基于改进遗传算法的反卷积信号识别

信号处理中的反卷积是一个不适定问题，在泛涵理论上求取反卷积正则解的关键足求距离的最小。遗传算法在优化方面具有优势，因此提出用遗传算法优化求取最小值进行反卷积信号诊断。但足由于传统的遗传算法存在着一些问题，易陷入局部极小点导致成熟前收敛，使得反卷积问题的解决有误差，恢复的波形具有波动性，精度还不够，由此我们对传统的遗传算法进行了改进，改进后模拟计算发现恢复的信号波形精度明显上升，和原信号波形很相象，比较准确地反映了原信号固有的特性。

Deconvolution Signal Indentification Based on Improved Genetic Algorithm

Generally, in signal processing, deconvolution is an improperly posed problem. From the opinion of functional analysis, the key to get the regular solution of deconvolution is to calculate out the minimum of the functional distance. In the paper a method of deconvolution is proposed by using genetic algorithm to calculate the minimum of the functional distance. Simple genetic algorithm is easy to converge early but usually lead to partial optimum. To solve deconvolution by using traditional genetic algorithm generally shows errors scattered over the exact solution. Another important work is to improve traditional genetic algorithm to make it fit for the calculation of deconvolution. Numerical example indicates that more accurate solution can be reached by using the improved genetic algorithm developed in the paper, and the recovered signal curve has quite similarity with the original signal curve.