling point.

3. The transformer load rate prediction method based on neural network according to claim 1 is characterized in that: The transformer load rate prediction model is constructed, and the prediction model is a back propagation neural network model optimized by a genetic algorithm, including: The back propagation neural network is used to construct the initial model for transformer load rate prediction; The genetic algorithm is used to optimize the weights and thresholds of the initial model for transformer load rate prediction to obtain a back propagation neural network model.

4. The transformer load rate prediction method based on neural network according to claim 1 is characterized in that: The optimization of the transformer load rate prediction model based on Bayesian theory includes: Bayesian theory is used to optimize the intermediate layer parameters of the transformer load rate prediction model.

5. The transformer load rate prediction method based on neural network according to claim 4 is characterized in that: The optimization of the intermediate layer parameters of the transformer load rate prediction model using the Bayesian theory includes: The maximum likelihood function of the network output follows a Gaussian distribution, whose mean is related to D and is determined by the trained network, and the variance is Then we have: The prior probability distribution of the weight w and the prior probability distribution of the bias b are set to Gaussian distribution, in the form of: For N times independent and identical distribution, the likelihood function is:

6. The transformer load rate prediction method based on neural network according to claim 2 is characterized in that: The XGboost feature selection model is: In the formula, for y i Prediction value, f k is a single regression tree, D represents the set of all regression trees, f k (x i ) represents the score of the i-th sample in the data set corresponding to the k-th tree; the objective function of the model is: Among them, l is the error function of the performance model fitting training samples, Ω(f k ) is the penalty function.

7. The method for predicting transformer load rate based on neural network according to claim 6, characterized in that: The penalty function is: Among them, γ and λ are model penalty coefficients, T and w represent the number of leaves and corresponding weights of the kth tree respectively.

8. A transformer load rate prediction device based on neural network, characterized in that: include: A collection module, used to collect parameter information of the transformer, wherein the parameter information includes basic information and operation information of the transformer; An acquisition module is used to construct a feature selection model and input the parameter information into the feature selection model for screening to obtain a feature subset; A construction module is used to construct a transformer load rate prediction model, wherein the prediction model is a back propagation neural network model optimized by a genetic algorithm; An optimization module, used for optimizing the transformer load rate prediction model based on Bayesian theory; The prediction module is used to train the optimized transformer load rate prediction model using the feature subset to obtain prediction data.

9. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the transformer load rate prediction method based on a neural network as described in any one of claims 1 to 7 are implemented.

10. An electronic device, comprising at least a memory and a processor, wherein a computer program is stored in the memory, wherein: The processor implements the steps of the transformer load rate prediction method based on neural network described in any one of claims 1 to 7 when executing the computer program on the memory.

Images