Research On Parallel Interaction Learning Of Artificial Bee Colony Algorithm And BP Neural Network

In the life and production process of human beings,what is often encountered is the optimization problem.Scholars have also studied many methods to solve optimization problems.As an important branch,swarm intelligence optimization algorithm provides a new idea for solving optimization problems.Artificial bee colony(ABC)algorithm is a novel swarm intelligence algorithm,it's a natural calculation method proposed by studying the information exchange and honey collecting behavior of bees in nature.It has a simple algorithm design process,few parameters,and can jump out of the local optimal solution easily,and don't need to calculate the gradient and other characteristics.However,in the real world,there are more discrete optimization problems,so it's urgent to study the discretization of ABC algorithm.BP neural network algorithm(BPNN)is also a method to solve the optimization problems.Since it was proposed,many experts and scholars at home and abroad have carried out a lot of improvements and research on it,and also achieved a lot of achievements.However,the initial weights and thresholds of the traditional BP neural network are random,which leads to inefficient search efficiency,premature convergence and inadequate convergence accuracy.Therefore,how to optimize the defects of BP neural network has become an important research direction.Based on the above problems,this thesis mainly studies the improvement of discrete artificial bee colony algorithm and its parallel integration learning with BP neural network algorithm.The research results are as follows: Firstly,a discrete binary artificial bee colony algorithm(IBABC)using Gauss variation function as probability mapping function is constructed,and proved the effectiveness of the algorithm theoretically,the experiments on Benchmark test functions also prove IBABC algorithm has better effect than other improved schemes.Secondly,the improved binary artificial bee colony algorithm(IBABC)is used to train BP neural network,a parallel integration learning algorithm based on improved binary artificial bee colony and BP neural network is proposed,namely IBABC-BP algorithm.The theoretical analysis and experimental results prove the effectiveness of the parallel integrated learning algorithm.Finally,a haze prediction model which based on IBABC-BP parallel integration learning algorithm is proposed,and then do the comparative experiments with haze data,the results prove the effectiveness of the haze prediction model proposed in this thesis.