【摘要】 针对遗传算法应用于时变时滞和参数在线辨识时 ,存在无法兼顾收敛速度与辨识精度的缺点 ,提出一种遗传算法和单神经元有机结合的混合遗传算法。利用GA的全局最优性在整个空间搜索可能的极值 ,而用单神经元的误差梯度下降特性在极值点附近快速搜索 ,从而达到全局最优与快速搜索的有机结合 ,提高了收敛速度和辨识精度。并对混合遗传算法进行了改进 ,使之更适用于在线辨识。仿真结果表明 ,改进的混合遗传算法用于在线辨识有效且实用。更多还原

【Abstract】 To solve the contradiction between the convergent speed and identifying accuracy at the application of genetic algorithm (GA) to on-line system identification, an improved hybrid algorithm, in which neuron’s local fast search and GA’s global convergent characteristic are combined perfectly, is proposed. In particular, the global convergent feature of the GA is used to find the possible extremums in the whole area, and the neuron with the feature of error descend in the direction of grads is used to fast search about the extremums. Thus, the better combination of global convergence and fast search is obtained. Besides, the accuracy of the on-line identification is greatly enhanced. Moreover, several improved measures are taken to meet the command of the on-line identification. Simulation results show that the improved hybrid algorithm is efficient and useable for on-line system identification.更多还原