基于RBF神经网络逆系统的注射速度控制

为了实现注射速度的精确控制，针对其非线性时变的动态特性，提出了基于神经网络逆系统的控制方法.采用M.Rafizadeh模型描述注射速度系统特性，通过求解该系统的相对阶证明了系统的可逆性.由于注射速度系统逆模型的解析形式难以获得，因此构造了基于RBF神经网络的注射速度逆系统，并将该系统与常规PID控制相结合，对注射速度实现复合控制，解决了基于RBF神经网络逆系统的开环控制效果不理想的问题.仿真实验表明，该控制系统具有良好的跟踪性能及抗干扰性能.

Injection Speed Control Based on RBF Neural Network Inverse System

Taking account of the nonlinear time-variant characteristics of injection speed, a control method based on a neural network inverse system was proposed to control injection speed precisely. The injection speed system was characterized by the M. Rafizadeh model, and its reversibility was confirmed through calculating the relative degree of the system. Considering the difficulty in obtaining the analytical form of reversible model for injection speed system, the RBF neural network (RBFNN) was used to build an inverse system of injection speed. The RBFNN-based inverse system, which has unsatisfied control effect due to open-loop control, was combined with the conventional PID control, thus realizing the composite control of injection speed. The simulation results show that the control system proposed has good tracking and anti-disturbance performance.