基于遗传算法和BP算法的电磁重构应用研究

Combining the BP and the Genetic Algorithm to Reconstruct the Electromagnetic Data

在电磁重构问题中,将BP神经网络算法中最速下降的思想与GA结合,构造BP算子.利用GA的杂交、变异选择算子在全变量空间大概率搜索全局解,在解点附近用BP算子快速搜索收敛,提高搜索性能.应用混合算法重构分层生物组织各层电导率和厚度.数值计算结果表明,改进后的算法在搜索速度和精确度上明显提高,并具有较高的抗噪性能.

In electromagnetic reconstruction, in order to improve both the global and local optimum capacity, combining the genetic algorithm with the BP operator enlightened by steepest descent algorithm in BP neutral network to reconstruct the conductivity distribution of stratified biological tissue. The experimental result shows that the improved algorithm largely enhances the performance and presents the higher noiseresistance.