欠驱动桥式吊车自适应PID控制

Adaptive PID Control for an Underactuated Overhead Crane System

桥式吊车是一类非线性、强耦合、多变量的复杂欠驱动机械系统,其控制器设计具有一定的难度;针对桥式吊车的防摇定位控制问题,提出了一种自适应PID控制策略,先用RBF神经网络在线辨识被控对象的Jacobian信息,然后用学习速率更快的Levenberg-Marquardt算法代替传统的梯度下降法整定PID控制器参数,很好地实现了桥式吊车的防摇定位控制;文中所提出的控制方法具有结构简单,容易实现的特点;最后通过三组仿真实验证实了这种自适应PID控制器具有良好的控制效果。

Overhead crane is essentially a kind of nonlinear, strong coupling and multivariable complex underaetuated mechanical system, and it is a little difficult to design a controller for an overhead crane. We proposed an adaptive PID controller to solve the problem of the anti --swing and positioning control for an overhead crane. The RBF neural network was used to perform the on--line identification of the controlled plant＇ s Jacobian information. The Levenberg--Marquardt algorithm with a faster learning rate was used to tune the parameters of the PID controller instead of the conventional gradient descent tuning method. The proposed controller had a simple structure and was easy to implement. The simulation results demonstrated the effectiveness of the proposed control algorithm.