基于改进遗传算法的六自由度并联平台动力学参数辨识OA

To improve the problems of premature convergence and high proportion of invalid solutions in traditional genetic algorithms for dynamic parameter identification of six degree-of-freedom parallel platforms,this study propo-ses an improved genetic algorithm incorporating dynamic priors.The algorithm first generates an initial population based on chaotic sequences,stratifies parameter sensitivity using the mRMR criterion,and filters valid solutions by combining parameter boundaries with physical constraints.Subpopulations are then divided based on cosine similar-ity,and collaborative evolution is achieved through a cyclic-chain dynamic competition mechanism,coupling-aware adaptive crossover,Cauchy mutation,and elitism.Finally,optimal identification parameters are obtained through iterative closed-loop optimization.Experimental results show that the algorithm reduces the average parameter error from 17.59%in traditional genetic algorithms to 3.69%,with all parameter errors controlled below 6.83%.Moreo-ver,the convergence time is improved from 0.9 s to about 0.03 s,ensuring rapid stabilization and a significantly improved convergence rate,thereby providing reliable parameter support for high-precision platform control.