CN111160520A - BP neural network wind speed prediction method based on genetic algorithm optimization - Google Patents

BP neural network wind speed prediction method based on genetic algorithm optimization Download PDF

Info

Publication number
CN111160520A
CN111160520A CN201911244226.5A CN201911244226A CN111160520A CN 111160520 A CN111160520 A CN 111160520A CN 201911244226 A CN201911244226 A CN 201911244226A CN 111160520 A CN111160520 A CN 111160520A
Authority
CN
China
Prior art keywords
neural network
wind speed
individual
initial
population
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
Application number
CN201911244226.5A
Other languages
Chinese (zh)
Inventor
李军
潘茹悦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Science and Technology
Original Assignee
Nanjing University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing University of Science and Technology filed Critical Nanjing University of Science and Technology
Priority to CN201911244226.5A priority Critical patent/CN111160520A/en
Publication of CN111160520A publication Critical patent/CN111160520A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Business, Economics & Management (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Economics (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Strategic Management (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • Human Resources & Organizations (AREA)
  • Evolutionary Biology (AREA)
  • Marketing (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Public Health (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Development Economics (AREA)
  • Primary Health Care (AREA)
  • Water Supply & Treatment (AREA)
  • Physiology (AREA)
  • Genetics & Genomics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Wind Motors (AREA)

Abstract

The invention discloses a BP neural network wind speed prediction method based on genetic algorithm optimization. The method comprises the following steps: firstly, collecting wind speed data of a wind power plant, establishing a BP neural network prediction model, and estimating an initial value range; then, carrying out real number coding on the weight and the threshold of the neural network, randomly generating a group of initial individuals to form an initial population, wherein each initial individual represents an initial solution of the problem; calculating the fitness of each individual in the population, performing selection, crossing and mutation operations to form a next generation population, evaluating the individual fitness of the new population, judging a convergence condition, selecting an optimal individual, and taking the optimal individual as an initial weight and a threshold of a neural network; and finally, training by utilizing matlab to obtain a wind speed predicted value. The invention improves the efficiency and the accuracy of the BP neural network wind speed prediction.

Description

BP neural network wind speed prediction method based on genetic algorithm optimization
Technical Field
The invention relates to the technical field of wind speed prediction of wind power plants, in particular to a BP neural network wind speed prediction method based on genetic algorithm optimization.
Background
With the development of economy and society, people have more and more requirements on clean energy, and wind energy has great development potential as a renewable clean energy. However, wind power generation is intermittent and time-varying, so that accurate wind speed prediction is particularly important for accurate prediction of wind speed, and the accurate wind speed prediction is helpful for ensuring safe, stable and economic operation of a power system.
The wind speed prediction model is mainly based on the establishment of a functional mapping relation between future data and historical data, namely, the future output is predicted according to the historical data. The artificial neural network has self-learning and self-organizing capabilities, the learning and training of the artificial neural network is the optimization of a network structure and weight coefficients, wherein the BP neural network is most widely and mature in application and has good function approximation capability, the number of neurons of the three-layer structure of the BP neural network is determined by researching historical wind speeds at the first moments influencing the wind speeds at the next moment, historical data is used as input, and a BP neural network wind speed prediction model is obtained by training the BP neural network. And then, improving by using a genetic algorithm, firstly carrying out real number coding on the weight and the threshold of the neural network, randomly generating a group of initial individuals to form an initial population, and enabling each initial individual to represent an initial solution of the problem. And then calculating the fitness of each individual in the population, performing selection, crossing and mutation operations to form a next generation population, finally judging a convergence condition and selecting an optimal individual, taking the optimal individual as an initial weight and a threshold of a neural network, and then training by utilizing matlab to finally obtain a 24-hour wind speed predicted value in 31 days of the month.
The BP neural network prediction model can be described as:
Figure BDA0002307077690000011
in the formula T1(T) to T24(T) represents the predicted wind speed values at 31 days 0 to 23, respectively, and T1(t-Δt),T1(t-2Δt),T1(t-3Δt),T1(t-4Δt),T1(T-5. DELTA.t) represents the same as T1(t) wind speed data of the first 1 hour, the first 2 hours, the first 3 hours, the first 4 hours and the first 5 hours adjacent to each other, and so on.
The BP neural network algorithm has the problem of local optimization, and the training process may be in local minimum. When bad data is encountered, the BP neural network training becomes difficult to converge, greatly affecting the efficiency and accuracy of prediction.
Disclosure of Invention
The invention aims to provide a BP neural network wind speed prediction method based on genetic algorithm optimization, which is simple in algorithm, high in efficiency and high in accuracy.
The technical solution for realizing the purpose of the invention is as follows: a BP neural network wind speed prediction method based on genetic algorithm optimization comprises the following steps:
step 1, acquiring wind speed data of a wind power plant, dividing the sample data into a training sample set and a test sample set, carrying out normalization processing, establishing a BP neural network prediction model, and estimating an initial value range;
step 2, carrying out real number coding on the weight and the threshold of the BP neural network, randomly generating a group of initial individuals to form an initial population, wherein each initial individual represents an initial solution of the problem;
step 3, calculating the fitness of each individual in the population;
step 4, carrying out selection, crossing and mutation operations to form a next generation population, and evaluating the individual fitness of the new population;
step 5, judging whether a convergence condition is reached, if so, entering step 6, otherwise, returning to step 4;
and 6, selecting the optimal individuals as initial weights and thresholds of the BP neural network to predict the wind speed.
Further, the wind speed data of the wind power plant obtained in step 1 is divided into a training sample set and a test sample set, normalization processing is performed, a BP neural network prediction model is established, and an initial value range is estimated, specifically as follows:
step 1.1, selecting input training sample data according to a rolling method, namely, arranging 24-hour wind speed data of each day in one month into a group according to time sequence, and predicting the wind speed of the next hour by using wind speed of every 5 adjacent hours;
step 1.2, normalization processing is respectively carried out on an input training sample, an output training sample and an input test sample by using a mapminmax function carried by matlab, a BP neural network prediction model is built, and an initial value range is estimated.
Further, the BP neural network described in step 1 specifically includes the following steps:
the number of input layer neurons R is 5, and the number of output layer neurons S is S21, number of hidden layer neurons according to empirical formula
Figure BDA0002307077690000021
a is a natural number of 1 to 10, S1The value is 12.
Further, the step 2 of performing real number coding on the weight and the threshold of the BP neural network specifically includes:
using real number encoding, i.e. floating-point number encoding, the encoding length, i.e. the individual length, being S ═ R × S1+S1+S1*S2+S2(ii) a When floating-point number coding is adopted, each gene value of an individual is represented by one floating-point number in a set range; r is the number of neurons in the input layer, S2Is the number of neurons in the output layer, S1The number of hidden layer neurons is, and a is a natural number in 1-10.
Further, the calculating of the fitness of each individual in the population in step 3 specifically includes the following steps:
first, an objective function is calculated, which is expressed by the sum of squared errors e (i), and the formula is:
Figure BDA0002307077690000031
wherein i 1.. N is the number of chromosomes; k is the number of output layer nodes; p is the number of training samples; t iskTo output a test sample; vkA predictor representing a kth output node;
and calculating each individual evaluation function, wherein the probability pi of the individual i being selected is as follows:
Figure BDA0002307077690000032
wherein f isiThe adaptation value, i.e. fitness, for an individual i is measured by the inverse of the sum of the squares of the errors, i.e.
Figure BDA0002307077690000033
Further, the determination of whether the convergence condition is reached in step 5 specifically includes the following steps:
and (3) presetting an evolution algebra, and when the evolution process reaches the set algebra, namely the termination condition is met, ending the optimization process of the BP neural network, and outputting the optimal individual as the initial weight and the threshold of the BP neural network.
Compared with the prior art, the invention has the remarkable advantages that: (1) an approximate optimal solution is found in the global range and is used as an initial weight and a threshold of the BP neural network, and then the optimal solution is found through the existing BP neural network training method, so that the situation that the optimal solution is trapped in local optimization is avoided, and the operation efficiency and the accuracy are improved; (2) the method has relatively low requirement on data, belongs to parallel search, has high search efficiency, effectively solves the phenomenon that a neural network is difficult to converge when bad data appears, and improves the stability of prediction.
Drawings
FIG. 1 is a schematic structural diagram of a wind speed prediction model based on a BP neural network.
Fig. 2 is a flow chart diagram of the BP algorithm.
FIG. 3 is a flow chart diagram of a BP neural network wind speed prediction method based on genetic algorithm optimization.
Fig. 4 is a graph of the predicted effect in an embodiment of the present invention.
FIG. 5 is a graph of the genetic algorithm squared error sum variation in an embodiment of the present invention.
Detailed Description
The invention relates to a BP neural network wind speed prediction method based on genetic algorithm optimization, which comprises the following steps:
step 1, acquiring wind speed data of a wind power plant, dividing the sample data into a training sample set and a test sample set, carrying out normalization processing, establishing a BP neural network prediction model, and estimating an initial value range;
step 2, carrying out real number coding on the weight and the threshold of the BP neural network, randomly generating a group of initial individuals to form an initial population, wherein each initial individual represents an initial solution of the problem;
step 3, calculating the fitness of each individual in the population;
step 4, carrying out selection, crossing and mutation operations to form a next generation population, and evaluating the individual fitness of the new population;
step 5, judging whether a convergence condition is reached, if so, entering step 6, otherwise, returning to step 4;
and 6, selecting the optimal individuals as initial weights and thresholds of the BP neural network to predict the wind speed.
Further, the wind speed data of the wind power plant obtained in step 1 is divided into a training sample set and a test sample set, normalization processing is performed, a BP neural network prediction model is established, and an initial value range is estimated, specifically as follows:
step 1.1, selecting input training sample data according to a rolling method, namely, arranging 24-hour wind speed data of each day in one month into a group according to time sequence, and predicting the wind speed of the next hour by using wind speed of every 5 adjacent hours;
step 1.2, normalization processing is respectively carried out on an input training sample, an output training sample and an input test sample by using a mapminmax function carried by matlab, a BP neural network prediction model is built, and an initial value range is estimated.
Further, the BP neural network described in step 1 specifically includes the following steps:
the number of input layer neurons R is 5, and the number of output layer neurons S is S21, number of hidden layer neurons according to empirical formula
Figure BDA0002307077690000041
a is a natural number of 1 to 10, S1The value is 12.
Further, the step 2 of performing real number coding on the weight and the threshold of the BP neural network specifically includes:
using real number encoding, i.e. floating-point number encoding, the encoding length, i.e. the individual length, being S ═ R × S1+S1+S1*S2+S2(ii) a When floating-point number coding is adopted, each gene value of an individual is represented by one floating-point number in a set range; r is the number of neurons in the input layer, S2Is the number of neurons in the output layer, S1The number of hidden layer neurons is, and a is a natural number in 1-10.
Further, the calculating of the fitness of each individual in the population in step 3 specifically includes the following steps:
first, an objective function is calculated, which is expressed by the sum of squared errors e (i), and the formula is:
Figure BDA0002307077690000042
wherein i 1.. N is the number of chromosomes; k is the number of output layer nodes; p is the number of training samples; t iskTo output a test sample; vkA predictor representing a kth output node;
and calculating each individual evaluation function, wherein the probability pi of the individual i being selected is as follows:
Figure BDA0002307077690000051
wherein f isiThe adaptation value, i.e. fitness, for an individual i is measured by the inverse of the sum of the squares of the errors, i.e.
Figure BDA0002307077690000052
Further, the determination of whether the convergence condition is reached in step 5 specifically includes the following steps:
and (3) presetting an evolution algebra, and when the evolution process reaches the set algebra, namely the termination condition is met, ending the optimization process of the BP neural network, and outputting the optimal individual as the initial weight and the threshold of the BP neural network.
The invention is described in further detail below with reference to the figures and the embodiments.
With reference to fig. 1, the wind speed prediction model diagram based on the BP neural network is divided into three layers of structures 5-12-1, namely, 5 neurons in an input layer, 12 neurons in a hidden layer, and 1 neuron in an output layer, wherein R, S1, and S2 respectively represent dimensions of the input layer, the hidden layer, and the output layer, and IW and LW respectively represent a weight matrix from the input layer to the hidden layer and a weight matrix from the hidden layer to the output layer.
When the BP neural network is used for wind speed prediction, the input layer comprises the hour wind speed of the first 5 moments, the hour wind speed of the next moment is output, and the output layer is one-dimensional.
Fig. 2 is a flow chart of the BP algorithm of the present invention. The following describes the execution procedure specifically:
(1) firstly, data acquisition and preprocessing: acquiring wind speed data of 5 months of a wind power plant in a certain place every day in whole hour, dividing the sample data into a training sample set and a testing sample set, and carrying out normalization processing on the training sample set and the testing sample set.
(2) And (3) constructing a BP neural network by using the obtained sample set, and predicting the wind speed: this includes determining the number of hidden layer neurons, the number of input and output layer neurons of the BP neural network from the input-output data dimensions. Then BP network training is carried out to obtain 24-hour wind speed prediction data of 31 th day before improvement; wherein, the input training sample data is selected according to a rolling method. That is, 744 data of 24-hour wind speeds of all 31 days are arranged in a time-series order, and wind speed of the next hour is predicted every 5 adjacent hour wind speeds. A total of 715 training samples and 24 test samples were generated.
FIG. 3 is a flowchart of a BP neural network wind speed prediction algorithm based on genetic algorithm optimization. The initial weight and the threshold of the neural network are improved by utilizing a genetic algorithm, firstly, the weight and the threshold of the neural network are subjected to real number coding, a group of initial individuals are randomly generated to form an initial population, and each initial individual represents an initial solution of the problem. And then calculating the fitness of each individual in the population, performing selection, crossing and mutation operations to form a next generation population, finally judging a convergence condition and selecting an optimal individual, taking the optimal individual as an initial weight and a threshold of a neural network, and then training by utilizing matlab to finally obtain a 24-hour wind speed predicted value in 31 days of the month.
(1) Data acquisition: training sample data and test sample data after normalization; the parameters of the genetic algorithm: the number of population individuals (population size), maximum evolution algebra, cross probability, mutation probability and the like; parameters of the BP algorithm: target error, maximum number of iterations.
(2) Encoding, forming an initial population: w1 is the weight from input layer to hidden layer, there are S1R; b1 is hidden layer threshold, there are S1; w2 is weight from hidden layer to output layer, and there are S2 × S1; b2 is the output layer threshold, and there are S2.
(3) Calculating an individual fitness value:
first, an objective function is calculated, which can be expressed by the sum of squared errors, and the formula is:
Figure BDA0002307077690000061
wherein i 1.. N is the number of chromosomes; k is the number of output layer nodes; p is the number of training samples; vkA predictor representing a kth output node; t iskTo output a test sample. Calculating each individual evaluation function, and then selecting the probability that the individual i is:
Figure BDA0002307077690000062
wherein f isiThe fitness value (fitness) of an individual i can be measured by the inverse of the sum of the squares of the errors mentioned above, i.e.
Figure BDA0002307077690000063
(4) Genetic manipulation: respectively carrying out selection, crossing and mutation operations on individuals in the population to generate a new population, and evaluating the individual fitness of the new population by adopting the method;
(5) judging an optimization termination condition: the evolution algebra is preset, when the evolution process reaches a certain algebra, namely the termination condition is met, the optimization process of the BP neural network is ended, and the optimal individual is output and used as the initial weight and the threshold of the BP neural network.
(6) And (4) establishing a BP neural network by using the obtained optimized initial weight and threshold, initializing the BP neural network, and predicting the wind speed of 24 hours in 31 days.
Example 1
In this embodiment, wind speed data of 5 months of a wind farm in whole hour each day is divided into training sample data and test sample data.
Step 1, acquiring wind speed data of a wind power plant, dividing the sample data into a training sample set and a testing sample set, carrying out normalization processing, establishing a BP neural network prediction model, and estimating an initial value range, wherein the method specifically comprises the following steps:
step 1.1, inputting data: (tables 1-1, tables 1-2, tables 1-3)
TABLE 1-1
Figure BDA0002307077690000071
Tables 1 to 2
Figure BDA0002307077690000072
Tables 1 to 3
Figure BDA0002307077690000081
Step 1.2, constructing a BP neural network, estimating an initial value range:
the input training sample data is selected according to a rolling method, namely 744 data of wind speed in 24 hours of all 31 days are arranged into a group according to time sequence, wind speed in the next hour is predicted according to wind speed in every 5 adjacent hours, and 715 groups of training samples and 24 groups of test samples are generated in total. The input training samples are (5 × 715) matrix, the output training samples are (1 × 715) matrix, the input testing samples are (5 × 24) matrix, and the output testing samples are (1 × 24) matrix. Carrying out normalization and inverse normalization processing on the sample data by using a mapminmax function of Matlab to obtain a wind speed predicted value of 5 months, 31 days and 24 hours based on a BP algorithm, wherein the predicted value is as follows:
BPOutput_test＝
Columns 1through 15
10.8564 10.6518 12.6806 11.5194 11.6081 11.3704 10.2768
10.1029 10.4036 9.9832 10.9769 10.5026 9.0682 8.0546 7.6115
Columns 16through 24
9.1529 7.4169 10.1395 10.7974 9.3148 10.5313 10.4000
5.3905 8.0092
step 1.3, performing normalization processing on an input training sample, an output training sample and an input test sample by using a mapminmax function carried by matlab, establishing a BP neural network prediction model, and estimating an initial value range;
w1 is the weight from input layer to hidden layer, there are S1R; b is1For the hidden layer threshold, there are S1; w2 is weight from hidden layer to output layer, and there are S2 × S1; b2 is the output layer threshold, and there are S2. Using real number encoding, i.e. floating point encoding, the encoding length, i.e. the individual length, i.e. the length of the chromosome, is S ═ R × S1+S1+S1*S2+S2In this embodiment, S is 5 × 12+12+12 × 1+1 is 85.
Step 2, carrying out real number coding on the weight and the threshold of the BP neural network, randomly generating a group of initial individuals to form an initial population, wherein each initial individual represents an initial solution of the problem;
step 3, calculating the fitness of each individual in the population;
and 4, carrying out selection, crossing and mutation operations to form a next generation population, and evaluating the individual fitness of the new population, wherein the method specifically comprises the following steps:
and (3) selecting, crossing and mutating the formed population individuals to finally obtain the optimal individual, which comprises the following steps:
Figure BDA0002307077690000091
W2＝[-0.0419 0.1999 0.5783 -0.5555 -0.0480 0.0871 0.0459 -0.0345 -0.3965 0.7523 0.6577 -0.0730]B2＝[-0.2531]，val＝[0.0620]；
step 5, judging whether a convergence condition is reached, if so, entering step 6, otherwise, returning to step 4;
and 6, selecting the optimal individuals as the initial weight and the threshold of the BP neural network to predict the wind speed, wherein the method specifically comprises the following steps:
and (3) carrying out BP neural network wind speed prediction based on genetic algorithm optimization by using the obtained optimal initial weight and threshold, and obtaining a wind speed predicted value of 24 hours in 31 days in 5 months as follows:
GABPOutput_test＝
Columns 1through 15
11.0932 10.8554 13.0732 11.8340 12.3356 11.6376 10.2962 10.013110.3260 9.8504 10.9748 10.3161 8.7381 7.7269 7.0412
Columns 16through 24
9.0398 7.6116 10.1849 10.3671 9.3636 10.3485 10.1292 4.9127 8.3211
the results are shown in tables 2 and 3:
TABLE 2
Figure BDA0002307077690000101
TABLE 3
Figure BDA0002307077690000102
And (3) error analysis: as can be seen from fig. 4, the line with a dotted sign represents the actual wind speed value at 24 hours in 31 days, the line with a circle represents the predicted value obtained by the BP algorithm, and the line with a dotted sign represents the predicted value obtained by the BP algorithm optimized based on the genetic algorithm. The mean square error of the BP algorithm is 2.7222, while the mean square error of the GA-BP algorithm is 2.6509. As can be seen from fig. 5, after approximately 45 times or so of evolution, the square sum error changes smoothly and no longer changes greatly. In conclusion, the BP neural network wind speed prediction method based on genetic algorithm optimization is more efficient and accurate than a single BP neural network wind speed prediction method.

Claims (6)

1. A BP neural network wind speed prediction method based on genetic algorithm optimization is characterized by comprising the following steps:
step 1, acquiring wind speed data of a wind power plant, dividing the sample data into a training sample set and a test sample set, carrying out normalization processing, establishing a BP neural network prediction model, and estimating an initial value range;
step 2, carrying out real number coding on the weight and the threshold of the BP neural network, randomly generating a group of initial individuals to form an initial population, wherein each initial individual represents an initial solution of the problem;
step 3, calculating the fitness of each individual in the population;
step 4, carrying out selection, crossing and mutation operations to form a next generation population, and evaluating the individual fitness of the new population;
step 5, judging whether a convergence condition is reached, if so, entering step 6, otherwise, returning to step 4;
and 6, selecting the optimal individuals as initial weights and thresholds of the BP neural network to predict the wind speed.
2. The genetic algorithm optimization-based BP neural network wind speed prediction method according to claim 1, characterized in that, the wind speed data of the wind farm is obtained in step 1, the sample data is divided into a training sample set and a testing sample set, normalization processing is performed, a BP neural network prediction model is established, and an initial value range is estimated, specifically as follows:
step 1.1, selecting input training sample data according to a rolling method, namely, arranging 24-hour wind speed data of each day in one month into a group according to time sequence, and predicting the wind speed of the next hour by using wind speed of every 5 adjacent hours;
step 1.2, normalization processing is respectively carried out on an input training sample, an output training sample and an input test sample by using a mapminmax function carried by matlab, a BP neural network prediction model is built, and an initial value range is estimated.
3. The BP neural network wind speed prediction method based on genetic algorithm optimization according to claim 1, wherein the BP neural network in the step 1 is specifically as follows:
the number of input layer neurons R is 5, and the number of output layer neurons S is S21, number of hidden layer neurons according to empirical formula
Figure FDA0002307077680000011
a is a natural number of 1 to 10, S1The value is 12.
4. The BP neural network wind speed prediction method based on genetic algorithm optimization according to claim 1, wherein the weight and the threshold of the BP neural network are real number encoded in step 2, specifically as follows:
using real number encoding, i.e. floating-point number encoding, the encoding length, i.e. the individual length, being S ═ R × S1+S1+S1*S2+S2(ii) a When floating-point number coding is adopted, each gene value of an individual is represented by one floating-point number in a set range; r is the number of neurons in the input layer, S2Is the number of neurons in the output layer, S1The number of hidden layer neurons is, and a is a natural number in 1-10.
5. The BP neural network wind speed prediction method based on genetic algorithm optimization according to claim 1, wherein the calculation of fitness of each individual in a population in step 3 is as follows:
first, an objective function is calculated, which is expressed by the sum of squared errors e (i), and the formula is:
Figure FDA0002307077680000021
wherein i 1.. N is the number of chromosomes; k is the number of output layer nodes; p is the number of training samples; t iskTo output a test sample; vkRepresenting the kth output nodePredicting a value;
and calculating each individual evaluation function, wherein the probability pi of the individual i being selected is as follows:
Figure FDA0002307077680000022
wherein f isiThe adaptation value, i.e. fitness, for an individual i is measured by the inverse of the sum of the squares of the errors, i.e.
Figure FDA0002307077680000023
6. The BP neural network wind speed prediction method based on genetic algorithm optimization according to claim 1, wherein the judgment of whether the convergence condition is reached in step 5 is as follows:
and (3) presetting an evolution algebra, and when the evolution process reaches the set algebra, namely the termination condition is met, ending the optimization process of the BP neural network, and outputting the optimal individual as the initial weight and the threshold of the BP neural network.
CN201911244226.5A 2019-12-06 2019-12-06 BP neural network wind speed prediction method based on genetic algorithm optimization Pending CN111160520A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911244226.5A CN111160520A (en) 2019-12-06 2019-12-06 BP neural network wind speed prediction method based on genetic algorithm optimization

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911244226.5A CN111160520A (en) 2019-12-06 2019-12-06 BP neural network wind speed prediction method based on genetic algorithm optimization

Publications (1)

Publication Number Publication Date
CN111160520A true CN111160520A (en) 2020-05-15

Family

ID=70556394

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911244226.5A Pending CN111160520A (en) 2019-12-06 2019-12-06 BP neural network wind speed prediction method based on genetic algorithm optimization

Country Status (1)

Country Link
CN (1) CN111160520A (en)

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111726349A (en) * 2020-06-16 2020-09-29 桂林电子科技大学 GRU parallel network flow abnormity detection method based on GA optimization
CN111859807A (en) * 2020-07-23 2020-10-30 润电能源科学技术有限公司 Initial pressure optimizing method, device, equipment and storage medium for steam turbine
CN111898827A (en) * 2020-08-03 2020-11-06 西安石油大学 Motorcycle helmet wearing condition prediction evaluation model based on neural network
CN112289392A (en) * 2020-10-28 2021-01-29 广西贺州市桂东电子科技有限责任公司 Method for predicting performance of anode aluminum foil for aluminum electrolytic capacitor
CN112487726A (en) * 2020-12-21 2021-03-12 北京航天测控技术有限公司 Training method and device for target test model
CN112651176A (en) * 2020-12-24 2021-04-13 广西大学 GA-BP algorithm-based shielded door motor output force prediction method
CN112749792A (en) * 2021-02-02 2021-05-04 南京信息工程大学 Wind speed prediction method based on BP algorithm
CN113095477A (en) * 2021-03-17 2021-07-09 西安理工大学 Wind power prediction method based on DE-BP neural network
CN113159299A (en) * 2021-04-30 2021-07-23 杭州电子科技大学 Fractional order depth BP neural network optimization method based on extremum optimization
CN113158586A (en) * 2021-05-26 2021-07-23 华能新能源股份有限公司 Wind power plant numerical simulation method and device combined with neural network algorithm and storage medium
CN113483473A (en) * 2021-03-29 2021-10-08 南方环境有限公司 Welding workshop environment control method based on genetic-neural network (GA-BP) model
CN113610297A (en) * 2021-08-06 2021-11-05 浙江工业大学之江学院 Air quality prediction method, device, equipment and storage medium
CN113640712A (en) * 2021-08-06 2021-11-12 中国人民解放军海军工程大学 Prediction method for vertical component of vertical induced magnetic field of ship
CN113762610A (en) * 2021-08-30 2021-12-07 武汉理工大学 Method, system and equipment for predicting maximum development potential of channel bearing capacity
CN113971517A (en) * 2021-10-25 2022-01-25 中国计量大学 GA-LM-BP neural network-based water quality evaluation method
CN114462321A (en) * 2022-04-12 2022-05-10 武汉博赛环保能源科技有限公司 Neural network prediction method based on boiler combustion characteristics
CN115310025A (en) * 2022-10-12 2022-11-08 珠海翔翼航空技术有限公司 Airplane safe landing method and device based on fusion model
WO2022247049A1 (en) * 2021-05-24 2022-12-01 苏州大学 Method for predicting wind speed based on complex-valued forward neural network

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101706335A (en) * 2009-11-11 2010-05-12 华南理工大学 Wind power forecasting method based on genetic algorithm optimization BP neural network
CN108710288A (en) * 2018-04-19 2018-10-26 东华大学 The control method of the anti-drift of rotor craft hovering based on forecasting wind speed
CN109165764A (en) * 2018-06-26 2019-01-08 昆明理工大学 A kind of line loss calculation method of genetic algorithm optimization BP neural network

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101706335A (en) * 2009-11-11 2010-05-12 华南理工大学 Wind power forecasting method based on genetic algorithm optimization BP neural network
CN108710288A (en) * 2018-04-19 2018-10-26 东华大学 The control method of the anti-drift of rotor craft hovering based on forecasting wind speed
CN109165764A (en) * 2018-06-26 2019-01-08 昆明理工大学 A kind of line loss calculation method of genetic algorithm optimization BP neural network

Cited By (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111726349B (en) * 2020-06-16 2022-07-08 桂林电子科技大学 GRU parallel network flow abnormity detection method based on GA optimization
CN111726349A (en) * 2020-06-16 2020-09-29 桂林电子科技大学 GRU parallel network flow abnormity detection method based on GA optimization
CN111859807A (en) * 2020-07-23 2020-10-30 润电能源科学技术有限公司 Initial pressure optimizing method, device, equipment and storage medium for steam turbine
CN111898827A (en) * 2020-08-03 2020-11-06 西安石油大学 Motorcycle helmet wearing condition prediction evaluation model based on neural network
CN112289392A (en) * 2020-10-28 2021-01-29 广西贺州市桂东电子科技有限责任公司 Method for predicting performance of anode aluminum foil for aluminum electrolytic capacitor
CN112487726A (en) * 2020-12-21 2021-03-12 北京航天测控技术有限公司 Training method and device for target test model
CN112651176A (en) * 2020-12-24 2021-04-13 广西大学 GA-BP algorithm-based shielded door motor output force prediction method
CN112749792A (en) * 2021-02-02 2021-05-04 南京信息工程大学 Wind speed prediction method based on BP algorithm
CN112749792B (en) * 2021-02-02 2023-07-07 南京信息工程大学 BP algorithm-based wind speed prediction method
CN113095477A (en) * 2021-03-17 2021-07-09 西安理工大学 Wind power prediction method based on DE-BP neural network
CN113095477B (en) * 2021-03-17 2024-02-09 西安理工大学 Wind power prediction method based on DE-BP neural network
CN113483473A (en) * 2021-03-29 2021-10-08 南方环境有限公司 Welding workshop environment control method based on genetic-neural network (GA-BP) model
CN113159299B (en) * 2021-04-30 2024-02-06 杭州电子科技大学 Fractional order depth BP neural network optimization method based on extremum optimization
CN113159299A (en) * 2021-04-30 2021-07-23 杭州电子科技大学 Fractional order depth BP neural network optimization method based on extremum optimization
WO2022247049A1 (en) * 2021-05-24 2022-12-01 苏州大学 Method for predicting wind speed based on complex-valued forward neural network
CN113158586A (en) * 2021-05-26 2021-07-23 华能新能源股份有限公司 Wind power plant numerical simulation method and device combined with neural network algorithm and storage medium
CN113640712A (en) * 2021-08-06 2021-11-12 中国人民解放军海军工程大学 Prediction method for vertical component of vertical induced magnetic field of ship
CN113610297A (en) * 2021-08-06 2021-11-05 浙江工业大学之江学院 Air quality prediction method, device, equipment and storage medium
CN113640712B (en) * 2021-08-06 2023-11-24 中国人民解放军海军工程大学 Prediction method for vertical component of vertical induction magnetic field of ship
CN113762610A (en) * 2021-08-30 2021-12-07 武汉理工大学 Method, system and equipment for predicting maximum development potential of channel bearing capacity
CN113971517A (en) * 2021-10-25 2022-01-25 中国计量大学 GA-LM-BP neural network-based water quality evaluation method
CN114462321A (en) * 2022-04-12 2022-05-10 武汉博赛环保能源科技有限公司 Neural network prediction method based on boiler combustion characteristics
CN115310025A (en) * 2022-10-12 2022-11-08 珠海翔翼航空技术有限公司 Airplane safe landing method and device based on fusion model

Similar Documents

Publication Publication Date Title
CN111160520A (en) BP neural network wind speed prediction method based on genetic algorithm optimization
CN110070226B (en) Photovoltaic power prediction method and system based on convolutional neural network and meta-learning
CN109214575B (en) Ultrashort-term wind power prediction method based on small-wavelength short-term memory network
CN112949945B (en) Wind power ultra-short-term prediction method for improving bidirectional long-term and short-term memory network
CN106779151B (en) A kind of line of high-speed railway wind speed multi-point multi-layer coupling prediction method
CN115293415A (en) Multi-wind-farm short-term power prediction method considering time evolution and space correlation
CN105574615B (en) wavelet-BP neural network wind power prediction method based on spatial correlation and GA
CN104573879A (en) Photovoltaic power station output predicting method based on optimal similar day set
CN110751318A (en) IPSO-LSTM-based ultra-short-term power load prediction method
CN110070228B (en) BP neural network wind speed prediction method for neuron branch evolution
CN113141008B (en) Data-driven power distribution network distributed new energy consumption capability assessment method
CN107609774B (en) Photovoltaic power prediction method for optimizing wavelet neural network based on thought evolution algorithm
CN111027775A (en) Step hydropower station generating capacity prediction method based on long-term and short-term memory network
CN114358213B (en) Error ablation processing method, system and medium for nonlinear time series data prediction
CN112330951A (en) Method for realizing road network traffic data restoration based on generation of countermeasure network
CN113516271A (en) Wind power cluster power day-ahead prediction method based on space-time neural network
CN114169445A (en) Day-ahead photovoltaic power prediction method, device and system based on CAE and GAN hybrid network
CN112149905A (en) Photovoltaic power station short-term power prediction method based on wavelet transformation and wavelet neural network
CN115146842A (en) Multivariate time series trend prediction method and system based on deep learning
Shang et al. Research on intelligent pest prediction of based on improved artificial neural network
CN109408896B (en) Multi-element intelligent real-time monitoring method for anaerobic sewage treatment gas production
CN115271225A (en) Wind power-wind power modeling method based on wavelet denoising and neural network
CN113151842B (en) Method and device for determining conversion efficiency of wind-solar complementary water electrolysis hydrogen production
CN113762591A (en) Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy
CN111488974A (en) Deep learning neural network-based ocean wind energy downscaling 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