Genetic algorithm approach to fixed-order mixed \(H_2/H_\infty\) optimal deconvolution filter designs (Q2734435)

The design procedure of a fixed-order mixed \( H_2 / H_{\infty} \) deconvolution filter is divided into two steps. In the first step, based on Jury's stability criterion, the stability domain of the coefficients of the denominator in a fixed-order deconvolution filter is specified. In the second step, the mixed \( H_2 / H_{\infty} \) optimal solution in the stability domain of the coefficients (parameter space) will be searched via the genetic algorithm. The simulation results indicate that the genetic-based design algorithms converge exponentially and the reconstruction performance is acceptable even if the order of the deconvolution filter is lower. The proposed design methods are suitable for lower-order optimal deconvolution filter design with the simplicity of implementation as well as saving of operation time and are useful for practical application in signal reconstruction problems with a high order channel and signal model.