钢丝螺套装配神经网络自适应扭矩跟踪控制系统研究

Research on the Neural Network Adaptive Torque Tracking Control System for Wire Thread Insert Tightening

钢丝螺套装配过程是一个典型的非线性和动态过程,传统的装配控制方法难以保证自动装配质量的可靠性和一致性。系统的轻微扰动即会使装配力矩产生大幅波动,从而引起跳扣、卡滞等。针对以上问题,提出一种基于自适应神经网络的扭矩跟踪控制算法。该算法利用动态神经网络对装配系统进行离线辨识,解决被控制对象变化引起的失稳问题。在装配过程中,动态神经网络通过在线学习,进一步提高辨识精度;同时利用RBF神经网络调节装配的速度,实现钢丝螺套装配力矩的在线自适应控制。动态神经网络和RBF神经网络协同进行在线系统辨识和实时控制,保证控制系统的控制参数最优化。最终针对钢丝螺套装配过程的近似模型进行仿真实验,结果表明,所提出的神经网络自适应控制系统具有良好的自适应、自学习能力。该方法能够有效提高钢丝螺套装配的质量和稳定性。

The tightening of wire thread insert is a severely nonlinear and dynamic process.It is difficult for conventional control methods to ensure the reliability and consistency of equipment quality.The slight disturbances will cause large fluctuations of assembly torque,resulting in tripping,jamming,etc.To solve these problems,this paper proposed an adaptive neural network torque tracking controller.The dynamic neural network was firstly used to identify the assembly system offline to solve the problem of instability caused by the change of plant.Then,during the assembly process,the dynamic neural network was used to further improve the identification accuracy through online learning.At the same time,the RBF neural network was used to adjust the rotation speed to achieve online torque adaptive control of the assembly process.The two neural networks cooperated to carry out online system model identification and real-time control,so that the control parameters of the control system were optimized.Finally,a simulation experiment was carried out for the tightening process of wire thread insert.The results show that the adaptive neural network control method in this paper can have good adaptability and self-learning ability under the condition of dynamic plant.This method can effectively improve the assembly quality and stability of the wire thread.