基于改进遗传算法优化的RBFNN防摆控制研究

Research on anti-swing control of RBFNN optimized by improved genetic algorithm

针对起重机吊运过程中吊重摇摆以及其数学模型高阶非线性的问题,提出利用改进遗传算法优化径向基函数神经网络(RBFNN)监督控制方法对起重机进行防摇摆和定位控制。采用拉格朗日方程建立起重机的数学模型,在传统PD反馈控制的基础上,设计了RBFNN摆角和位移监督控制器,利用RBFNN强大的自学习能力对PD控制器的输出进行在线学习并逐步取代,实现监督控制。采用改进的遗传算法对RBFNN的参数进行全局优化,摆脱了局部极值的困扰。实验结果表明,该方法能够实现在起重机精确定位的同时快速消除摆动,与模糊PID控制和模糊神经网络控制相比,获得了更好的控制效果,证明了该方法的有效性。

Aiming at the problem of the swaying of the lifting weight and the high order nonlinear of its mathematical model in the process of lifting, an improved genetic algorithm optimized radial basis neural network(RBFNN) supervisory control method was proposed to control the swaying and positioning of the crane. The Lagrange equation was used to establish the mathematical model of the controlled object. Based on the traditional PD feedback control, the RBFNN supervised controller for pendulum angle and displacement was designed. The powerful self-learning ability of RBFNN was utilized to carry out online learning and gradually replace the output of PD controller to realize the supervised control. The parameters of RBFNN are optimized globally by improved genetic algorithm, and the trouble of local extremum is eliminated. The experimental results illustrate that the proposed method can achieve pinpoint positioning of the hoist and quickly prevent the swing. Compared with fuzzy PID control and fuzzy neural network control, better control effect is obtained, which proves the validity of the proposed method.