Improvements And Applications Of Arithmetic Optimization Algorithm

With the development of society,it is gradually becoming a trend to use meta-heuristic algorithms to solve optimization problems.The Arithmetic Optimization Algorithm,as a recently proposed meta-heuristic algorithm,was designed inspired by the basic operators,i.e.,addition,subtraction,multiplication and division.The Arithmetic Optimization Algorithm has the advantages of simple principle,few parameters and easy operation,and can be used to solve various kinds of optimization problems,but there are similar problems with other meta-heuristic algorithms,such as easy to fall into the local optima and slow convergence speed.Therefore,in order to improve the optimization performance of the algorithm and expand the application fields,this thesis addresses the improvements of the Arithmetic Optimization Algorithm and uses it to solve problems in different fields,i.e.,typical optimization problems,feature selection problem and state of health of lithium battery estimation problem.The related research is divided into three parts as follows.(1)An improved Arithmetic Optimization Algorithm for solving typical optimization problems.The Arithmetic Optimization Algorithm is improved and a multi-strategy Arithmetic Optimization Algorithm is proposed.Firstly,a sinusoidal chaotic strategy is used to generate the initial population in the initializing stage to achieve the purpose of improving the population diversity.Secondly,the S-shaped transfer function is used in the calculating stage to map the values from continuous search space to discrete search space,and then dynamic weight is used to balance exploration and exploitation.Then,in the updating stage,a mutation strategy is used to enhance the algorithm’s optimization-seeking ability and avoid falling into the local optima.Finally,the improved algorithm is compared with other meta-heuristic algorithms,and the experimental results show that the improved algorithm has better optimization capability than the compared meta-heuristic algorithms.(2)An improved Arithmetic Optimization Algorithm for solving feature selection problem.The Arithmetic Optimization Algorithm is improved and the binary Arithmetic Optimization Algorithm is proposed.Firstly,six algorithms are formed by converting the continuous search space into a discrete search space based on six different transfer functions.Secondly,by combining the transfer functions and Lévy flight strategy,six other algorithms are developed to improve the search speed and the ability to jump out of the local optima.Finally,with respect to various evaluation metrics,the performance of the proposed algorithms is evaluated based on 20 common datasets from the machine learning library,and the proposed algorithms are compared with other meta-heuristic algorithms,and the results show that the proposed algorithms outperform other meta-heuristic algorithms in solving feature selection problem.(3)An improved Arithmetic Optimization Algorithm is applied to solve the state of health of lithium battery estimation problem.Firstly,the improved Arithmetic Optimization Algorithm is proposed by introducing an adaptive strategy in the algorithm.Secondly,the improved algorithm is used to optimize the weights and thresholds of the Back Propagation neural network to form a new method to estimate the health state of lithium battery.Then,three health features are extracted as the inputs of the Back Propagation neural network by analyzing the Pearson correlation coefficients.Then,to evaluate the optimization ability of the improved algorithm,several other metaheuristic algorithms are also used to optimize the weights and thresholds of the Back Propagation neural network,and the results show that the improved Arithmetic Optimization Algorithm is more suitable for optimizing the Back Propagation neural network.Finally,the method is compared with other methods,and the estimation results show that the proposed method is able to estimate the health state of lithium batteries with the higher accuracy.This thesis first analyzes the shortcomings of the Arithmetic Optimization Algorithm and improves the algorithm for typical optimization problems,feature selection problem and state of health of lithium battery estimation problem,respectively.The experimental results show that the improved Arithmetic Optimization Algorithm has better performance and can be used to solve problems in several fields,such as grid optimization,energy forecasting,transportation scheduling and other optimization problems.