傅里叶空间变换处理带噪声进化算法的研究

Research on evolutionary algorithm with noisy fitness based on Fourier space transform

鉴于进化算法处理实际优化问题时受到的噪声干扰，提出了一种新的数学去噪方法（FourierSpaceTransform，FST）。建立噪声环境下进化计算中新的适应函数计算模型；对该模型下计算所得的个体适应值进行傅氏空间变换，运用滤波方法处理；通过傅氏逆变换得到处理后的适应值，通过比较它们模值的大小，选出优秀个体。实验结果表明，FST方法不仅对噪声处理有很好的效果，而且计算代价低，稳定性好。

It is inevitable to meet noisy for Evolutionary Algorithms（EAs） when optimizing the practical problems.This pa- per proposes a new mathematic denoising method（Fourier Space Transform,FST）.A new computational model of fitness function in noisy environment is established.The noisy fitnesses of solutions are calculated from the new model and the filtering approach is used to deal with the noisy fitness after Fourier space transform.By the inverse FST, The disposed fitnesses are obtained whose module values are decided whether they are needed excellent solutions.The simulation experiment shows that the FST is not only efficiency but also low computational complexity and high stability.