Research And Application Of Fuzzy Cognitive Map Based On Memetic And Particle Swarm Algorithm

Fuzzy cognitive map is a kind of model of knowledge representation,reasoning and soft computation,which quantifies causality between nodes by introducing fuzzy measure into the cognitive map model.In recent decades,fuzzy cognitive map has been successfully applied in many fields such as reasoning,decision making and time series analysis.However,in the aspect of multivariable time series prediction,fuzzy cognitive map has yet to overcome the challenge of adjacency matrix quantization,which is faced with data purification,lack of prior knowledge,local extreme value and high computational complexity.Concerning the abovementioned challenges,this article develops research on fuzzy cognitive map based on memetic and particle swarm optimization algorithm,and applies it to the prediction field of atmospheric environmental monitoring time series data from the perspective of multi-discipline integration of fuzzy cognitive map and evolutionary learning,optimization theory and time series analysis.The specific research contents are as follows:(1)Aiming at the challenges of lack of prior knowledge,local extremum and high computational complexity faced by fuzzy cognitive map based on evolutionary learning algorithm,this article proposes a fuzzy cognitive map based on memetic and particle swarm learning algorithm.According to the framework of memetic algorithm based on population global search and individual local heuristic search,the particle swarm optimization algorithm is introduced to carry out global search,which effect ively solves the problems of lack of prior knowledge and high computational complexity.The simulated annealing algorithm is used to solve the local extremum problem.The experimental results show that the memetic and particle swarm learning algorithm proposed in this paper has the overall advantages of fast convergence,high accuracy and low computational complexity compared with the three typical learning algorithms.(2)Aiming at the challenges of sparse samples,interference among attributes and large errors in multivariable atmospheric environmental monitoring time series,this paper proposes a fuzzy cognitive map based atmospheric monitoring data prediction algorithm.Firstly,systematically comb and analyze the atmospheric environmental monitoring data,analyze the characteristics and correlation of the time series data,and obtained the purified data set.Then,based on the proposed fuzzy cognitive map based on memetic and particle swarm learning algorithm,a fuzzy cognitive map based atmospheric monitoring data prediction algorithm is constructed.Finally,the data predicted value output by fuzzy cognitive map iteration.Experimental results show that the proposed algorithm has overall advantages over the five typical time series prediction algorithms in the six monitoring data.