CN109002878A - A kind of GA Optimized BP Neural Network - Google Patents

A kind of GA Optimized BP Neural Network Download PDF

Info

Publication number
CN109002878A
CN109002878A CN201810727710.2A CN201810727710A CN109002878A CN 109002878 A CN109002878 A CN 109002878A CN 201810727710 A CN201810727710 A CN 201810727710A CN 109002878 A CN109002878 A CN 109002878A
Authority
CN
China
Prior art keywords
neural network
individual
fitness
chromosome
formula
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
CN201810727710.2A
Other languages
Chinese (zh)
Inventor
俞阿龙
诸飞
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.)
Huaiyin Normal University
Original Assignee
Huaiyin Normal University
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 Huaiyin Normal University filed Critical Huaiyin Normal University
Priority to CN201810727710.2A priority Critical patent/CN109002878A/en
Publication of CN109002878A publication Critical patent/CN109002878A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/12Computing arrangements based on biological models using genetic models
    • G06N3/126Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Genetics & Genomics (AREA)
  • Physiology (AREA)
  • Feedback Control In General (AREA)

Abstract

A kind of GA Optimized BP Neural Network, it is related to a kind of BP neural network technical field.It is comprised the following steps: step 1, initialization population select suitable coding mode;The selection of step 2, fitness function;Step 3, selection operation use roulette method;Step 4, crossover operation;Step 5, mutation operation;New chromosome is replaced original chromosome by step 6, is calculated fitness and is otherwise gone to step 3 if meeting condition jumps to step 7 and continue to optimize;Best initial weights and threshold values are assigned to BP neural network and are used to train by step 7, the error amount until reaching setting;Step 8, the evaluation that trained BP neural network is used for plant effuent matter, obtain evaluation result.After adopting the above technical scheme, the invention has the following beneficial effects: can effectively solve that BP network easily sinks into the slow even not convergence problem of local minimum, convergence rate, and hidden layer neuron number can be reasonably selected.

Description

A kind of GA Optimized BP Neural Network
Technical field
The present invention relates to BP neural network technical fields, and in particular to a kind of GA Optimized BP Neural Network.
Background technique
Although BP neural network has very strong nonlinearity mapping ability and simple network structure.But BP neural network is also deposited In many disadvantages: hidden layer neuron number is difficult to determine, increases calculation amount too much, influence convergence rate, is difficult to very little accurately Prediction;Algorithm the convergence speed is slow;It is easily trapped into locally optimal solution.
The natural selection of genetic algorithm (Genetic Algorithm) genetic manipulation and " survival of the fittest " is come guidance learning With determine the direction of search, the topological structure and its weight and threshold values of network can be optimized simultaneously during Optimized BP Neural Network, Select network model according to sample knowledge.And change with the complexity of problem, the dynamic of BP network may be implemented Adaptivity.GA has of overall importance, concurrency, rapidity, well adapting to property and robustness, can effectively solve that BP network is easy Sink into the slow even not convergence problem of local minimum, convergence rate, and hidden layer neuron number can be reasonably selected.
Summary of the invention
In view of the defects and deficiencies of the prior art, the present invention intends to provide a kind of GA Optimized BP Neural Network, energy Effective solution BP network easily sinks into the slow even not convergence problem of local minimum, convergence rate, and can reasonably select hidden layer mind Through first number.
To achieve the above object, the present invention is using following technical scheme: it is comprised the following steps:
Step 1, initialization population select suitable coding mode;
The selection of step 2, fitness function obtains the initial weight and threshold values of BP neural network according to individual, with training Data training BP neural network after forecasting system export, using prediction output desired output between Error Absolute Value and E as Ideal adaptation angle value F, calculation formula are as follows:
In formula, n is network output node number, and yi is the desired output of i-th of node of BP neural network, and oi is i-th of section The prediction output of point, k is coefficient;
Step 3, selection operation use roulette method, that is, are based on fitness ratio strategy, the select probability Pi of each individual i Are as follows:
fi=k/Fi (2)
In formula, Fi is the fitness of individual i, since fitness value is the smaller the better, so to fitness before individual choice It is worth inverted, N is population invariable number, and k is coefficient;
Step 4, crossover operation, since individual uses real coding, so crossover operation method uses real number interior extrapolation method, the K chromosome ak and first of chromosome al are as follows in j crossover operations:
In formula, b is the random number between [0,1];
Step 5, mutation operation, j-th of gene aij for choosing i-th of individual make a variation, and mutation operation method is as follows:
In formula, amax is the upper bound of gene aij, and amin is the lower bound of gene aij, and f (g)=r2 (1-g/Gmax) 2, r2 is One random number, g are current iteration number, and Gmax is maximum evolution number, random number of the r between [0,1];
New chromosome is replaced original chromosome by step 6, fitness is calculated, if meeting condition jumps to step 7, otherwise Step 3 is gone to continue to optimize;
Best initial weights and threshold values are assigned to BP neural network and are used to train by step 7, the error amount until reaching setting;
Step 8, the evaluation that trained BP neural network is used for plant effuent matter, obtain evaluation result.
The coding mode uses real coding.
The working principle of the invention: being searched in global scope, optimized first the weight and threshold values of BP using GA, then will Best initial weights and threshold values are assigned to BP and are trained, and training reaches the error of setting, and training is completed, and carry out simulation and prediction.
After adopting the above technical scheme, the invention has the following beneficial effects: can effectively solve BP network easily sink into local minimum, The slow even not convergence problem of convergence rate, and hidden layer neuron number can be reasonably selected.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this Some embodiments of invention without any creative labor, may be used also for those of ordinary skill in the art To obtain other drawings based on these drawings.
Fig. 1 is schematic process flow diagram of the invention.
Specific embodiment
Referring to shown in Fig. 1, present embodiment the technical solution adopted is that: it is comprised the following steps:
Step 1, initialization population select suitable coding mode;
The selection of step 2, fitness function obtains the initial weight and threshold values of BP neural network according to individual, with training Data training BP neural network after forecasting system export, using prediction output desired output between Error Absolute Value and E as Ideal adaptation angle value F, calculation formula are as follows:
In formula, n is network output node number, and yi is the desired output of i-th of node of BP neural network, and oi is i-th of section The prediction output of point, k is coefficient;
Step 3, selection operation use roulette method, that is, are based on fitness ratio strategy, the select probability Pi of each individual i Are as follows:
fi=k/Fi (7)
In formula, Fi is the fitness of individual i, since fitness value is the smaller the better, so to fitness before individual choice It is worth inverted, N is population invariable number, and k is coefficient;
Step 4, crossover operation, since individual uses real coding, so crossover operation method uses real number interior extrapolation method, the K chromosome ak and first of chromosome al are as follows in j crossover operations:
In formula, b is the random number between [0,1];
Step 5, mutation operation, j-th of gene aij for choosing i-th of individual make a variation, and mutation operation method is as follows:
In formula, amax is the upper bound of gene aij, and amin is the lower bound of gene aij, and f (g)=r2 (1-g/Gmax) 2, r2 is One random number, g are current iteration number, and Gmax is maximum evolution number, random number of the r between [0,1];
New chromosome is replaced original chromosome by step 6, fitness is calculated, if meeting condition jumps to step 7, otherwise Step 3 is gone to continue to optimize;
Best initial weights and threshold values are assigned to BP neural network and are used to train by step 7, the error amount until reaching setting;
Step 8, the evaluation that trained BP neural network is used for plant effuent matter, obtain evaluation result.
The coding mode uses real coding.For the particularity of trade effluent, convenient for the genetic search of larger space, Genetic algorithm required precision is improved, the complexity of coding and decoding is avoided.
The working principle of the invention: being searched in global scope, optimized first the weight and threshold values of BP using GA, then will Best initial weights and threshold values are assigned to BP and are trained, and training reaches the error of setting, and training is completed, and carry out simulation and prediction.
After adopting the above technical scheme, the invention has the following beneficial effects: can effectively solve BP network easily sink into local minimum, The slow even not convergence problem of convergence rate, and hidden layer neuron number can be reasonably selected.
The above is only used to illustrate the technical scheme of the present invention and not to limit it, and those of ordinary skill in the art are to this hair The other modifications or equivalent replacement that bright technical solution is made, as long as it does not depart from the spirit and scope of the technical scheme of the present invention, It is intended to be within the scope of the claims of the invention.

Claims (2)

1. a kind of GA Optimized BP Neural Network, it is characterised in that it is comprised the following steps:
Step 1, initialization population select suitable coding mode;
The selection of step 2, fitness function obtains the initial weight and threshold values of BP neural network according to individual, uses training data Forecasting system exports after training BP neural network, using the Error Absolute Value and E predicted between output and desired output as individual Fitness value F, calculation formula are as follows:
In formula, n is network output node number, and yi is the desired output of i-th of node of BP neural network, and oi is i-th of node Prediction output, k is coefficient;
Step 3, selection operation use roulette method, that is, are based on fitness ratio strategy, the select probability Pi of each individual i are as follows:
fi=k/Fi (2)
In formula, Fi is the fitness of individual i, since fitness value is the smaller the better, so taking before individual choice to fitness value Inverse, N are population invariable number, and k is coefficient;
Step 4, crossover operation, since individual uses real coding, so crossover operation method use real number interior extrapolation method, k-th Chromosome ak and first of chromosome al are as follows in j crossover operations:
In formula, b is the random number between [0,1];
Step 5, mutation operation, j-th of gene aij for choosing i-th of individual make a variation, and mutation operation method is as follows:
In formula, amax is the upper bound of gene aij, and amin is the lower bound of gene aij, and f (g)=r2 (1-g/Gmax) 2, r2 is one Random number, g are current iteration number, and Gmax is maximum evolution number, random number of the r between [0,1];
New chromosome is replaced original chromosome by step 6, is calculated fitness and is otherwise gone to if meeting condition jumps to step 7 Step 3 continues to optimize;
Best initial weights and threshold values are assigned to BP neural network and are used to train by step 7, the error amount until reaching setting;
Step 8, the evaluation that trained BP neural network is used for plant effuent matter, obtain evaluation result.
2. a kind of GA Optimized BP Neural Network according to claim 1, it is characterised in that: the coding mode uses real number Coding.
CN201810727710.2A 2018-07-05 2018-07-05 A kind of GA Optimized BP Neural Network Pending CN109002878A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810727710.2A CN109002878A (en) 2018-07-05 2018-07-05 A kind of GA Optimized BP Neural Network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810727710.2A CN109002878A (en) 2018-07-05 2018-07-05 A kind of GA Optimized BP Neural Network

Publications (1)

Publication Number Publication Date
CN109002878A true CN109002878A (en) 2018-12-14

Family

ID=64598683

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810727710.2A Pending CN109002878A (en) 2018-07-05 2018-07-05 A kind of GA Optimized BP Neural Network

Country Status (1)

Country Link
CN (1) CN109002878A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109726817A (en) * 2018-12-21 2019-05-07 河北工业大学 The WPT system impedance matching methods of genetic algorithm optimization BP neural network
CN110084354A (en) * 2019-04-09 2019-08-02 浙江工业大学 A method of based on genetic algorithm training ANN Control game role behavior
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

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103699873A (en) * 2013-09-22 2014-04-02 杭州电子科技大学 Lower-limb flat ground walking gait recognition method based on GA-BP (Genetic Algorithm-Back Propagation) neural network
CN104820977A (en) * 2015-05-22 2015-08-05 无锡职业技术学院 BP neural network image restoration algorithm based on self-adaption genetic algorithm
CN106482938A (en) * 2016-10-14 2017-03-08 温州大学 Brake fluid system multi-source fusion fault predicting method based on GA BP network

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103699873A (en) * 2013-09-22 2014-04-02 杭州电子科技大学 Lower-limb flat ground walking gait recognition method based on GA-BP (Genetic Algorithm-Back Propagation) neural network
CN104820977A (en) * 2015-05-22 2015-08-05 无锡职业技术学院 BP neural network image restoration algorithm based on self-adaption genetic algorithm
CN106482938A (en) * 2016-10-14 2017-03-08 温州大学 Brake fluid system multi-source fusion fault predicting method based on GA BP network

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
刘浩然 等: "一种基于改进遗传算法的神经网络优化算法研究", 《仪器仪表学报》 *
周德红 等: "基于GA-BP神经网络的化工园区应急救援能力", 《安全与环境工程》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109726817A (en) * 2018-12-21 2019-05-07 河北工业大学 The WPT system impedance matching methods of genetic algorithm optimization BP neural network
CN110084354A (en) * 2019-04-09 2019-08-02 浙江工业大学 A method of based on genetic algorithm training ANN Control game role behavior
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

Similar Documents

Publication Publication Date Title
CN108764540B (en) Water supply network pressure prediction method based on parallel LSTM series DNN
CN103226741B (en) Public supply mains tube explosion prediction method
CN106600059A (en) Intelligent power grid short-term load predication method based on improved RBF neural network
CN109615011A (en) A kind of middle and small river short time flood forecast method based on LSTM
CN109146121A (en) The power predicating method stopped in the case of limited production based on PSO-BP model
CN104636801A (en) Transmission line audible noise prediction method based on BP neural network optimization
CN108920812B (en) Machining surface roughness prediction method
CN110516831A (en) A kind of short-term load forecasting method based on MWOA algorithm optimization SVM
CN108416366A (en) A kind of power-system short-term load forecasting method of the weighting LS-SVM based on Meteorological Index
CN102622418B (en) Prediction device and equipment based on BP (Back Propagation) nerve network
CN103105246A (en) Greenhouse environment forecasting feedback method of back propagation (BP) neural network based on improvement of genetic algorithm
CN104636985A (en) Method for predicting radio disturbance of electric transmission line by using improved BP (back propagation) neural network
CN110751318A (en) IPSO-LSTM-based ultra-short-term power load prediction method
CN104978611A (en) Neural network photovoltaic power generation output prediction method based on grey correlation analysis
CN109002878A (en) A kind of GA Optimized BP Neural Network
CN109034223A (en) A kind of Study on prediction technology of chaotic series and system
CN104504442A (en) Neural network optimization method
Ning et al. GA-BP air quality evaluation method based on fuzzy theory.
CN105224987A (en) A kind of change strategy colony global optimization method based on dynamic Lipschitz Lower Bound Estimation
CN109583588A (en) A kind of short-term wind speed forecasting method and system
CN108229750A (en) A kind of stock yield Forecasting Methodology
CN115049292B (en) Intelligent single reservoir flood control scheduling method based on DQN deep reinforcement learning algorithm
CN114580762A (en) Hydrological forecast error correction method based on XGboost
Hu et al. Radial basis function neural network with particle swarm optimization algorithms for regional logistics demand prediction
CN109408896B (en) Multi-element intelligent real-time monitoring method for anaerobic sewage treatment gas production

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20181214

RJ01 Rejection of invention patent application after publication