基于改进BP-PID控制的磨矿回路控制系统的研究

Research on Grinding Circuit Control System Based on Improved BP-PID Control

由于磨矿过程的非线性以及时变性等特性,常规的PID控制很难达到理想的效果,而利用神经网络能够实现最优控制参数基础上的PID控制。因此,构建了一种改进BP神经网络自学习PID控制算法。为了验证算法的性能,选取三个具有传递函数的复杂模型作为测试对象,测试结果表明,改进型BP-PID控制器跟踪误差更小,鲁棒性和抗干扰能力更强,比传统Z-N方法具有更好的性能。最后,将其应用在磨矿过程中,实际应用表明,不仅实现了给矿、给水的稳定控制,而且改进后的控制器动态和静态性能更好,磨机运行更加稳定。

Due to the non-linearity and time-varying properties in the grinding process, the conventional PID control was difficult to achieve the desired results. Whereas the neural network could be used to realize the PID control based on the optimal control parameters. Therefore, a network self-learning PID control algorithm of improved BP neural was established. In order to verify the algorithm performance, three complex models with transfer functions were selected as test objects. Then the test results showed that the improved BP-PID controller had less tracking error, stronger robustness and anti-interference ability, as well as had better control performance than the traditional Z-N method. Finally, by applying it into the grinding process, the actual application results showed that it could achieve the stable control of feeding ore and water, the improved controller had better dynamic and static performances, and the mill ran more stable than before.