CN110309603B - Short-term wind speed prediction method and system based on wind speed characteristics - Google Patents
Short-term wind speed prediction method and system based on wind speed characteristics Download PDFInfo
- Publication number
- CN110309603B CN110309603B CN201910601512.6A CN201910601512A CN110309603B CN 110309603 B CN110309603 B CN 110309603B CN 201910601512 A CN201910601512 A CN 201910601512A CN 110309603 B CN110309603 B CN 110309603B
- Authority
- CN
- China
- Prior art keywords
- wind speed
- prediction
- component
- remainder
- model
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention discloses a short-term wind speed prediction method and a short-term wind speed prediction system based on wind speed characteristics. The prediction method comprises the following steps: acquiring an original wind speed sequence; singular value decomposition is carried out on the wind speed sequence, and a denoising sequence and a noise remainder are obtained; acquiring an optimal mode number for performing variation mode decomposition on the denoising sequence; performing variation modal decomposition on the denoising sequence to obtain each component sequence; obtaining information related to high output wind speed in the environmental factors by using an improved average influence value method, and establishing a long-term and short-term memory network model by combining the component sequences or the remainder; optimizing a network model through an improved extremum optimization method to obtain a prediction model; predicting each component sequence and the noise sequence through the prediction model to obtain each component and remainder prediction results; accumulating the predicted results of each component and the remainder to obtain a wind speed predicted result; the wind speed prediction method or system provided by the invention obviously improves the reliability of the prediction model, and can obtain a high-precision prediction result.
Description
Technical Field
The invention relates to the field of wind speed prediction, in particular to a short-term wind speed prediction method and system based on wind speed characteristics.
Background
In recent years, global energy situation is becoming severe, and energy demands are increasing. The global energy demand in 2018 has increased by 2.3% as reported by the international energy organization (IEA), which is the fastest year for the last decade. However, as non-renewable resources continue to be consumed, environmental problems continue to be exacerbated, and energy systems should be shifted to more sustainable directions in order to maintain a balance of supply and demand, improve the global environment. The core of energy system transformation is to develop renewable energy sources, and wind energy is one of renewable energy sources, so that the development speed is also accelerated.
At the present stage, the global wind energy resources are rich, the wind power generation cost is greatly reduced, and the wind power generation is rapidly developed. The Global Wind Energy College (GWEC) report on global wind power development of 2018 indicates that the new installed capacity of the global wind energy industry in 2018 is 51.3GW and the total installed capacity is 519GW, wherein the installed capacity of china is the first place in the world. The difficulty of real-time dispatching of the power grid is aggravated by large-scale wind power access to the power system, and challenges are brought to the safety and stability of the power grid. Wind power prediction is one of the main approaches to solve the problems, so that wind power prediction research is very important in countries around the world. And because the accurate wind speed prediction is favorable for solving the problems of wind power output power control, power grid safety economic dispatch and the like, the wind speed prediction has very important research significance and value.
Disclosure of Invention
The invention aims to provide a short-term wind speed prediction method and a short-term wind speed prediction system based on wind speed characteristics, which are used for obtaining a high-precision prediction result and improving the reliability of a prediction model.
In order to achieve the above object, the present invention provides the following solutions:
a short-term wind speed prediction method based on wind speed characteristics, the method comprising:
acquiring an original wind speed sequence; singular value decomposition is carried out on the wind speed sequence, and a denoising sequence and a noise remainder are obtained;
according to a mode number optimization method, obtaining the optimal mode number for carrying out variation mode decomposition on the denoising sequence;
decomposing the denoising sequence according to a variation modal decomposition method of the optimal modal number to obtain a plurality of component sequences;
acquiring other environmental factors influencing wind speed prediction, acquiring information with high correlation with output wind speed in the environmental factors by using an improved average influence value method, and establishing a long-term and short-term memory network model by combining the component sequences or remainder;
optimizing a network model through an improved extremum optimization algorithm, and obtaining an optimized prediction model;
predicting a plurality of component sequences through the prediction model to obtain a component prediction result;
Predicting the noise remainder through the prediction model to obtain a remainder prediction result;
and accumulating the component and remainder predicted results to obtain a wind speed predicted result.
Optionally, the acquiring the denoising sequence and the noise remainder specifically includes:
acquiring original wind speed data;
constructing a Hankel matrix for the wind speed sequence, acquiring singular values of the Hankel matrix by a singular value decomposition method, and sequencing the singular values from large to small; obtaining a differential spectrum of the singular values;
and determining the number of effective singular values according to the maximum mutation points in the singular value differential spectrum, and reconstructing the wind speed signal to obtain a denoising signal and a noise remainder.
Optionally, the obtaining the optimal mode number for performing variation mode decomposition on the denoising sequence includes:
constructing a constraint variation model under the current mode number for the denoising sequence;
introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into an unconstrained problem;
the adaptive decomposition of the signals is realized through Fourier equidistant transformation and other processes; iterative updating, judging whether the Fourier transform of the modal component meets the convergence condition; if yes, obtaining a decomposition result under the mode number condition; if not, adjusting the center frequency and the bandwidth of the modal component until the Fourier transform of the modal component meets the convergence condition, and acquiring a decomposition result of the signal;
Calculating the energy sum of each component under the current mode number;
and (3) circulating the steps, and calculating the energy difference under the condition of the adjacent mode numbers, wherein the mode number corresponding to the obviously increased energy difference is the optimal mode number of the acquired variation mode decomposition.
Optionally, performing a variation modal decomposition of the denoising sequence with a determined modal number to obtain a plurality of component sequences, which specifically includes:
constructing a constraint variation model under the condition of the optimal modal number for the denoising sequence;
introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into an unconstrained problem;
the adaptive decomposition of the signals is realized through Fourier equidistant transformation and other processes; iterative updating, judging whether the Fourier transform of the modal component meets a convergence condition; if yes, obtaining a decomposition result of the signal; if not, the center frequency and the bandwidth of the modal component are adjusted until convergence conditions are met, and a plurality of component sequences of the signal are obtained.
Optionally, the obtaining other environmental factors affecting wind speed prediction, obtaining information with high correlation with output wind speed in the environmental factors by using an improved average influence value method, and building a long-term and short-term memory network model by combining the component sequences or the remainder, including:
Acquiring other environmental factors such as wind direction, temperature and the like which influence wind speed prediction, and establishing a radial basis function neural network model for the environmental factors;
adding and subtracting 10% from each environmental factor to obtain a new sample; training the radial basis function neural network on the sample, and acquiring an average influence value between the factors and the model output variable;
according to the average influence value, obtaining information with high correlation with the output wind speed in the environmental factors; normalizing the selected information to obtain a dimensionless data set;
acquiring a correlation coefficient matrix and an accumulated variance contribution rate of the data set, and selecting all main components with the accumulated variance contribution rate more than 85%;
and the main component combines the component sequences or the remainder to establish a long-term and short-term memory network model.
Optionally, the obtaining information with high correlation with the output wind speed in the environmental factors by using an improved average influence value method, and predicting with the component sequence or the remainder sequence specifically includes:
optimizing a network model through an improved extremum optimization algorithm, and obtaining an optimized prediction model;
and determining a long-term and short-term memory network model through the optimal initialization parameters to obtain an optimized long-term and short-term memory network prediction model.
Optionally, predicting a plurality of the component sequences through the prediction model to obtain a component prediction result; predicting the noise remainder through the prediction model to obtain a remainder prediction result; accumulating the component and remainder prediction results to obtain a wind speed prediction result, which specifically comprises:
predicting each modal component through the prediction model to obtain a plurality of component prediction results;
predicting the remainder sequence through the prediction model to obtain a remainder prediction result;
and linearly superposing a plurality of component prediction results and the remainder prediction results to obtain a wind speed prediction result.
The invention also provides a short-term wind speed prediction system based on wind speed characteristics, which comprises:
the wind speed data acquisition and separation module is used for acquiring an original wind speed sequence; singular value decomposition is carried out on the wind speed sequence, and a denoising sequence and a noise remainder are obtained;
the optimal mode number acquisition module is used for acquiring the optimal mode number for performing variation mode decomposition on the denoising sequence according to a mode number optimization method;
the variation mode decomposition module is used for decomposing the denoising sequence according to a variation mode decomposition method of the optimal mode number to obtain a plurality of component sequences;
The environment information screening module is used for acquiring other environment factors influencing wind speed prediction, acquiring information with high relativity with output wind speed in the environment factors by utilizing an improved average influence value method, and establishing a long-term and short-term memory network model by combining the component sequences or the remainder;
the parameter optimization module is used for optimizing parameters of the network model by the improved extremum optimization algorithm to obtain an optimized prediction model;
the prediction module is used for predicting a plurality of component sequences through the prediction model to obtain a component prediction result; predicting the noise remainder through the prediction model to obtain a remainder prediction result; and accumulating each component predicted result and each remainder predicted result to obtain a wind speed predicted result.
Optionally, the wind speed data acquisition and separation module specifically includes:
the wind speed data acquisition unit is used for acquiring original wind speed data;
the singular value differential spectrum acquisition unit is used for constructing a Hankel matrix for the wind speed sequence, acquiring singular values of the Hankel matrix through a singular value decomposition method, and sequencing the singular values from large to small; obtaining a differential spectrum of the singular values;
And the signal reconstruction unit is used for determining the number of the effective singular values according to the maximum mutation points in the singular value differential spectrum, reconstructing the wind speed signal and obtaining a denoising signal and a noise residual.
Optionally, the optimal modality number acquisition module specifically includes:
the constraint variation model acquisition unit is used for constructing a constraint variation model under the current mode number for the denoising sequence;
the problem conversion unit is used for introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into a non-constraint problem;
the modal component acquisition unit is used for realizing the self-adaptive decomposition of the signals through Fourier equidistant transformation and other processes; iterative updating, judging whether the Fourier transform of the modal component meets the convergence condition; if yes, obtaining a decomposition result under the mode number condition; if not, the center frequency and the bandwidth of the modal component are adjusted until the Fourier transform of the modal component meets the convergence condition, and the decomposition result of the signal is obtained.
The energy acquisition unit is used for calculating the energy sum of each component under the current mode number;
and the energy difference acquisition unit is used for circulating the steps, calculating the energy difference under the condition of adjacent modal numbers, and the corresponding modal number when the energy difference is obviously increased is the acquired optimal modal number of the variation modal decomposition.
Optionally, the variation modal decomposition module specifically includes:
the constraint variation model construction unit is used for constructing a constraint variation model under the condition of the optimal modal number for the denoising sequence;
the problem conversion unit is used for introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into a non-constraint problem;
the modal component acquisition unit is used for realizing the self-adaptive decomposition of the signals through Fourier equidistant transformation and other processes; iterative updating, judging whether the Fourier transform of the modal component meets a convergence condition; if yes, obtaining a decomposition result of the signal; if not, the center frequency and the bandwidth of the modal component are adjusted until convergence conditions are met, and a plurality of component sequences of the signal are obtained.
Optionally, the environmental factor screening module specifically includes:
the environment factor acquisition unit is used for acquiring other environment factors such as wind direction, temperature and the like which influence wind speed prediction, and establishing a radial basis function neural network model for the environment factors;
the average influence value obtaining unit is used for adding and subtracting 10% from each environmental factor to obtain a new sample; training the radial basis function neural network on the sample, and acquiring an average influence value between the factors and the model output variable;
The normalization unit is used for acquiring information with high correlation with the output wind speed in the environmental factors according to the average influence value; carrying out normalization processing on the useful information to obtain a dimensionless data set;
a principal component obtaining unit, configured to obtain a correlation coefficient matrix and an accumulated variance contribution rate of the dataset, and select all principal components with accumulated variance contribution rates greater than 85%;
and the network model building unit is used for building a long-term and short-term memory network model by combining the main components with the component sequences or the remainder.
Optionally, the parameter optimization module specifically includes:
the basic prediction model acquisition unit is used for acquiring a long-period and short-period memory network model;
the parameter optimization unit is used for optimizing the long-period memory network model through an improved extremum optimization algorithm to obtain optimal initialization model parameters;
and the wind speed prediction model determining unit is used for determining a long-period memory network model through the optimal initialization parameters to obtain an optimized long-period memory network wind speed prediction model.
Optionally, the prediction module specifically includes:
the component prediction unit is used for predicting each modal component through the prediction model to obtain a plurality of component prediction results;
The remainder prediction unit is used for predicting the remainder sequence through the prediction model to obtain a remainder prediction result;
and the superposition unit is used for linearly superposing the plurality of component prediction results and the remainder prediction results to obtain a wind speed prediction result.
Compared with the prior art, the invention has the following technical effects:
the wind speed prediction method and the wind speed prediction system are short-term wind speed prediction processes based on wind speed characteristics. Firstly, a new mixed mode decomposition method is adopted to decompose a non-stationary time sequence into a plurality of relatively stationary mode components and a remainder sequence, noise information is separated from a non-stationary original wind speed sequence, and a variation mode decomposition method with optimized mode numbers is utilized to decompose a denoising sequence to obtain a plurality of components with obvious periodic characteristics. Meanwhile, environmental factors influencing wind speed are introduced to participate in prediction, and an improved average influence value method is provided for screening variables participating in model input, so that the complexity of the model is reduced, and meanwhile, the prediction precision is improved. Finally, the invention selects the long-period memory network capable of memorizing the effective information for a long time as a prediction model, and utilizes the improved extremum optimization algorithm to optimize the parameters of the long-period memory network, thereby accelerating the model training speed and improving the prediction precision. And accumulating the predicted result by using the model predicted component sequence and the remainder sequence to obtain the final predicted wind speed. According to the method, the fluctuation and uncertainty of the wind speed are considered, so that the short-term wind speed prediction accuracy of the long-term and short-term memory network model is effectively improved, and meanwhile, the reliability of the prediction model is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a short-term wind speed prediction method based on wind speed characteristics according to embodiment 1 of the present invention;
FIG. 2 is a flowchart of a prediction method according to embodiment 2 of the present invention;
FIG. 3 is a block diagram of a prediction system according to embodiment 3 of the present invention;
FIG. 4 is a graph of the original wind speed data distribution of the Sotavento wind farm of the present invention;
FIG. 5 is a graph showing the original wind speed data distribution of the Changma wind farm of the present invention;
FIG. 6 is a schematic diagram of the result of the denoising sequence decomposition of the Sotavento wind farm according to the present invention;
FIG. 7 is a schematic diagram of the denoising sequence decomposition result of the Changma wind farm according to the present invention;
FIG. 8 is a comparison box diagram of prediction errors before and after optimization of parameters of a Sotavento wind farm long-short term memory network model;
FIG. 9 is a graph of the comparison of prediction errors before and after optimization of parameters of the Chang Ma Fengdian field long-term memory network model according to the present invention;
FIG. 10 is a graph of three model predictions for a Sotavento wind farm according to the present invention;
FIG. 11 is a graph showing the predicted results of three models after the field Chang Ma Fengdian of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
FIG. 1 is a flow chart of a wind speed prediction method according to the present invention. As shown in fig. 1, the short-term wind speed prediction method based on wind speed characteristics includes the steps of:
step 11: and obtaining an original wind speed sequence, performing singular value decomposition on the wind speed sequence, and obtaining a denoising sequence and a noise remainder.
At present, singular value decomposition is mainly used for noise reduction and periodic component extraction of vibration signals. According to the invention, the singular value decomposition differential spectrum method is utilized to extract noise information in a wind speed sequence, so as to obtain a noise reduction signal and a noise remainder. The method comprises the following specific steps:
Step 111: and acquiring original wind speed data, and constructing a Hankel matrix for the wind speed sequence.
Step 112: singular values of the Hankel matrix are obtained through a singular value decomposition method, and the singular values are ordered from large to small.
Step 113: and obtaining a differential spectrum of the singular values.
Step 114: and determining the number of effective singular values according to the maximum mutation points in the singular value differential spectrum, and reconstructing the wind speed signal to obtain a denoising signal and a noise remainder.
Step 12: and obtaining the optimal mode number for carrying out variation mode decomposition on the denoising sequence according to the mode number optimization method.
In 2014, the Variational Modal Decomposition (VMD) method proposed by Dragomiltskiy et al overcomes a series of disadvantages of Empirical Mode Decomposition (EMD) modal aliasing, end-point effect and the like, and has higher operation efficiency and good noise robustness. The method seeks the optimal solution of the variation model in a completely non-recursive mode, and realizes the effective separation of the frequency domain decomposition of the signal and each component according to the central frequency and the bandwidth of each decomposition component. It is found that the number of modes has a great influence on the analysis result, and that too much or insufficient modes can influence the accuracy of the analysis result. In order to make up for the defect of subjective selection of influence parameters, the invention provides a mode number optimization method for overcoming the defect caused by the random selection of the mode number. The components obtained by VMD decomposition are in an orthogonal relationship, so that the sum of the energy of the components is equal to the energy of the original signal if the decomposition is proper. When the number of modes exceeds a certain proper value, imaginary components are generated, and the sum of component energy linearities is larger than the energy sum of the original decomposed components. The method comprises the following specific steps:
Step 121: the denoising sequence constructs a constraint variation model under the current mode number: first, for each modal componentAnd->Respectively represent u k Instantaneous amplitude and instantaneous phase of (t), +.>Non-subtraction function, t is sampling time), and obtaining an analytic signal thereof through Hilbert transformation; then, predicting the center frequency of each analytic signal, and converting the frequency spectrum of each analytic signal to a base frequency band; and finally, estimating the bandwidth of each modal component by using Gaussian smoothing indexes of the frequency shift signals, and constructing a constraint variation model shown in the formula (1).
In the method, in the process of the invention,for the component signal u k Center frequency of (t),>representing the partial derivative of the function t, j 2 = -1, x (t) is the de-noising sequence.
Step 122: the constraint problem is changed into the unconstrained problem by introducing a quadratic penalty factor alpha and Lagrange multiplication operator lambda (t), and the extended Lagrange expression is as follows:
step 123: and obtaining a formula (3) through Fourier equidistant transformation and other processes, and realizing self-adaptive decomposition of signals.
Wherein, the liquid crystal display device comprises a liquid crystal display device,are respectively->Is a fourier transform of (a). The center frequency is updated by the following formula:
step 124: and (5) iteratively updating until convergence meets the following conditions:
step 125: and calculating the energy sum of each component under the current mode number.
Step 126: and circulating the steps, calculating the energy difference under the condition of adjacent modal numbers, and obtaining the optimal modal number of the fractional-change modal decomposition. The calculation formula of the energy of the denoising signal and the energy difference is as follows
Wherein E is k In the case of the number of modes being k, all component energies sum, E k-1 Is the sum of the component energy values after the last decomposition, eta k，k-1 Representing the energy difference, x j (i) And n is the number of sampling points for the j-th component sequence under the current mode number.
As can be seen from (6), η k，k-1 The smaller the value of (2), the signal may be under-decomposed; η (eta) k，k-1 The larger the value of (2), the more pronounced the decomposition phenomenon of the VMD. Over-decomposition occurs with increasing parameter k, η k，k-1 There will be a significant increase where k-1 of the turning point can be used as the optimal number of modes for VMD decomposition.
Step 13: carrying out variation modal decomposition of modal number determination on the denoising sequence to obtain a plurality of component sequences, wherein the method specifically comprises the following steps:
step 131: constructing a constraint variation model under the condition of the optimal modal number for the denoising sequence;
step 132: introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into an unconstrained problem;
step 133: the adaptive decomposition of the signals is realized through Fourier equidistant transformation and other processes;
Step 134: iterative updating, judging whether the Fourier transform of the modal component meets a convergence condition; if yes, obtaining a decomposition result of the signal; if not, the center frequency and the bandwidth of the modal component are adjusted until convergence conditions are met, and a plurality of component sequences of the signal are obtained.
Step 14: and obtaining information with high correlation with the output wind speed in the environmental factors by using an improved average influence value method, and predicting the information with the component sequence or the remainder sequence.
And acquiring other environmental factors such as wind direction, temperature and the like which influence wind speed prediction, and selecting information with high correlation with the output wind speed from the environmental factors by utilizing an improved average influence value method. The existing average influence value (MIV) method mainly adopts a Back Propagation (BP) neural network to calculate the correlation between the independent variable and the output variable, but the BP neural network has higher randomness and unstable result, so that the Radial Basis Function (RBF) neural network is adopted to replace the BP neural network. The method reduces the input dimension of the prediction model, improves the prediction performance, and is named as an improved average influence value (IMIV) method. The method comprises the following specific steps:
step 141: for the environmental factor raw data L (L 1 ，l 2 ，…，l p ) Establishing a radial basis function neural network model, wherein p is the number of introduced factors;
step 142: for each factor of the original data i Adding or subtracting 10% to obtain two new samples
Step 143: for newTraining the radial basis function neural network to obtain training results y respectively i ，z i And calculate y i ，z i The difference of (a) representing the change value of the influence of the variation of the characteristic on the output result is denoted as a i (see formula 8) calculating the mean of the examples to obtain MIV value, denoted as b i ；
Step 144: repeating steps 141-143, calculating MIV values of all independent variables, and selecting factors with relatively large absolute values to form a data set Y= { Y ij I=1, 2, …, n.j =1, 2, …, m, m is the data size, n is the number of variables selected to participate in the analysis.
Step 145: normalization of Y is performed, the formula is shown in (9), and the result is X= { X ij I=1, 2, …, n.j =1, 2, … m, y i Is the ith feature dataset.
Step 146: and calculating a correlation coefficient matrix R of X and accumulating variance contribution rate eta, and selecting the number p of main components with eta more than or equal to 85 percent.
Where r is the data size, λ i Is the i-th eigenvalue of the correlation coefficient matrix R.
Step 15: parameters of a long-short-term memory network model (LSTM) are optimized through an improved extremum optimization algorithm.
With the continuous development of deep learning technology, the concept of time series is applied in the structural design of Recurrent Neural Networks (RNNs). Thus, RNNs have good time series analysis capabilities. The LSTM is used as an improved RNN, inherits the analysis capability of the RNN on time sequence data, and overcomes the defect of the RNN in long-term memory. Because the LSTM neural network model can effectively keep long-time memory, certain results are achieved in the field of wind speed prediction. The memory unit of an LSTM neural network is internally provided with 3 gates, namely an input gate, a forget gate and an output gate. The LSTM update unit calculation process is as follows:
first, the t-th time inputThe information of (1) passes through the input gate, and the value of the input layer of the memory unit is i t Candidate value of hidden layer stateThen, the information passes through the forgetting gate, and the value of the forgetting layer of the memory unit is f t : at this time, the hidden layer status update value is C t : finally, the information is updated to o through the output gate, the value of the output layer and the output value of the final memory unit t ，h t The calculation formula is shown in formula (11):
wherein x is t Is the input vector, h t Is the output vector, α represents the gate activation function (typically a logical sigmoid function), x is the element multiplication (Hadamard product) between the two vectors, W represents the corresponding weight matrix, and b is the associated bias vector.
Therefore, a long-short-term memory network model can be acquired. And then optimizing the long-term and short-term memory network model through an improved extremum optimization algorithm to obtain optimal initialization model parameters. And determining a long-term and short-term memory network model through the optimal initialization parameters to obtain an optimized long-term and short-term memory network prediction model.
The idea of extremum optimizing algorithm (EO) is derived from self-organizing critical theory, and the outstanding characteristic is unbalance. Unlike genetic algorithm, simulated annealing algorithm, ant colony algorithm, particle swarm optimization algorithm, etc., the algorithm will not converge to a balanced state, and the generated fluctuation will make the algorithm have better ability to search continuously and jump out the optimal solution. The algorithm is easy to realize, small in calculated amount, good in algorithm effect and applicable to a plurality of projects. Because of the long training time caused by the free generation of LSTM initial parameters, the invention combines the EO optimization idea to optimize LSTM, quickens the training speed, improves the prediction performance of LSTM, and names the optimization algorithm as an improved mechanism optimization algorithm (IEO). The method comprises the following specific steps:
step 151: aiming at parameters required to be trained by LSTM, defining a proper population size, wherein the corresponding fitness function is f;
In the method, in the process of the invention,x represents a wind speed predicted value and an actual value respectively, and n is the number of samples.
Step 152: initializing parameters, setting the maximum iteration number as T max =50, randomly generating an initial solution pop;
step 153: calculating a fitness value f of the population, and sequencing species pop according to the fitness value;
step 154: selecting species x with the smallest fitness value m And x m-1 Generating random number rand according to a certain probability distribution x1 And rand x2 To replace the original x m And x m-1 ；
Step 155: recalculating rand x1 And rand x2 The fitness value f of (a);
if f (rand) x1 ) > max (f), then x max ＝rand x1 ，max(f)＝f(rand x1 ) Updating the population pop;
if f (rand) x2 ) > max (f), then x max ＝rand x2 ，max(f)＝f(rand x2 ) Updating the population pop;
step 156: if the stopping criterion is met, the algorithm stops, the result is output, otherwise the loop calculation is returned to step 153.
Step 16: and predicting each modal component through the prediction model to obtain a plurality of component prediction results. And predicting the remainder sequence through the prediction model to obtain a remainder prediction result. And linearly superposing the plurality of component prediction results and the remainder prediction result to obtain a wind speed prediction result.
In order to verify that the method has good prediction performance on wind speed data of an actual wind power plant, a data simulation experiment is performed by adopting wind speed of a Chinese wind power plant, and fig. 2 is a flow chart of the prediction method provided by the embodiment. As shown in fig. 2, the specific process includes:
Step 21: acquiring an original wind speed sequence
In order to better verify the prediction capability of the method provided by the invention, two data sets are selected to respectively predict the short-term wind speed in the embodiment. The two data sets are respectively from spanish Sotavento and China Chang Ma Fengdian, and in 2013, 250 sample data are respectively selected, wherein the first 200 data are used as training sets, and the last 50 data are used as test sets. The two data sets herein select different influencing factors and different sampling times, thereby verifying the broad applicability of the model herein. The wind speed of the Sotavento wind power plant is sampled once every 10 minutes, and 7 factors including wind speed, wind direction, air temperature, air pressure, specific volume, specific humidity and surface roughness are selected to participate in prediction; the wind speed collection interval of the China Chang horse wind power plant is 5 minutes, and environmental factors comprise wind direction, air temperature, motor rotating speed, pitch angle and daily power generation quantity accumulation. The wind speeds are studied from the point of view of random fluctuation, and the change characteristics of the wind speeds of the two wind power stations are respectively shown in fig. 4 and 5. As can be seen from the figures: the wind speed of the Sotavento wind power plant mostly fluctuates in the interval [6m/s,16m/s ], no obvious periodic characteristics exist, and strong random fluctuation exists; the wind speed of the Chinese Chang horse wind power plant fluctuates at [6m/s,14m/s ], no obvious periodic characteristics exist, and the random fluctuation is strong, so that the research on the fluctuation characteristic of the wind speed is very important.
Step 22: mixed modal decomposition process
Step 221: in order to better extract the periodic characteristics in the original wind speed sequence, SVD noise reduction processing is carried out on the original wind speed, and a denoising sequence and a noise remainder are obtained.
Step 222: when the energy and the energy difference of the denoising sequence are calculated by using the mode number selection method, k=3-9, the result of the energy difference is shown in table 1, and when the mode number is changed from 7 to 8, the energy of the two wind power plant data sets is changed greatly, and the energy difference is increased. Thus, in both datasets, we choose a VMD with a mode number of 7.
TABLE 1 energy differences between different modality numbers
Step 223: VMD decomposition was performed on the denoised sequences in the Sotavento and chang Ma Fengdian fields to obtain 7 modal components, respectively, as shown in fig. 6 and 7. From fig. 6 and 7, it can be seen that the periodic characteristics of the components are more apparent when compared to the original wind speed sequence. At this time, the mixed mode decomposition process ends.
Step 224: to improve the prediction accuracy, we use the IMIV method to handle multiple factors. First, a correlation analysis is performed on multiple factors, and the correlation between each factor and the output, that is, the MIV value is calculated, and the results are shown in tables 2 and 3. The method comprises the steps of respectively filtering environmental factors with smaller influence on wind speed prediction results of two wind power plants, namely eliminating 5 th and 6 th factors specific volume and specific humidity with relatively smaller absolute value of a Sotavento wind power plant MIV, and eliminating 5 th factors of a Chang horse wind power plant: pitch angle.
TABLE 2 MIV values between factors and output sequences of Sotavento wind farms
TABLE 3 MIV values between factors and output sequences for Changma wind farm
Step 225: and the principal component analysis is carried out on other factors, the dimension of the model is reduced, redundant information is removed, and the calculation efficiency and the prediction precision are improved. All main components with the accumulated variance contribution rate reaching 85% are selected by the two wind farms respectively and used as fixed inputs of a prediction model. The principal component analysis results are shown in Table 4, and we can see that the characteristic values of the first two principal components of the Sotavento wind farm and the Chang Ma Fengdian farm are respectively: 1.00981, 0.99325, 1.03934 and 0.96260 all conform to principal component selection principles, and the cumulative variance contribution rate at this time reaches 91.29% and 85.32% respectively, so that the first two principal components are selected for prediction by both wind farms.
Step 23: IEO optimization model prediction
Predicting future values of each component and future values of the remainder respectively by using an IEO optimized prediction model; and reconstructing the independent predicted value to obtain a final original wind speed predicted result.
Step 24: error index analysis
The prediction results were comprehensively evaluated using the Mean Absolute Error (MAE), the absolute mean percent error (MAPE), and the Root Mean Square Error (RMSE) as evaluation indexes (see formula 13). The smaller the application index value is, the closer the prediction result is to the actual wind speed value, and the higher the prediction performance of the model is.
Wherein V represents the actual wind speed sequence,and (3) representing a predicted sequence, wherein n is the number of samples, and t is the sampling time.
TABLE 4 eigenvalue and cumulative variance contribution ratio
Step 25: wind speed prediction and error analysis
In this embodiment, the LSTM, the IEO-optimized LSTM model, the IEO-IMIV-LSTM model, and the HMD-based IEO-IMIV-LSTM model are compared to verify the effectiveness of the IEO, introduce multiple factors, and perform IMIV processing, and the necessity and effectiveness of performing HMD decomposition on wind speed data first;
step 251: comparing LSTM with the IEO-LSTM model, and verifying the effect of the IEO on optimizing the LSTM. We analyze from the point of view of the error magnitude and the error fluctuation range:
(1) In Table 5 we present MAE/MAPE/RMSE error results for Sotavento and Chang horse wind farms, from which it can be seen that in both wind farms the IEO-LSTM error index is smaller than the LSTM model compared to the original LSTM;
(2) Analysis is carried out from RMSE indexes alone, and in the prediction of a Sotavento wind power plant, the RMSE error index of an LSTM model after IEO optimization is reduced by 10.08%; similarly, the error index of the RMSE in the Chang Ma Fengdian field is reduced by 16.96%, so that the prediction performance of the LSTM model is improved by IEO;
(3) To show the superiority of IEO more directly, we draw an error box plot between the actual values and the prediction errors in the prediction of both Sotavento and changma wind farms, as shown in fig. 8 and 9. It is clear from fig. 8 that in the prediction of the Sotavento wind farm, the IEO-LSTM prediction error distribution is all closer to zero, and the error fluctuation range is reduced compared to the LSTM model. It can also be derived from fig. 9 that IEO optimizes the LSTM model in the wind speed prediction of the chang horse wind farm. This phenomenon shows that for wind speed data of two wind farms, the IEO-optimized LSTM model can obtain higher accuracy than the LSTM model in short-term wind speed prediction, and has certain advantages in capturing the data fluctuation characteristics. Taken together, we can conclude that IEO improves the predictive performance of LSTM neural networks.
Step 252: to verify the effect of the IMIV and HMD, MATLAB was used to draw graphs of the wind speed predictions of IEO-LSTM, IEO-IMIV-LSTM, and the HMD-based IEO-IMIV-LSTM model, as shown in FIGS. 10 and 11. Wherein fig. 10 and 11 show the results of the Sotavento and the chang horse wind power plants, respectively. In each figure, the abscissa represents the same meaning. The abscissa represents the prediction period and the ordinate represents the wind speed value. Wherein, the band + curve represents the actual wind speed sequence, the band-line represents the IEO-LSTM model prediction result, the black curve represents the IEO-IMIV-LSTM model prediction result, and the band ≡curve represents the HMD-based IEO-IMIV-LSTM model prediction result.
Step 2521: we compared the IEO-LSTM model with the IEO-IMIV-LSTM model to verify the effect of introducing multiple factors and performing IMIV treatment.
(1) As can be seen from fig. 10, in the Sotavento wind farm, the wind speed prediction curve of the IEO-IMIV-LSTM model is closer to the original wind speed sequence in trend; compared with the IEO-LSTM model, the predicted result is closer to the actual wind speed value in value;
(2) As can be seen from fig. 11, at the chang Ma Fengdian field we can also derive: and multiple factors are introduced and processed, so that a prediction model is optimized, the stability of the prediction model is improved, and the prediction precision is improved.
By predicting the wind speeds of two different wind farms, we can see that in the prediction of the wind speeds of the Sotavento and Chang horse wind farms, multiple factors are introduced, and the prediction effect is improved in terms of values and trends. At the same time we calculate three error indicators for both models as shown in table 5. It can be seen that in the wind speed prediction of the Sotavento wind farm, the error index of the IEO-IMIV-LSTM model is obviously reduced compared with that of the IEO-LSTM, and the same conclusion can be obtained in the Change Ma Fengdian farm as can be seen from the table 5. In summary, the effect of introducing multiple factors into the IMIV treatment is further verified.
Step 2522: we compare the black curve and the band ≡curve in fig. 10, and can see that in the wind speed prediction of the Sotavento wind farm, the wind speed prediction result of the HMD-IEO-IMIV-LSTM model is closer to the actual wind speed in both value and trend, compared with the IEO-IMIV-LSTM model. We compare the black curve with the band ≡curve in figure 11 and draw the same conclusion on the chang horse wind farm. Through verification of the two data sets, it is fully explained that the prediction performance of the IEO-IMIV-LSTM neural network can be improved by performing HMD decomposition on wind speed data. From the perspective of error, the error indexes of HMD-IEO-IMIV-LSTM at Sotavento and Chang Ma Fengdian fields are smaller than that of the untreated wind speed IEO-IMIV-LSTM model, which indicates that HMD decomposition of wind speed is essential for the IEO-IMIV-LSTM model proposed herein. Thus, the validity and necessity of HMID are verified.
TABLE 5 error index comparison results
Models 1-4 in Table 5 represent LSTM, IEO-IMIV-LSTM, HMD-IEO-IMIV-LSTM models, respectively.
FIG. 3 is a schematic diagram of a wind speed prediction system according to the present invention. As shown in fig. 3, the prediction system includes:
a wind speed data acquisition and separation module 31 for acquiring an original wind speed sequence; and carrying out singular value decomposition on the wind speed sequence to obtain a denoising sequence and a noise remainder.
The wind speed data acquisition and separation module 31 specifically includes:
the wind speed data acquisition unit is used for acquiring original wind speed data;
the singular value differential spectrum acquisition unit is used for constructing a Hankel matrix for the wind speed sequence, acquiring singular values of the Hankel matrix through a singular value decomposition method, and sequencing the singular values from large to small; obtaining a differential spectrum of the singular values;
and the signal reconstruction unit is used for determining the number of the effective singular values according to the maximum mutation points in the singular value differential spectrum, reconstructing the wind speed signal and obtaining a denoising signal and a noise residual.
The optimal mode number obtaining module 32 is configured to obtain an optimal mode number for performing variant mode decomposition on the denoising sequence according to a mode number optimization method.
The optimal modality number acquisition module 32 specifically includes:
the constraint variation model acquisition unit is used for constructing a constraint variation model under the current mode number for the denoising sequence;
the problem conversion unit is used for introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into a non-constraint problem;
the modal component acquisition unit is used for realizing the self-adaptive decomposition of the signals through Fourier equidistant transformation and other processes; iterative updating, judging whether the Fourier transform of the modal component meets the convergence condition; if yes, obtaining a decomposition result under the mode number condition; if not, the center frequency and the bandwidth of the modal component are adjusted until the Fourier transform of the modal component meets the convergence condition, and the decomposition result of the signal is obtained.
The energy acquisition unit is used for calculating the energy sum of each component under the current mode number;
and the energy difference acquisition unit is used for circulating the steps, calculating the energy difference under the condition of adjacent modal numbers, and the corresponding modal number when the energy difference is obviously increased is the acquired optimal modal number of the variation modal decomposition.
The variation mode decomposition module 33 is configured to decompose the denoising sequence according to a variation mode decomposition method of the optimal mode number, so as to obtain a plurality of component sequences;
The variation mode decomposition module 33 specifically includes:
the constraint variation model construction unit is used for constructing a constraint variation model under the condition of the optimal modal number for the denoising sequence;
the problem conversion unit is used for introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into a non-constraint problem;
the modal component acquisition unit is used for realizing the self-adaptive decomposition of the signals through Fourier equidistant transformation and other processes; iterative updating, judging whether the Fourier transform of the modal component meets a convergence condition; if yes, obtaining a decomposition result of the signal; if not, the center frequency and the bandwidth of the modal component are adjusted until convergence conditions are met, and a plurality of component sequences of the signal are obtained.
The environmental information screening module 34 is configured to obtain other environmental factors that affect wind speed prediction, obtain information with high correlation with the output wind speed from the environmental factors by using an improved average influence value method, and perform wind speed prediction together with the wind speed sequence.
The environmental information screening module 34 specifically includes:
the environment factor acquisition unit is used for acquiring other environment factors such as wind direction, temperature and the like which influence wind speed prediction, and establishing a radial basis function neural network model for the environment factors;
The average influence value obtaining unit is used for adding and subtracting 10% from each environmental factor to obtain a new sample; training the radial basis function neural network on the sample, and acquiring an average influence value between the factors and the model output variable;
the normalization unit is used for acquiring information with high correlation with the output wind speed in the environmental factors according to the average influence value; carrying out normalization processing on the useful information to obtain a dimensionless data set;
a principal component selection unit, configured to obtain a correlation coefficient matrix and an accumulated variance contribution rate of the dataset, and select all principal components with accumulated variance contribution rates greater than 85%;
and the network model building unit is used for building a long-term and short-term memory network model by combining the main components with the component sequences or the remainder.
And the parameter optimization module 35 is used for acquiring the long-term and short-term memory network model, optimizing the network model through an improved extremum optimization algorithm, and acquiring an optimized prediction model.
The parameter optimization module 35 specifically includes:
the parameter optimization unit is used for optimizing the long-period memory network model through an improved extremum optimization algorithm to obtain optimal initialization model parameters;
And the wind speed prediction model determining unit is used for determining a long-period memory network model through the optimal initialization parameters to obtain an optimized long-period memory network prediction model.
A prediction module 36, configured to predict a plurality of the component sequences by using the prediction model, so as to obtain a component prediction result; predicting the noise remainder through the prediction model to obtain a remainder prediction result; and accumulating each component predicted result and each remainder predicted result to obtain a wind speed predicted result.
The prediction module 36 specifically includes:
the component prediction unit is used for predicting each modal component through the prediction model to obtain a plurality of component prediction results;
the remainder prediction unit is used for predicting the remainder sequence through the prediction model to obtain a remainder prediction result;
and the superposition unit is used for linearly superposing the plurality of component prediction results and the remainder prediction results to obtain a wind speed prediction result.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present invention and the core ideas thereof; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (12)
1. A short-term wind speed prediction method based on wind speed characteristics, the prediction method comprising: acquiring an original wind speed sequence; singular value decomposition is carried out on the wind speed sequence, and a denoising sequence and a noise remainder are obtained; acquiring the optimal mode number for performing variation mode decomposition on the denoising sequence according to a mode number optimization method; decomposing the denoising sequence according to a variation modal decomposition method of the optimal modal number to obtain a plurality of component sequences; acquiring environmental factors influencing wind speed prediction, acquiring information with high correlation with output wind speed in the environmental factors by using an improved average influence value method, and establishing a long-term and short-term memory network model by combining the component sequences or remainder; optimizing a network model through an improved extremum optimization algorithm, and obtaining an optimized prediction model; predicting a plurality of component sequences through the prediction model to obtain a component prediction result; predicting the noise remainder through the prediction model to obtain a remainder prediction result; accumulating the predicted results of each component and the remainder to obtain a wind speed predicted result;
The method for acquiring the denoising sequence and the noise remainder specifically comprises the following steps: acquiring original wind speed data; constructing a Hankel matrix for the wind speed sequence, acquiring singular values of the Hankel matrix by a singular value decomposition method, and sequencing the singular values from large to small; obtaining a differential spectrum of the singular values; and determining the number of effective singular values according to the maximum mutation points in the singular value differential spectrum, and reconstructing the wind speed signal to obtain a denoising signal and a noise remainder.
2. The method according to claim 1, wherein the obtaining the optimal modal number for performing a variational modal decomposition on the denoising sequence comprises: constructing a constraint variation model under the current mode number for the denoising sequence; introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into an unconstrained problem; the self-adaptive decomposition of the signals is realized through the Fourier equidistant transformation process; iterative updating, judging whether the Fourier transform of the modal component meets the convergence condition; if yes, obtaining a decomposition result under the mode number condition; if not, adjusting the center frequency and bandwidth of the modal component until the Fourier transform of the modal component meets the convergence condition, obtaining the decomposition result of the signal, and calculating the energy sum of each component under the current modal number; and (3) circulating the steps, and calculating the energy difference under the condition of the adjacent mode numbers, wherein the mode number corresponding to the obviously increased energy difference is the optimal mode number of the acquired variation mode decomposition.
3. The short-term wind speed prediction method according to claim 1, wherein the denoising sequence is decomposed to obtain a plurality of component sequences according to a variation modal decomposition of an optimal modal number, and the method specifically comprises: constructing a constraint variation model under the condition of the optimal modal number for the denoising sequence; introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into an unconstrained problem; the self-adaptive decomposition of the signals is realized through the Fourier equidistant transformation process; iterative updating, judging whether the Fourier transform of the modal component meets the convergence condition; if yes, obtaining a decomposition result of the signal; if not, the center frequency and the bandwidth of the modal component are adjusted until convergence conditions are met, and a plurality of component sequences of the signal are obtained.
4. The method according to claim 1, wherein the obtaining of the environmental factors affecting the wind speed prediction, the obtaining of the information of the environmental factors having a high correlation with the output wind speed by using the improved average influence value method, and the building of the long-term and short-term memory network model by combining the component sequences or the remainder, specifically comprises: acquiring wind direction and temperature environmental factors influencing wind speed prediction, and establishing a radial basis function neural network model for the environmental factors; adding and subtracting 10% from each environmental factor to obtain a new sample; training the radial basis function neural network on the sample, and acquiring an average influence value between the factors and the model output variable; according to the average influence value, obtaining information with high correlation with the output wind speed in the environmental factors; normalizing the information to obtain a dimensionless data set; acquiring a correlation coefficient matrix and an accumulated variance contribution rate of the data set, and selecting all main components with the accumulated variance contribution rate more than 85%; and the main component combines the component sequences or the remainder to establish a long-term and short-term memory network model.
5. The method for predicting short-term wind speed according to claim 1, wherein the optimizing the long-term and short-term memory network model by the improved extremum optimizing algorithm, and obtaining the optimized prediction model, specifically comprises: optimizing the long-term and short-term memory network model through an improved extremum optimization algorithm to obtain optimal initialization model parameters; and determining a long-term and short-term memory network model through the optimal initialization parameters to obtain an optimized long-term and short-term memory network prediction model.
6. The method according to claim 1, wherein the predicting a plurality of the component sequences by the prediction model obtains a component prediction result; predicting the noise remainder through the prediction model to obtain a remainder prediction result; accumulating the component and remainder prediction results to obtain a wind speed prediction result, which specifically comprises: predicting each modal component through the prediction model to obtain a plurality of component prediction results; predicting the remainder sequence through the prediction model to obtain a remainder prediction result; and linearly superposing a plurality of component prediction results and the remainder prediction results to obtain a wind speed prediction result.
7. A short-term wind speed prediction system, the prediction system comprising: the wind speed data acquisition and separation module is used for acquiring an original wind speed sequence; singular value decomposition is carried out on the wind speed sequence, and a denoising sequence and a noise remainder are obtained; the optimal mode number acquisition module is used for acquiring the optimal mode number for performing variation mode decomposition on the denoising sequence according to a mode number optimization method; the variation mode decomposition module is used for decomposing the denoising sequence according to a variation mode decomposition method of the optimal mode number to obtain a plurality of component sequences; the environment information screening module is used for acquiring environment factors influencing wind speed prediction, acquiring information with high relativity with output wind speed in the environment factors by utilizing an improved average influence value method, and establishing a long-term and short-term memory network model by combining the component sequences or the remainder; the parameter optimization module is used for optimizing the network model through an improved extremum optimization algorithm and obtaining an optimized prediction model; the prediction module is used for predicting a plurality of component sequences through the prediction model to obtain a component prediction result; predicting the noise remainder through the prediction model to obtain a remainder prediction result; accumulating each component prediction result and each remainder prediction result to obtain a wind speed prediction result;
The wind speed data acquisition and separation module specifically comprises: the wind speed data acquisition unit is used for acquiring original wind speed data; the singular value differential spectrum acquisition unit is used for constructing a Hankel matrix for the wind speed sequence, acquiring singular values of the Hankel matrix through a singular value decomposition method, and sequencing the singular values from large to small; obtaining a differential spectrum of the singular values; and the signal reconstruction unit is used for determining the number of the effective singular values according to the maximum mutation points in the singular value differential spectrum, reconstructing the wind speed signal and obtaining a denoising signal and a noise residual.
8. The short-term wind speed prediction system according to claim 7, wherein the optimal modal number acquisition module specifically comprises: the constraint variation model acquisition unit is used for constructing a constraint variation model under the current mode number for the denoising sequence; the problem conversion unit is used for introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into an unconstrained problem; the modal component acquisition unit is used for realizing the self-adaptive decomposition of the signals through a Fourier equidistant transformation process; iterative updating, judging whether the Fourier transform of the modal component meets the convergence condition; if yes, obtaining a decomposition result under the mode number condition; if not, adjusting the center frequency and bandwidth of the modal component until the Fourier transform of the modal component meets a convergence condition, and acquiring a decomposition result of the signal, wherein the energy acquisition unit is used for calculating the energy sum of all components under the current modal number; and the energy difference acquisition unit is used for circulating the steps, calculating the energy difference under the condition of adjacent modal numbers, and the corresponding modal number when the energy difference is obviously increased is the acquired optimal modal number of the variation modal decomposition.
9. The short-term wind speed prediction system according to claim 7, wherein the variation modal decomposition module specifically comprises: the constraint variation model construction unit is used for constructing a constraint variation model under the condition of the optimal modal number for the denoising sequence; the problem conversion unit is used for introducing a secondary penalty factor and Lagrange multiplication operator to change the constraint problem into an unconstrained problem; the modal component acquisition unit is used for realizing the self-adaptive decomposition of the signals through a Fourier equidistant transformation process; iterative updating, judging whether the Fourier transform of the modal component meets a convergence condition; if yes, obtaining a decomposition result of the signal; if not, the center frequency and the bandwidth of the modal component are adjusted until convergence conditions are met, and a plurality of component sequences of the signal are obtained.
10. The short-term wind speed prediction system according to claim 7, wherein the environmental information screening module specifically comprises: the environment factor acquisition unit is used for acquiring wind direction and temperature environment factors affecting wind speed prediction and establishing a radial basis function neural network model for the environment factors; the average influence value acquisition unit is used for adding and subtracting 10% from each environmental factor to acquire a new sample; training the radial basis function neural network on the sample, and acquiring an average influence value between the factors and the model output variable; the normalization unit is used for acquiring information with high correlation with the output wind speed in the environmental factors according to the average influence value; normalizing the information to obtain a dimensionless data set; a principal component selection unit, configured to obtain a correlation coefficient matrix and an accumulated variance contribution rate of the dataset, and select all principal components with accumulated variance contribution rates greater than 85%; and the network model building unit is used for building a long-term and short-term memory network model by combining the main components with the component sequences or the remainder.
11. The short-term wind speed prediction system according to claim 7, wherein the parameter optimization module specifically comprises:
the parameter optimization unit is used for optimizing the long-period memory network model through an improved extremum optimization algorithm to obtain optimal initialization model parameters; and the wind speed prediction model determining unit is used for determining a long-period memory network model through the optimal initialization parameters to obtain an optimized long-period memory network prediction model.
12. The short-term wind speed prediction system according to claim 7, wherein the prediction module specifically comprises: the component prediction unit is used for predicting each modal component through the prediction model to obtain a plurality of component prediction results; the remainder prediction unit is used for predicting the remainder sequence through the prediction model to obtain a remainder prediction result; and the superposition unit is used for linearly superposing the plurality of component prediction results and the remainder prediction results to obtain a wind speed prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910601512.6A CN110309603B (en) | 2019-07-05 | 2019-07-05 | Short-term wind speed prediction method and system based on wind speed characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910601512.6A CN110309603B (en) | 2019-07-05 | 2019-07-05 | Short-term wind speed prediction method and system based on wind speed characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110309603A CN110309603A (en) | 2019-10-08 |
| CN110309603B true CN110309603B (en) | 2023-10-24 |
Family
ID=68079223
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910601512.6A Active CN110309603B (en) | 2019-07-05 | 2019-07-05 | Short-term wind speed prediction method and system based on wind speed characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110309603B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110826791A (en) * | 2019-10-31 | 2020-02-21 | 东北电力大学 | Hybrid wind power prediction method based on long-time and short-time memory neural network |
| CN111007559B (en) * | 2019-11-25 | 2023-09-15 | 中国辐射防护研究院 | Gamma energy spectrum smoothing method based on Hilbert-Huang transformation |
| CN111523728B (en) * | 2020-04-23 | 2022-05-10 | 东北电力大学 | Four-stage hybrid short-term wind direction prediction method |
| CN111998935B (en) * | 2020-08-11 | 2022-08-19 | 南京理工大学 | Time/frequency decomposition method for explosive shock wave signal based on variational modal decomposition |
| CN112559962B (en) * | 2020-11-20 | 2023-12-26 | 中国石油天然气股份有限公司 | Digital signal noise reduction method and device |
| CN112505486B (en) * | 2020-12-03 | 2024-02-20 | 山西世纪中试电力科学技术有限公司 | Source-load-storage integrated grid-connected power quality testing system |
| CN114336612B (en) * | 2021-12-31 | 2023-12-22 | 阳光智维科技有限公司 | Power station and electric quantity loss calculation method of fault equipment of power station and related application equipment |
| CN116796890A (en) * | 2023-05-29 | 2023-09-22 | 南方电网能源发展研究院有限责任公司 | Carbon emission right cost prediction method, device, equipment, medium and product |
| CN117536800A (en) * | 2023-11-13 | 2024-02-09 | 无锡学院 | Wind power equipment data acquisition system |
| CN117408164B (en) * | 2023-12-13 | 2024-03-22 | 西安热工研究院有限公司 | Intelligent wind speed prediction method and system for energy storage auxiliary black start |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102938093A (en) * | 2012-10-18 | 2013-02-20 | 安徽工程大学 | Wind power forecasting method |
| CN106446829A (en) * | 2016-09-22 | 2017-02-22 | 三峡大学 | Hydroelectric generating set vibration signal noise reduction method based on mode autocorrelation analysis of SVD and VMD |
| CN108197648A (en) * | 2017-12-28 | 2018-06-22 | 华中科技大学 | A kind of Fault Diagnosis Method of Hydro-generating Unit and system based on LSTM deep learning models |
| CN108615097A (en) * | 2018-05-10 | 2018-10-02 | 广东工业大学 | A kind of wind speed forecasting method, system, equipment and computer readable storage medium |
| CN109146162A (en) * | 2018-08-07 | 2019-01-04 | 河海大学 | A kind of probability wind speed forecasting method based on integrated Recognition with Recurrent Neural Network |
| CN109948833A (en) * | 2019-02-25 | 2019-06-28 | 华中科技大学 | A kind of Hydropower Unit degradation trend prediction technique based on shot and long term memory network |
-
2019
- 2019-07-05 CN CN201910601512.6A patent/CN110309603B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102938093A (en) * | 2012-10-18 | 2013-02-20 | 安徽工程大学 | Wind power forecasting method |
| CN106446829A (en) * | 2016-09-22 | 2017-02-22 | 三峡大学 | Hydroelectric generating set vibration signal noise reduction method based on mode autocorrelation analysis of SVD and VMD |
| CN108197648A (en) * | 2017-12-28 | 2018-06-22 | 华中科技大学 | A kind of Fault Diagnosis Method of Hydro-generating Unit and system based on LSTM deep learning models |
| CN108615097A (en) * | 2018-05-10 | 2018-10-02 | 广东工业大学 | A kind of wind speed forecasting method, system, equipment and computer readable storage medium |
| CN109146162A (en) * | 2018-08-07 | 2019-01-04 | 河海大学 | A kind of probability wind speed forecasting method based on integrated Recognition with Recurrent Neural Network |
| CN109948833A (en) * | 2019-02-25 | 2019-06-28 | 华中科技大学 | A kind of Hydropower Unit degradation trend prediction technique based on shot and long term memory network |
Non-Patent Citations (1)
| Title |
|---|
| A novel hybrid model based on VMD-WT and PCA-BP-RBF neural network for short-term wind speed forecasting;Yagang Zhang et al.;《ELSEVIER》;20190509;第180-197页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110309603A (en) | 2019-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110309603B (en) | Short-term wind speed prediction method and system based on wind speed characteristics | |
| Qu et al. | Multi-step wind speed forecasting based on a hybrid decomposition technique and an improved back-propagation neural network | |
| Tian | Short-term wind speed prediction based on LMD and improved FA optimized combined kernel function LSSVM | |
| CN105391083B (en) | Wind power interval short term prediction method based on variation mode decomposition and Method Using Relevance Vector Machine | |
| CN110543929B (en) | Wind speed interval prediction method and system based on Lorenz system | |
| CN112686464A (en) | Short-term wind power prediction method and device | |
| CN109886464B (en) | Low-information-loss short-term wind speed prediction method based on optimized singular value decomposition generated feature set | |
| CN112434848B (en) | Nonlinear weighted combination wind power prediction method based on deep belief network | |
| Lv et al. | Short-term wind speed forecasting based on non-stationary time series analysis and ARCH model | |
| CN109255728B (en) | Photovoltaic power generation power neural network prediction method based on chaotic phase space optimization reconstruction | |
| CN112100911B (en) | Solar radiation prediction method based on depth BILSTM | |
| CN111144644A (en) | Short-term wind speed prediction method based on variation variance Gaussian process regression | |
| CN110942170A (en) | Short-term wind speed prediction method and system based on information processing | |
| CN111553510A (en) | Short-term wind speed prediction method | |
| CN112434891A (en) | Method for predicting solar irradiance time sequence based on WCNN-ALSTM | |
| CN111222689A (en) | LSTM load prediction method, medium, and electronic device based on multi-scale temporal features | |
| CN114676622A (en) | Short-term photovoltaic power prediction method based on self-encoder deep learning model | |
| CN114266416A (en) | Photovoltaic power generation power short-term prediction method and device based on similar days and storage medium | |
| Sun et al. | A compound structure for wind speed forecasting using MKLSSVM with feature selection and parameter optimization | |
| CN113361782A (en) | Photovoltaic power generation power short-term rolling prediction method based on improved MKPLS | |
| CN111563631A (en) | Wind power generation power prediction method and device based on variational modal decomposition | |
| CN116665483A (en) | Novel method for predicting residual parking space | |
| CN117114160A (en) | Short-term photovoltaic power prediction method | |
| CN114298132A (en) | Wind power prediction method and device and electronic equipment | |
| Chenyu et al. | Ultra-short-term wind power prediction based on variational mode decomposition and clustering method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |