基于改进反向差分进化算法的多股簧响应模型参数辨识

针对传统算法在辨识多股簧响应模型参数方面存在的不足，提出一种改进反向差分进化算法。改进的算法采用新的反向学习机制引导种群逼近全局最优解，并使用正弦混沌序列计算缩放因子以提高种群的多样性。这两种机制协同操作可以较好地平衡算法的全局勘探和局部开采能力。通过建立参数辨识的目标函数和进行多股簧的动态试验，然后使用改进算法优化目标函数得到辨识结果。计算结果表明，改进算法能够有效地辨识多股簧模型参数，参数的收敛速度和计算的成功率优于标准反向差分进化算法和其他算法；即使在噪声级别较高的情况下，改进反向差分进化算法也可以准确地求出多股簧的模型参数。
Abstract：Aiming at shortcomings of traditional algorithms in identifying the response model parameters of a stranded wire helical spring, an improved reverse difference evolution algorithm was proposed.A novel reverse learning strategy was used to guide populations approaching the global optimal solution.In order to improve populations’ diversity, the sinusoidal chaotic sequence was used to calculate the scaling factor.These two mechanisms cooperatively operated to better balance the global exploration ability and the local mining one of the proposed algorithm.Through establishing the objective function of parametric identification and conducting dynamic tests of a stranded wire helical spring, the improved algorithm was used to optimize the objective function, and obtain the model parameters.The results showed that the proposed algorithm can effectively recognize the response model parameters of the stranded wire helical spring; the parameters’ convergence speed and the success rate of computation are superior to those of the standard reverse difference evolution algorithm and other algorithms; even in the case of higher noise level, the proposed algorithm can correctly recognize model parameters of stranded wire helical springs.