In order to further improve the performance of Infinite Impulse Response (IIR) digital filter, a design method of the IIR digital filter based on structure evolution and parameter evolution was proposed. Firstly, initial filter structure was got by using Genetic Algorithm (GA). Then, Differential Evolution (DE) was used to optimize the parameters of the filter. Finally, an improved optimization strategy was used to further optimize the parameters of the filter by using adjustment search-step and bidirectional heuristic search. Furthermore, the proposed method was applied to the design of low-pass filter and high-pass filter. Compared with the design method based on GA, the pass-band performance of low-pass filter based on the proposed method is not much different from that of the previous algorithm, however, the transition zone width of it is reduced by 65%, the minimum stop-band attenuation of it was reduced by 36.48 dB; the pass-band ripple of high-pass filter based on the proposed method is reduced by 75%, the transition zone width of it is reduced by 44%, and the minimum stop-band attenuation of it is reduced by 12.13 dB. Simulation results show that the proposed method can get effective filters with better performance, therefore it is suitable for the IIR digital filter design.

Focused on the issue that the transfer function of Infinite Impulse Response (IIR) digital filters is not optimal in the entire design process adopting traditional filter design methods, a structure evolution based design method for IIR digital filters using Genetic Algorithm (GA) was proposed. The method evolved the filter structure directly without the preparation of the transfer function. Firstly, the Structure Generation Instruction Sequences (SGIS) were generated randomly. Those SGIS not only controlled the process of structure generation but also represented those structures. Then, the SGIS were coded and seemed as chromosomes. Finally, GA was used to optimize those chromosomes to obtain a best filer. In the comparison experiments with the traditional coefficient evolution based design method for IIR digital filters using GA, the pass-band ripple of the proposed algorithm decreased by 40.58%, the transition zone width of it decreased by 87.62%, and the minimum stop-band attenuation of it declined by 9.22%.