圆柱-球体三自由度超声电机神经网络控制策略仿真

Simulation of Neural Network Control Strategy for Cylinder-sphere 3-DOF Ultrasonic Motor

针对圆柱-球体三自由度超声电机难以建立精确数学模型的特点,设计了神经网络控制系统。该控制系统采用包含输入层、中间层和输出层3层网络的BP神经网络控制器。应用该控制器,实现了电机精确定位控制。结果表明,神经网络控制器的性能稳定,克服了常规控制存在的不足,得到了满意的控制效果。

A neural network control system was designed for cylinder-sphere 3-dof ultrasonic motors which are difficult to be controlled by exact mathematic models. The control system used a 3-layer BP neural network controller which consists of an input layer, an immediate layer and an output layer. The precise orientation and control of motors was realized by using this controller. The results showed that the neural network controller has overcome the insufficiency of normal control and present a steady and satisfied performance.