Design Of Active Noise Reduction Algorithm Based On Gene Expression Programming

With the development of Industrial society and the improvement of People’s living standard,active noise reduction headphones have gradually become an indispensable daily necessities.However,during the production of active noise reduction headphones,the initial parameters of the IIR Digital filter in the headphones are burned to the fixed filter parameters,and each pair of headphones can not achieve the best noise reduction effect.Therefore,it is necessary to optimize the IIR Digital filter coefficient in the detection process of the active noise reduction headphones,so that the headphones can have the maximum active noise reduction effect.Gene Expression Programming(GEP)is a new evolutionary Algorithm which inherits and develops the advantages of Genetic Algorithm(GA)and Genetic Programming(GP)based on the theory of biological Gene Expression,it has powerful computing power,but it is seldom used in IIR Digital filter optimization and active noise reduction.In this paper,based on gene expression programming algorithm,the optimization method of IIR Digital filter coefficients is studied in depth.Firstly,the active noise reduction structure of headphone feed-forward is analyzed,and the error function of the actual frequency response and the expected frequency response of the IIR Digital filter is derived from the formula as the fitness function of the GEP algorithm,the constraint range of the coefficient was determined according to its stability condition,and gene chromosomes were designed according to the characteristics of IIR Digital filter coefficient.Secondly,in order to apply the GEP algorithm to the optimization of IIR Digital filter coefficients,the standard GEP algorithm is improved.In order to improve the performance of GEP algorithm,convex transformation operator and inner recombination operator are added in this paper.In addition,the probability adaptation of genetic operators is proposed to solve the problem of slow late evolution of population.Finally,the experimental platform is built to carry out relevant experiments on the proposed algorithm.In the actual application process,the proposed GEP algorithm has a good noise reduction effect and consistency,the noise reduction in most frequency bands reaches 20～30d B,some frequency bands can exceed 30 d B,which proves that the improved GEP algorithm proposed in this paper has better performance than the classical GEP algorithm.Compared with other genetic algorithms,GEP has good efficiency and stability in active noise reduction.