Research On Improved Quantum-Behaved Particle Swarm Optimization And Its Application

Particle swarm optimization(PSO)is a group intelligent optimization algorithm which has fewer setup parameters and easy to implement.It has been already widely used in function optimization,neural network training and fuzzy system control.However,the traditional PSO has the shortcomings of weak convergence and easy to fall into the local optimal value.Based on the principle of quantum mechanics,a proposed quantum-behaved particle swarm optimization(QPSO)algorithm solves the problem that the global convergence performance of PSO is not effective.Based on the QPSO algorithm,an adaptive quantum-behaved particle swarm optimization(AQPSO)algorithm is proposed.Then we proposed a new T-S fuzzy nueral network that use the AQPSO as training algorithm,and applied this fuzzy neural network to water quality evaluation.Finally,based on the parallel computing model of MapReduce,a parallel AQPSO algorithm for big data is proposed.The main research work of this paper is as follows:(1)Due to the contraction expansion coefficient ? in the standard quantum particle swarm algorithm cannot adapt to the problems in the complex and nonlinear search process in the iterative operation,and in the calculation of the average position of the individual do not take into account the excellent particles in the optimization process played an important guiding role,an adaptive quantum particle swarm optimization algorithm(AQPSO)is proposed,which improves the convergence performance of the algorithm by adding the concept of aggregation degree and central weight.(2)The traditional T-S fuzzy neural network is based on genetic algorithm and BP algorithm as the learning algorithm,but the required parameters of the genetic algorithm are too many,and the convergence speed of the BP algorithm is slow and easy to fall into the local optimal value.Aiming at this problem,AQPSO algorithm is proposed as T-S fuzzy neural network learning algorithm to improve the learning ability of neural network model.(3)In the background of big data,the AQPSO algorithm based on MapReduce parallel computing model is proposed for shortcomings of traditional serial intelligent optimization algorithm which cannot deal with high dimensional data.And the feasibility of the new algorithm is verified by the test of the reference function under high dimensional data.