CN115660194A - Insulator wind deflection angle prediction method and system - Google Patents
Insulator wind deflection angle prediction method and system Download PDFInfo
- Publication number
- CN115660194A CN115660194A CN202211375768.8A CN202211375768A CN115660194A CN 115660194 A CN115660194 A CN 115660194A CN 202211375768 A CN202211375768 A CN 202211375768A CN 115660194 A CN115660194 A CN 115660194A
- Authority
- CN
- China
- Prior art keywords
- factor
- time sequence
- insulator
- phase space
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention belongs to the technical field of power systems, and discloses a method and a system for predicting an insulator wind deflection angle, wherein the method comprises the following steps: determining influence factors of the wind deflection angle of the insulator of the power transmission line, and calculating and sequencing the influence factors to obtain main influence factor variables; determining a variable time sequence of wind deflection angle changes in the attack time of the pre-selected power transmission line insulator from the typhoon according to the main influence factor variables; constructing a multivariable phase space of the wind deflection angle of the insulator according to the variable time sequence; inputting the multivariate phase space into a preset wavelet neural network, and predicting the variable time sequence through the wavelet neural network to obtain an initial predicted value; and carrying out interval division on the initial predicted value based on a nonparametric probability estimation method to obtain a final predicted value, and taking the final predicted value as a predicted value of the wind deflection angle of the insulator. The method can solve the problems of large error of the wind deflection angle prediction result and low reliability in the prior art.
Description
Technical Field
The invention relates to the technical field of power systems, in particular to a method and a system for predicting an insulator wind deflection angle.
Background
The suspension insulator string of the power transmission line can deflect under the action of wind load, and if the wind deflection angle is too large, the clearance between a wire suspended by the insulator string and a tower can be smaller than an insulation clearance allowable value, so that accidents such as wind deflection flashover, tripping and wire burning are caused, the stability and reliability of the power transmission line are seriously influenced, and huge economic loss is caused. Therefore, the prediction of the wind deflection angle of the insulator of the power transmission line is carried out, and the method has practical significance and application value for safe and stable operation of the power transmission line. At present, the study on the wind deflection angle of the insulator mainly comprises the calculation of a wind deflection discharge mechanism and the wind deflection angle, and the prediction study on the wind deflection angle of the insulator is relatively less. The existing power transmission line insulator wind drift angle prediction technology mainly comprises two technologies, wherein one technology is used for predicting the wind drift angle according to the self trend of the insulator wind drift angle; the other method is to predict the wind deflection angle of the insulator according to a BP neural network model, but the two existing methods have the following defects: 1. the wind drift angle of the insulator of the power transmission line is influenced by various factors and has complex relation, and the prediction of the self trend of the wind drift angle of the insulator is inevitable and lacks reliability; 2. the insulator wind deflection angle is predicted by using a traditional BP neural network model, and the degree of freedom, elasticity and plasticity are low, so that the method has no more sensitive approximation capability and stronger fault-tolerant capability, and the wind deflection angle prediction error is larger.
Therefore, how to provide a method and a system for predicting an insulator windage yaw angle with more accurate prediction results is a problem to be solved urgently at present.
Disclosure of Invention
The embodiment of the invention provides a method and a system for predicting a wind deflection angle of an insulator, and aims to solve the problems of large error and low reliability of a wind deflection angle prediction result in the prior art.
The following presents a simplified summary in order to provide a basic understanding of some aspects of the disclosed embodiments. This summary is not an extensive overview and is intended to neither identify key/critical elements nor delineate the scope of such embodiments. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.
According to a first aspect of the embodiment of the invention, a method for predicting an insulator windage yaw angle is provided.
The insulator windage yaw angle prediction method comprises the following steps:
determining influence factors of the wind drift angle of the insulator of the power transmission line, and calculating and sequencing the influence factors to obtain main influence factor variables;
determining a variable time sequence of wind deflection angle changes in the attack time of the pre-selected power transmission line insulator from the typhoon according to the main influence factor variables; constructing a multivariable phase space of the wind deflection angle of the insulator according to the variable time sequence;
inputting the multivariate phase space into a preset wavelet neural network, and predicting the variable time sequence through the wavelet neural network to obtain an initial predicted value;
and carrying out interval division on the initial predicted value based on a nonparametric probability estimation method to obtain a final predicted value, and taking the final predicted value as a predicted value of the wind deflection angle of the insulator.
In one embodiment, the influencing factors include: wind speed, wind power angle, rainfall intensity and humidity.
In one embodiment, the step of calculating and sorting the influencing factors to obtain the main influencing factor variables comprises: and calculating and sequencing the influence factors by a fuzzy analytic hierarchy process to obtain main influence factor variables.
In one embodiment, the step of calculating and sorting the influencing factors by the fuzzy analytic hierarchy process to obtain the main influencing factor variables comprises: comparing every two influencing factors of the same layer according to expert experience to obtain a triangular fuzzy judgment matrix; calculating the proportion of each influence factor of the triangular fuzzy judgment matrix according to a sum-average method to obtain a factor weight matrix, calculating an expected value of the factor weight matrix according to a preset adjusting coefficient, and normalizing the expected value to obtain a factor weight matrix weight vector; and calculating the global weight of each influence factor according to the weight vector of the factor weight matrix, sequencing the obtained weight of each influence factor, and selecting D factors with larger influence as main influence factor variables.
In one embodiment, the step of calculating and sorting the influencing factors by the fuzzy analytic hierarchy process to obtain the main influencing factor variables further comprises: before calculating the global weight of each influencing factor according to the weight vector of the factor weight matrix, checking the consistency of the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix; and when the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix meets consistency check, calculating the global weight of each influencing factor.
In one embodiment, constructing a multivariate phase space of the wind deflection angle of the insulator according to the variable time series comprises: calculating delay time and embedding dimension of each univariate time sequence in the variable time sequence based on a univariate time sequence phase space reconstruction method; and constructing a multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
In one embodiment, constructing the multivariate phase space of the wind drift angle of the insulator according to the variable time series further comprises: before constructing a multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension, performing parameter optimization on the delay time and the embedding dimension of the constructed multivariable phase space based on a chaos time sequence phase space reconstruction C-C method.
In one embodiment, inputting the multivariate phase space into a preset wavelet neural network, and predicting the variable time series through the wavelet neural network to obtain an initial predicted value includes: dividing the wavelet neural network prediction step size into a front N step and a back N-N step; predicting and correcting the first n steps of the variable time sequence by using a wavelet neural network to obtain a predicted time sequence and a predicted value of the first n steps; performing phase space reconstruction on the variable time sequence and the obtained first n-step prediction time sequence, and calculating a maximum Lyapunov exponent of the reconstructed phase space based on a small data amount algorithm; predicting the last N-N steps of the variable time sequence based on the maximum Lyapunov index to obtain a predicted value; and combining the predicted values of the previous N steps and the next N-N steps to obtain the initial predicted value of the variable time sequence.
According to a second aspect of the embodiments of the present invention, there is provided an insulator windage yaw prediction system.
This insulator windage yaw prediction system includes:
the wind deflection angle factor determining module is used for determining the influence factors of the wind deflection angle of the insulator of the power transmission line, and calculating and sequencing the influence factors to obtain main influence factor variables;
the multivariate space construction module is used for determining a variable time sequence of wind deflection angle change in the attack time of the pre-selected power transmission line insulator from the typhoon according to the main influence factor variable; constructing a multivariable phase space of the wind deflection angle of the insulator according to the variable time sequence;
the wind deflection angle prediction module is used for inputting the multivariate phase space into a preset wavelet neural network and predicting the variable time sequence through the wavelet neural network to obtain an initial prediction value;
and the wind deflection angle determining module is used for segmenting the initial predicted value based on a nonparametric probability estimation method to obtain a final predicted value, and taking the final predicted value as the predicted value of the wind deflection angle of the insulator.
In one embodiment, the influencing factors include: wind speed, wind power angle, rainfall intensity and humidity.
In one embodiment, the wind deflection angle factor determining module calculates and sorts the influence factors to obtain the main influence factor variables by a fuzzy analytic hierarchy process when calculating and sorting the influence factors to obtain the main influence factor variables.
In one embodiment, the wind deflection angle factor determination module comprises: the device comprises a judgment matrix determining submodule, a weight vector calculating submodule and a factor sorting submodule, wherein the judgment matrix determining submodule is used for comparing every two influencing factors of the same layer according to expert experience to obtain a triangular fuzzy judgment matrix; the weight vector calculation sub-module is used for calculating the proportion of each influence factor of the triangular fuzzy judgment matrix according to a sum-average method to obtain a factor weight matrix, calculating the expected value of the factor weight matrix according to a preset adjusting coefficient, and normalizing the expected value to obtain a factor weight matrix weight vector; and the factor sorting submodule is used for calculating the global weight of each influence factor according to the factor weight matrix weight vector, sorting the obtained weight of each influence factor and selecting D factors with larger influence as main influence factor variables.
In one embodiment, the wind slip factor determination module further comprises: the matrix checking submodule is used for checking the consistency of the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix before calculating the global weight of each influencing factor according to the factor weight matrix weight vector; and when the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix meets consistency check, the factor sorting submodule calculates the global weight of each influence factor.
In one embodiment, the multivariate spatial construction module comprises: the time and dimension calculation submodule and the phase space construction submodule, wherein the parameter calculation submodule is used for calculating the delay time and the embedding dimension of each univariate time sequence in the variable time sequences based on a univariate time sequence phase space reconstruction method; and the phase space construction submodule is used for constructing a multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
In one embodiment, the multivariate spatial construction module comprises: and the parameter optimization submodule is used for carrying out parameter optimization on the delay time and the embedding dimension of the constructed multivariable phase space based on a chaotic time sequence phase space reconstruction C-C method before constructing the multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
In one embodiment, the wind slip angle prediction module comprises: the wavelet neural network prediction device comprises a step size dividing sub-module, a wavelet prediction sub-module, an index prediction sub-module and a merging sub-module, wherein the step size dividing sub-module is used for dividing the wavelet neural network prediction step size into a front N step and a rear N-N step; the wavelet prediction sub-module is used for predicting and correcting the first n steps of the variable time sequence by using a wavelet neural network to obtain a predicted time sequence and a predicted value of the first n steps; the index prediction submodule is used for carrying out phase space reconstruction on the variable time sequence and the obtained first n-step prediction time sequence and calculating the maximum Lyapunov index of the reconstructed phase space based on a small data amount algorithm; predicting the last N-N steps of the variable time sequence based on the maximum Lyapunov index to obtain a predicted value; and the merging submodule is used for merging the predicted values of the previous N steps and the next N-N steps to obtain the initial predicted value of the variable time sequence.
The technical scheme provided by the embodiment of the invention has the following beneficial effects:
the method calculates and sequences the influence factors of the wind drift angle of the insulator of the power transmission line to obtain the main influence factor variables, so that the wind drift angle of the insulator can be predicted more scientifically and reasonably, and meanwhile, the wind drift angle of the insulator is predicted based on multivariate phase space reconstruction and wavelet neural network, so that the influence of various factors is considered, the advantages of a small neural model are brought into play, and the problem of large prediction deviation in the later period is solved. By using nonparametric probability density estimation, the predicted value can be compartmentalized, and the reliability of the predicted value is improved.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention.
FIG. 1 is a schematic flow diagram illustrating a method for insulator windage yaw prediction, according to an exemplary embodiment;
FIG. 2 is a block diagram of a compact wavelet neural network shown in accordance with an exemplary embodiment;
fig. 3 is a block diagram illustrating a system for predicting an insulator windage yaw according to an exemplary embodiment.
Detailed Description
The following description and the drawings sufficiently illustrate specific embodiments herein to enable those skilled in the art to practice them. Portions and features of some embodiments may be included in or substituted for those of others. The scope of the embodiments herein includes the full ambit of the claims, as well as all available equivalents of the claims. The terms "first," "second," and the like, herein are used solely to distinguish one element from another element without requiring or implying any actual such relationship or order between such elements. In fact, a first element could be termed a second element, and vice versa. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a structure, apparatus, or device that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such structure, apparatus, or device. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another like element in a structure, device, or apparatus that comprises the element. The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
The terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like herein, as used herein, are defined as orientations or positional relationships based on the orientation or positional relationship shown in the drawings, and are used for convenience in describing and simplifying the description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the present invention. In the description herein, unless otherwise specified and limited, the terms "mounted," "connected," and "connected" are to be construed broadly, and may include, for example, mechanical or electrical connections, and communication between two elements, and may include direct connection and indirect connection through intervening media, where the meaning of the terms is to be understood by those skilled in the art as appropriate.
Herein, the term "plurality" means two or more, unless otherwise specified.
Herein, the character "/" indicates that the preceding and following objects are in an "or" relationship. For example, A/B represents: a or B.
Herein, the term "and/or" is an associative relationship describing objects, meaning that three relationships may exist. For example, a and/or B, represents: a or B, or A and B.
It should be understood that, although the steps in the flowchart are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in the figures may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.
The various modules in the apparatus or system of the present application may be implemented in whole or in part by software, hardware, and combinations thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
The embodiments and features of the embodiments of the present invention may be combined with each other without conflict.
Fig. 1 shows an embodiment of a method for predicting an insulator windage yaw according to the present invention.
In this optional embodiment, the insulator windage yaw angle prediction method includes:
step S101, determining influence factors of an insulator wind drift angle of the power transmission line, and calculating and sequencing the influence factors to obtain main influence factor variables;
step S103, determining a variable time sequence of wind deflection angle changes in the typhoon attack time of a pre-selected power transmission line insulator according to the main influence factor variable; constructing a multivariable phase space of the wind deflection angle of the insulator according to the variable time sequence;
step S105, inputting the multivariate phase space into a preset wavelet neural network, and predicting the variable time sequence through the wavelet neural network to obtain an initial predicted value;
and S107, the initial predicted value is partitioned based on a nonparametric probability estimation method to obtain a final predicted value, and the final predicted value is used as the predicted value of the wind deflection angle of the insulator.
In one embodiment, the influencing factors include: wind speed, wind power angle, rainfall intensity and humidity. And when the influence factors are calculated and sorted to obtain the main influence factor variables, the influence factors can be calculated and sorted by a fuzzy analytic hierarchy process to obtain the main influence factor variables. Specifically, the step of calculating and sequencing the influence factors by a fuzzy analytic hierarchy process to obtain the main influence factor variables comprises the following steps:
step 1: and constructing a fuzzy judgment matrix. Comparing every two influencing factors of the same layer according to expert experience to obtain a triangular fuzzy judgment matrix
Is recorded as:
in the formula: (l) ij ,m ij ,u ij ) Representing a fuzzy quantization relationship, matrix, between factors i and j
In (l) ij ,m ij ,u ij ) And (l) ji ,m ji ,u ji ) Reciprocal of each other; l ij 、u ij Are respectively (l) ij ,m ij ,u ij ) Lower and upper bounds of support at factors i and j, m ij Is (l) ij ,m ij ,u ij ) Median value of support at factors i and j, and l ij ≤m ij ≤u ij 。
Step 2: and (5) calculating a weight value. Calculating the proportion of each influence factor of the triangular fuzzy judgment matrix according to a sum-average method to obtain a factor weight matrix H i The following:
and in order to eliminate the influence of subjective consciousness of experts on the judgment matrix, introducing an adjusting coefficient theta, and taking theta>0.5 indicates that the judge has a pessimistic character or a negative emotion at the time of judgment; theta<0.5 indicates that the judge is positive. Calculating the expected value E (H) i ) After normalization, the generated weight vector is:
in the formula I i 、m i 、u i Are each l ij 、m ij 、u ij A normalized value; e is an expected value of the factor weight matrix Hi; e L Factor weight matrix Hi in interval l i ，m i ]Average expected value of; e R Factor weight matrix Hi in interval m i ，u i ]Average expected value of; w is a i Is a factor weight vector.
And step 3: and (5) checking consistency. Judgment matrix
I = j, i is always the diagonal element of ij ＝l ji 、m ij ＝m ji And u ij ＝u ji (ii) a When i ≠ j, there is l ji ≤m ji ≤u ji . Matrix of
The weight of each factor necessarily meets the condition: l ji ≤(w i /w j )≤u ji (ii) a Introducing matrix
Middle amount m in ji Calculating a ratio function G of the weights ji (w i /w j ) Defining a matrix
The consistency check index of (1):
in the formula: when gamma is>e –1 =0.3679, indicating that the weight of the factor satisfies the consistency check; γ =1 indicates that the evaluation matrix is completely consistent. The larger the value of gamma is, the better the corresponding consistency is, and gamma refers to a fuzzy matrix
The consistency check index.
And 4, step 4: when the judgment matrix of each factor layer of the factor global weight meets the consistency check, the global weight of the sub-factors is calculated and set
Weighting the target layer for the first-level criterion layer;
weights for the first-level criterion layer to its sub-criterion layers; then the global weight
Comprises the following steps:
and sorting the weights of the various influence factors obtained by calculation, and selecting D factors with larger influence.
In one embodiment, when a multivariable phase space of the wind drift angle of the insulator is constructed according to a variable time sequence, the delay time and the embedding dimension of each univariate time sequence in the variable time sequence can be calculated based on a univariate time sequence phase space reconstruction method; and constructing a multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
Specifically, according to the correlation analysis of meteorological factors, main factors influencing the wind deflection angle of the insulator of the power transmission line are determined: wind speed x 1 Angle of wind power x 2 Rainfall intensity x 3 Humidity x 4 Isoconstituent D-dimensional multivariate time series { x } i I =1,2, …, D }, where x i ＝[x i (1),x i (2),…,x i (N)] T N is the length of the time sequence, T is a time variable, and the coupling and mutual influence relation among all weather factors is mined. Respectively calculating the delay time t of each univariate time sequence by using the univariate time sequence phase space reconstruction method i And embedding dimension m i Obtaining a phase space of the D-dimensional multivariate time sequence:
in the formula: x is the number of i (t)＝[x i (t),x i (t-t i ),...,x i (t-(m i -1)t i )] T (ii) a t is the time variable t = L, L +1, …, N;
wherein m is i And t i (i =1,2, …, D) are the embedding dimension and delay time, respectively, of the ith univariate chaotic time series.
Before the phase space is constructed, the phase space reconstruction parameters can be optimized, for example, the delay time and the embedding dimension of the constructed multivariable phase space are optimized on the basis of a chaos time sequence phase space reconstruction C-C method. The principle of the chaotic time series-based phase space reconstruction C-C method is as follows:
let chaotic time series x = { x i I =1,2, …, N } to embed dimension m i And a delay time t i Reconstructing the phase space, one can obtain:
X＝{X 1 ,X 2 ,...,X i ,...,X M }
in the formula: x is the reconstruction phase space, X i Phase points in m-dimensional phase space; x i ＝{x i ,x i+t ,…,x i+(m-1)t }; m = N- (M-1) t is the number of points in the phase space, and N is the time series length. Defining the correlation integral of the reconstruction time series:
in the formula: c (m, N, r, t) is an association section; m is the number of points in the phase space; n is the time sequence length; r is the size of the radius of the field, r>0; t is a time delay; d ij ＝||X i -X j || ∞ Is composed of ∞ A norm; theta (-) is a Heaviside unit function, and satisfies:
partitioning a given time series x (N) (N =1,2, …, N) into t disjoint sub-time series, respectively:
calculate statistics S (m, N, r, t) for each subsequence:
in the formula, C s And (4) integrating the correlation of the S-th subsequence. Define the maximum deviation with respect to r:
ΔS(m,t)＝max(S(m,N,r i ,t))-min(S(m,N,r j ,t)),i≠j；
according to the statistical principle, when m is more than or equal to 2 and less than or equal to 5 and sigma/2 is more than or equal to r and less than or equal to 2 sigma (sigma is the variance of the time series), the progressive distribution can be well approximated by a finite sequence, and the following 3 statistics are calculated:
in the formula (I), the compound is shown in the specification,
is the mean of the statistics S (m, N, r, t),
is the mean square error of the statistic S (m, N, r, t); optimum delay time correspondence
1 st zero or
1 st minimum value, S cor (t) the time corresponding to the minimum value is the optimum embedding window width m i 。
And the multivariable time sequence of the wind deflection angle of the insulator of the power transmission line consists of a plurality of univariate time sequences, and whether the multivariable time sequence has chaos characteristics or not needs to be determined before prediction by utilizing a chaos theory. The maximum Lyapunov (Lyapunov) index is one of the commonly used criteria to characterize whether a system is sensitive to initial values. Therefore, the maximum Lyapunov exponent can be calculated by adopting a small data method with strong anti-noise capability, and if the value of the maximum Lyapunov exponent is greater than 0, the multivariate time sequence has chaotic characteristics.
In one embodiment, the multivariate phase space is input into a preset wavelet neural network, and when the variable time sequence is predicted through the wavelet neural network to obtain an initial predicted value, the wavelet neural network prediction step length can be divided into N steps before and N-N steps after; predicting and correcting the first n steps of the variable time sequence by using a wavelet neural network to obtain a predicted time sequence and a predicted value of the first n steps; performing phase space reconstruction on the variable time sequence and the obtained first n-step prediction time sequence, and calculating a maximum Lyapunov exponent of the reconstructed phase space based on a small data amount algorithm; predicting the last N-N steps of the variable time sequence based on the maximum Lyapunov index to obtain a predicted value; and combining the predicted values of the previous N steps and the next N-N steps to obtain the initial predicted value of the variable time sequence.
Specifically, the wavelet neural network is formed by combining wavelet analysis multi-scale resolution, good time-frequency local characteristics of a wavelet function and a self-learning function of a traditional artificial neural network, and the result is more accurate by utilizing the strong noise and deformation data interference network generalization resistance of the wavelet neural network. The wavelet neural network structure with a compact structure adopted by the invention can be shown in fig. 2, and the wavelet neural network model prediction process is as follows:
(1) Setting a prediction step size: multi-variable phase space reconstruction phase transformation quantity based on power transmission line insulator windage yaw angle prediction
The time interval length of the time sequence X (T) and the time range to be predicted are set as the prediction step N, the time interval length is T, the total time interval length is T, and then the prediction step N = [ T/T ] after the mth time period of the initial prediction time sequence X (T) is set as the prediction step N]。
Since the chaos characteristic of the rear part of the multi-step improved prediction is more remarkable than that of the front part, the N step is divided into N steps (
And N is an integer), and the last N-N steps, and respectively performing distribution prediction on the first N steps and the last N-N steps.
(2) Predicting the first n steps:
step 1: the invention applies wavelet neural network theory, namely wavelet basis function (Mexican) as the transfer function of the hidden layer. The hidden layer output value is:
in the formula, h (j) is the output value of the j-th node in the hidden layer; omega ij The connection weight value between the ith node of the input layer and the jth node of the hidden layer is obtained;
is a wavelet basis function defined as
b j As a time-shift parameter of a function, c j Is a dimensional scaling parameter of the function.
The predicted value of the wavelet neural network output layer is as follows:
in the formula, ω jk The connection weight between the j node of the hidden layer and the k node of the output layer, and y (k) is the output value of the k node of the output layer。
Step 2: and (3) performing parameter correction on the output value by using a gradient correction method, and correcting the network weight and the wavelet basis function parameters by using self-adaptive adjustment eta proposed by Salomon. And (3) training the wavelet neural network model by using the obtained initial prediction time sequence X (t) to obtain the trained wavelet neural network model.
And 3, step 3: and predicting the effective result of the previous n steps by using the trained wavelet neural network model. Inputting the initial prediction time sequence X (t) into the trained wavelet neural network model to obtain a predicted value y 1 (1). Will y 1 (1) Combining with X (t) to obtain X (1，1) ＝[X(t)，y 1 (t)]Then, X is added (1，1) Inputting the predicted value y into a trained wavelet neural network model to obtain a predicted value y 1 (2) Successively calculating the predicted value according to the method, and iterating until the predicted value y of the nth step is obtained 1 (n), obtaining the prediction results of the previous n steps: y is 1 ＝{y(t)|t＝1,2,3,...,n}。
And 4, step 4: an m-dimensional phase space is reconstructed. Predicting the predicted value y of the previous n steps in the step 3 1 And combining with the initial prediction time sequence X (t) to obtain a prediction time sequence of the first n steps:
(3) Predicting in the last N-N steps:
step 1: to pair
And calculating the lag time tau and the embedding dimension m by using a C-C method, and then performing phase space reconstruction to obtain an m-dimensional phase space.
Step 2: calculating the maximum Lyapunov exponent of the phase space by using a small data quantity algorithm improved on the basis of a Walsh method and proposed by Rosensstein et al, and performing the next N-N steps of prediction on the m-dimensional phase space to obtain a prediction result of the next N-N steps:
1) Phase setting space midpoint d t ＝{X t ,X t+τ ,...,X t+(m-1)τ I is less than or equal to d and less than or equal to f (where d is the average period and f is a number)Number of data) of the nearest neighbor d i ＝{X i ,X i+τ ,...,X i+(m-1)τ }。
2) Calculating the maximum Lyapunov exponent:
where Δ t is the sample period, d t (i) And M is the iteration number in the evolution process, wherein the distance is the distance of the ith pair of nearest points passing through i discrete time steps.
3) Obtained by the maximum Lyapunov exponent method:
g(d t+1 ,d t )＝g(d i+1 ,d i )exp(λ)；
in the formula, g (d) t+1 ,d t ) Exp (λ) is the maximum Lyapunov exponent for the distance of the phase point between time t and time t + 1.
Thus, X t+(m-1)τ The predicted value of (A) is:
in the formula, y 2,t+(m-1)τ+1 The predicted value of the step 1 in the N-N steps after the wind deflection angle is obtained; x t+(m-1)τ+1 Is the phase point of the 1 st step in the N-N step phase space after the wind deflection angle.
4) Let y 2 (n+1)＝y 2,t+(m-1)τ+1 Then obtain the predicted value y 2 ＝{y 2 (t) | t = N +1, N +2,., N }. Will y 2 (n + 1) and the first n steps of prediction time sequence X (1) Combining to obtain X (1,n+1) ＝[X (1) ,y 2 (n+1)](ii) a In the formula, y 2 (N + 1) is a predicted value of the 1 st step in the N-N steps after the wind deflection angle; y is 2 Predicting value sets for N-N steps after the wind deflection angle; y is 2 (t) is a predicted value of the time t in the subsequent N-N steps; x (1,n+1) And predicting the time sequence after combination for the 1 st step in the previous N steps and the later N-N steps of the wind deflection angle.
5) To X (1,n+1) Performing phase space reconstruction to obtain new m-dimensional phase space, and application thereofPredicting the phase space by using the maximum Lyapunov exponent method to obtain a new prediction result y 2 (n + 2). The method iterates calculation one by one until y 2 Stopping after the value of (N), and obtaining the prediction result of the next N-N steps: y is 2 = { y (t) | t = N +1, N +2,.., N }; in the formula, y 2 The predicted value set of N-N steps after the wind deflection angle is obtained, t is the time of the N-N steps after the wind deflection angle, and y (t) is the predicted value of the t moment in the N-N steps after the wind deflection angle.
(4) The final result is:
step-by-step wavelet neural network is used for predicting value y of previous n steps 1 Predicting predicted value y with last N-N steps 2 And merging to obtain a prediction result:
Y＝{Y(t)＝[y 1 ,y 2 ]|t＝M+1,M+2,...,M+N}；
in the formula, Y is a final N-step prediction value set of the wavelet neural network, Y (t) is a wind deflection angle N-step prediction sequence, and Y 1 For the preceding n steps of prediction, y 2 And predicting the value of the next N-N steps.
In one embodiment, the insulator windage yaw has strong randomness and volatility, so that errors can occur even if the accuracy of a prediction model is high. The deterministic prediction method is not comprehensive enough in description of the oscillation rule of the insulator, so that on the basis of obtaining a predicted value of the wind drift angle, the fluctuation range of the predicted value is required to be obtained, and a prediction interval is constructed. The interval prediction of the wind deflection angle is to use a probability interval to express a difference value between an actual value and a predicted value of the wind deflection angle, namely a prediction error, and then combine a point predicted value of the obtained wind deflection angle to obtain a prediction interval.
Constructing an insulator wind drift angle prediction interval:
constructing a prediction interval of the wind drift angle based on the insulator wind drift angle prediction value and the probability distribution of the prediction error, setting the point prediction value as P, and the probability distribution function of the prediction error as F (), and then expressing the prediction interval under the confidence coefficient of 1-alpha as: [ P + F ] -1 (α 1 ),P+F -1 (α 2 )](ii) a Wherein alpha is 1 、α 2 Respectively an upper and a lower limit, alpha, in the probability distribution 1 ＝α/2，α 2 ＝1–α/2。
Non-parametric probability density estimation:
kernel Density Estimation (KDE) belongs to one of the non-parametric Estimation methods. Each data and bandwidth are used as parameters of a kernel function, a plurality of kernel functions are obtained, then a plurality of kernel functions are superposed to obtain a kernel density estimation function, and after normalization processing is carried out on the kernel density estimation function, a probability density function of the kernel density can be obtained and is used for determining probability distribution of prediction errors. The probability density function obtained by using KDE is as follows:
wherein N is the number of samples; h is the window width; k () is a kernel function; x is the number of i The sample value of the i-th prediction error. The kernel function selects an Epanechnikov function, and the formula is as follows:
wherein the content of the first and second substances,
is a sample probability density estimate.
Insulator windage yaw interval prediction: based on the obtained structure prediction interval, the specific steps are as follows: dividing the predicted point values of the wind deflection angle into a plurality of sections at equal intervals; obtaining the prediction error of each section, i.e. the difference e between the predicted value and the actual value i ＝y i -o i (ii) a Determining a probability density curve of a prediction error by adopting a kernel density estimation method; determining corresponding alpha according to the probability density function curve of each section 1 、α 2 Point values; calculating the alpha/2 and 1-alpha/2 quantile points of the corresponding curve, and combining the point predicted values at the corresponding time to construct a prediction interval.
FIG. 3 shows an embodiment of an insulator windage yaw prediction system of the present invention
In this optional embodiment, the insulator windage yaw prediction system includes:
the wind deflection angle factor determining module 301 is configured to determine influence factors of a wind deflection angle of an insulator of the power transmission line, and calculate and sort the influence factors to obtain a main influence factor variable;
the multivariate space construction module 303 is used for determining a variable time sequence of wind deflection angle change in the attack time of the pre-selected power transmission line insulator from the typhoon according to the main influence factor variable; constructing a multivariable phase space of the wind deflection angle of the insulator according to the variable time sequence;
the wind deflection angle prediction module 305 is configured to input the multivariate phase space to a preset wavelet neural network, and predict the variable time sequence through the wavelet neural network to obtain an initial prediction value;
and a windage yaw angle determining module 307, configured to perform interval transformation on the initial predicted value based on a nonparametric probability estimation method to obtain a final predicted value, and use the final predicted value as a predicted value of the windage yaw angle of the insulator.
In one embodiment, the influencing factors include: wind speed, wind power angle, rainfall intensity and humidity. When the wind deflection angle factor determining module 301 calculates and sorts the influence factors to obtain the main influence factor variables, the influence factors are calculated and sorted by a fuzzy analytic hierarchy process to obtain the main influence factor variables.
Correspondingly, in one embodiment, the wind slip factor determining module 301 includes: the device comprises a judgment matrix determining submodule (not shown in the figure), a weight vector calculating submodule (not shown in the figure) and a factor sorting submodule (not shown in the figure), wherein the judgment matrix determining submodule is used for comparing every two influencing factors of the same layer according to expert experience to obtain a triangular fuzzy judgment matrix; the weight vector calculation sub-module is used for calculating the proportion of each influence factor of the triangular fuzzy judgment matrix according to a sum-average method to obtain a factor weight matrix, calculating the expected value of the factor weight matrix according to a preset adjusting coefficient, and normalizing the expected value to obtain a factor weight matrix weight vector; and the factor sorting submodule is used for calculating the global weight of each influence factor according to the factor weight matrix weight vector, sorting the obtained weight of each influence factor and selecting D factors with larger influence as main influence factor variables.
In one embodiment, the wind slip factor determining module 301 further includes: a matrix checking submodule (not shown in the figure) for checking the consistency of the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix before calculating the global weight of each influencing factor according to the factor weight matrix weight vector; and when the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix meets consistency check, the factor sorting submodule calculates the global weight of each influence factor.
In one embodiment, the multivariate spatial construction module 303 comprises: the system comprises a time and dimension calculation submodule (not shown in the figure) and a phase space construction submodule (not shown in the figure), wherein the parameter calculation submodule is used for calculating delay time and embedding dimension of each univariate time sequence in the variable time sequence based on a univariate time sequence phase space reconstruction method; and the phase space construction submodule is used for constructing a multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
In one embodiment, the multivariate spatial construction module 303 comprises: and the parameter optimization submodule (not shown in the figure) is used for carrying out parameter optimization on the delay time and the embedding dimension of the constructed multivariable phase space based on a chaos time sequence phase space reconstruction C-C method before constructing the multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
In one embodiment, the wind slip angle prediction module 305 includes: a step size dividing sub-module (not shown in the figure), a wavelet prediction sub-module (not shown in the figure), an index prediction sub-module (not shown in the figure) and a merging sub-module (not shown in the figure), wherein the step size dividing sub-module is used for dividing the wavelet neural network prediction step size into the front N steps and the back N-N steps; the wavelet prediction submodule is used for predicting and correcting the first n steps of the variable time sequence by using a wavelet neural network to obtain a predicted time sequence and a predicted value of the first n steps; the index prediction submodule is used for carrying out phase space reconstruction on the variable time sequence and the obtained first n-step prediction time sequence and calculating the maximum Lyapunov index of the reconstructed phase space based on a small data amount algorithm; predicting the last N-N steps of the variable time sequence based on the maximum Lyapunov index to obtain a predicted value; and the merging submodule is used for merging the predicted values of the previous N steps and the next N-N steps to obtain the initial predicted value of the variable time sequence.
In conclusion, by means of the technical scheme, the method establishes the multivariate phase space reconstruction and wavelet neural network combination model based on which the wind drift angle of the insulator of the power transmission line is predicted. Firstly, an fuzzy analytic hierarchy process is used for calculating and sequencing influence factors, main influence factor variables are screened out, then a chaos theory and a C-C method are used for reconstructing a multivariable phase space of an insulator windage yaw angle, then a wavelet neural network combined prediction method is used for predicting the insulator windage yaw angle of the power transmission line to obtain a final prediction result, and finally a prediction value is segmented based on nonparametric probability density estimation to improve the reliability of the prediction value and obtain a final prediction interval result.
The method has the advantages that subjective consciousness can be weakened by the aid of a fuzzy analytic hierarchy process, index weights are reasonable, influence factors obtained according to the index weights are scientific and reasonable, and development reality is met; in addition, compared with the traditional BP neural network model, the wavelet neural network has higher degree of freedom, elasticity and plasticity, thereby having more sensitive approximation capability and stronger fault-tolerant capability. Therefore, a more effective prediction model can be obtained by using the wavelet neural network; and the maximum Lyapunov exponent is adopted to predict the data in the later period, so that the larger deviation of prediction in the later period can be effectively reduced. Meanwhile, random fluctuation of the combined model in a certain range of prediction results can be effectively eliminated by using nonparametric probability density estimation, so that more accurate prediction results are obtained.
The present invention is not limited to the structures that have been described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the invention is limited only by the appended claims.
Claims (16)
1. An insulator windage yaw prediction method is characterized by comprising the following steps:
determining influence factors of the wind deflection angle of the insulator of the power transmission line, and calculating and sequencing the influence factors to obtain main influence factor variables;
determining a variable time sequence of wind deflection angle changes in the attack time of the pre-selected power transmission line insulator from the typhoon according to the main influence factor variables; constructing a multivariable phase space of the wind deflection angle of the insulator according to the variable time sequence;
inputting the multivariate phase space into a preset wavelet neural network, and predicting the variable time sequence through the wavelet neural network to obtain an initial predicted value;
and carrying out interval division on the initial predicted value based on a nonparametric probability estimation method to obtain a final predicted value, and taking the final predicted value as a predicted value of the wind deflection angle of the insulator.
2. The insulator windage yaw prediction method of claim 1, wherein the influencing factors comprise:
wind speed, wind power angle, rainfall intensity and humidity.
3. The insulator windage yaw prediction method according to claim 1, wherein the step of calculating and sorting the influence factors to obtain the main influence factor variables comprises:
and calculating and sequencing the influence factors by a fuzzy analytic hierarchy process to obtain main influence factor variables.
4. The insulator windage yaw angle prediction method according to claim 3, wherein the influence factors are calculated and sequenced by a fuzzy analytic hierarchy process, and obtaining the main influence factor variables comprises:
comparing every two influencing factors of the same layer according to expert experience to obtain a triangular fuzzy judgment matrix;
calculating the proportion of each influence factor of the triangular fuzzy judgment matrix according to a sum-average method to obtain a factor weight matrix, calculating an expected value of the factor weight matrix according to a preset adjusting coefficient, and normalizing the expected value to obtain a factor weight matrix weight vector;
and calculating the global weight of each influence factor according to the factor weight matrix weight vector, sequencing the obtained weights of each influence factor, and selecting D factors with larger influence as main influence factor variables.
5. The insulator windage yaw angle prediction method according to claim 4, wherein the step of calculating and sorting the influence factors by a fuzzy analytic hierarchy process to obtain the main influence factor variables further comprises:
before calculating the global weight of each influencing factor according to the weight vector of the factor weight matrix, checking the consistency of the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix;
and when the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix meets consistency check, calculating the global weight of each influencing factor.
6. The insulator windage yaw prediction method of claim 1, wherein constructing a multivariate phase space of insulator windage yaw based on a variable time series comprises:
calculating delay time and embedding dimension of each univariate time sequence in the variable time sequence based on a univariate time sequence phase space reconstruction method;
and constructing a multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
7. The insulator windage yaw prediction method of claim 6, wherein constructing a multivariate phase space of insulator windage yaw based on a variable time series further comprises:
before constructing a multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension, performing parameter optimization on the delay time and the embedding dimension of the constructed multivariable phase space based on a chaos time sequence phase space reconstruction C-C method.
8. The insulator windage yaw prediction method according to claim 1, wherein the step of inputting the multivariate phase space into a preset wavelet neural network, and the step of predicting the variable time series through the wavelet neural network to obtain an initial predicted value comprises the steps of:
dividing the wavelet neural network prediction step size into a front N step and a back N-N step;
predicting and correcting the first n steps of the variable time sequence by using a wavelet neural network to obtain a predicted time sequence and a predicted value of the first n steps;
performing phase space reconstruction on the variable time sequence and the obtained first n-step prediction time sequence, and calculating a maximum Lyapunov exponent of the reconstructed phase space based on a small data amount algorithm; predicting the last N-N steps of the variable time sequence based on the maximum Lyapunov index to obtain a predicted value;
and combining the predicted values of the previous N steps and the next N-N steps to obtain the initial predicted value of the variable time sequence.
9. An insulator windage yaw prediction system, comprising:
the wind deflection angle factor determining module is used for determining the influence factors of the wind deflection angle of the insulator of the power transmission line, and calculating and sequencing the influence factors to obtain main influence factor variables;
the multivariate space construction module is used for determining a variable time sequence of wind deflection angle change in the attack time of the pre-selected power transmission line insulator from the typhoon according to the main influence factor variable; constructing a multivariable phase space of the wind deflection angle of the insulator according to the variable time sequence;
the wind deflection angle prediction module is used for inputting the multivariate phase space into a preset wavelet neural network and predicting the variable time sequence through the wavelet neural network to obtain an initial prediction value;
and the wind deflection angle determining module is used for segmenting the initial predicted value based on a nonparametric probability estimation method to obtain a final predicted value, and taking the final predicted value as the predicted value of the wind deflection angle of the insulator.
10. The insulator yaw angle prediction system of claim 9, wherein the influencing factors comprise:
wind speed, wind power angle, rainfall intensity and humidity.
11. The insulator wind deflection angle prediction system according to claim 9, wherein the wind deflection angle factor determination module calculates and sorts the influence factors to obtain the main influence factor variables by an fuzzy analytic hierarchy process when calculating and sorting the influence factors to obtain the main influence factor variables.
12. The insulator yaw angle prediction system of claim 11, wherein the yaw angle factor determination module comprises: a judgment matrix determining submodule, a weight vector calculating submodule and a factor sorting submodule, wherein,
the judgment matrix determining submodule is used for comparing every two influencing factors of the same layer according to expert experience to obtain a triangular fuzzy judgment matrix;
the weight vector calculation sub-module is used for calculating the proportion of each influence factor of the triangular fuzzy judgment matrix according to a sum-average method to obtain a factor weight matrix, calculating the expected value of the factor weight matrix according to a preset adjusting coefficient, and normalizing the expected value to obtain a factor weight matrix weight vector;
and the factor sorting submodule is used for calculating the global weight of each influence factor according to the factor weight matrix weight vector, sorting the obtained weight of each influence factor and selecting D factors with larger influence as main influence factor variables.
13. The insulator yaw angle prediction system of claim 12, wherein the yaw angle factor determination module further comprises: the matrix checking submodule is used for checking the consistency of the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix before calculating the global weight of each influencing factor according to the factor weight matrix weight vector; and when the triangular fuzzy judgment matrix of each factor layer of the factor weight matrix meets consistency check, the factor sorting submodule calculates the global weight of each influence factor.
14. The insulator windage yaw prediction system of claim 9, wherein the multivariate spatial construction module comprises: a time and dimension calculation sub-module, a phase space construction sub-module, wherein,
the parameter calculation submodule is used for calculating the delay time and the embedding dimension of each univariate time sequence in the variable time sequence based on a univariate time sequence phase space reconstruction method;
and the phase space construction submodule is used for constructing a multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
15. The insulator yaw angle prediction system of claim 14, wherein the multivariate spatial construction module comprises: and the parameter optimization submodule is used for carrying out parameter optimization on the delay time and the embedding dimension of the constructed multivariable phase space based on a chaotic time sequence phase space reconstruction C-C method before constructing the multivariable phase space corresponding to the variable time sequence according to the delay time and the embedding dimension.
16. The insulator drift angle prediction system of claim 9, wherein the drift angle prediction module comprises: a step size division sub-module, a wavelet prediction sub-module, an exponential prediction sub-module, and a merging sub-module, wherein,
the step size dividing submodule is used for dividing the wavelet neural network prediction step size into the front N steps and the back N-N steps;
the wavelet prediction submodule is used for predicting and correcting the first n steps of the variable time sequence by using a wavelet neural network to obtain a predicted time sequence and a predicted value of the first n steps;
the index prediction submodule is used for carrying out phase space reconstruction on the variable time sequence and the obtained first n-step prediction time sequence and calculating the maximum Lyapunov index of the reconstructed phase space based on a small data amount algorithm; predicting the last N-N steps of the variable time sequence based on the maximum Lyapunov index to obtain a predicted value;
and the merging submodule is used for merging the predicted values of the previous N steps and the next N-N steps to obtain the initial predicted value of the variable time sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211375768.8A CN115660194A (en) | 2022-11-04 | 2022-11-04 | Insulator wind deflection angle prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211375768.8A CN115660194A (en) | 2022-11-04 | 2022-11-04 | Insulator wind deflection angle prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115660194A true CN115660194A (en) | 2023-01-31 |
Family
ID=84995843
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211375768.8A Pending CN115660194A (en) | 2022-11-04 | 2022-11-04 | Insulator wind deflection angle prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115660194A (en) |
-
2022
- 2022-11-04 CN CN202211375768.8A patent/CN115660194A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110149237B (en) | Hadoop platform computing node load prediction method | |
| CN110232203B (en) | Knowledge distillation optimization RNN short-term power failure prediction method, storage medium and equipment | |
| CN112598248B (en) | Load prediction method, load prediction device, computer equipment and storage medium | |
| CN114565124A (en) | Ship traffic flow prediction method based on improved graph convolution neural network | |
| CN110413657B (en) | Average response time evaluation method for seasonal non-stationary concurrency | |
| CN112363896A (en) | Log anomaly detection system | |
| CN111985845B (en) | Node priority optimization method of heterogeneous Spark cluster | |
| CN107704426A (en) | Water level prediction method based on extension wavelet-neural network model | |
| CN110633859A (en) | Hydrological sequence prediction method for two-stage decomposition integration | |
| CN116187835A (en) | Data-driven-based method and system for estimating theoretical line loss interval of transformer area | |
| CN115766125A (en) | Network flow prediction method based on LSTM and generation countermeasure network | |
| CN110689140A (en) | Method for intelligently managing rail transit alarm data through big data | |
| CN114357670A (en) | Power distribution network power consumption data abnormity early warning method based on BLS and self-encoder | |
| CN112861418A (en) | Short-term icing thickness prediction method for stay cable based on GA-WOA-GRNN network | |
| CN115660194A (en) | Insulator wind deflection angle prediction method and system | |
| CN114463978B (en) | Data monitoring method based on track traffic information processing terminal | |
| CN116522747A (en) | Two-stage optimized extrusion casting process parameter optimization design method | |
| CN116089801A (en) | Medical data missing value repairing method based on multiple confidence degrees | |
| CN115879369A (en) | Coal mill fault early warning method based on optimized LightGBM algorithm | |
| CN114943328A (en) | SARIMA-GRU time sequence prediction model based on BP neural network nonlinear combination | |
| CN114784795A (en) | Wind power prediction method and device, electronic equipment and storage medium | |
| CN112667394A (en) | Computer resource utilization rate optimization method | |
| CN114647679A (en) | Hydrological time series motif mining method based on numerical characteristic clustering | |
| CN114118508A (en) | OD market aviation passenger flow prediction method based on space-time convolution network | |
| CN112330003B (en) | Periodic capacity data prediction method, system and storage medium based on bidirectional cyclic neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |