基于BP神经网络的高速开关阀多级电压控制策略

为了提高液压系统控制精度,通过分析几种常用驱动策略下阀芯的动态特性以及进油口压力对动态特性的影响,提出了一种可适应进油口压力变化的多级电压激励驱动策略,与常用的双电压激励策略相比具有更好的动态特性,阀芯开启、关闭时间分别降至2. 2、1. 7 ms,线圈热功率降低了68. 5%。设计了一种通过PWM调制、可输出0～60 V之间任一电压的驱动电路。采用BP神经网络对PID参数进行整定,可实现液压缸位移的精确控制。在自适应电压激励与BP神经网络联合控制策略下,恒流量液压系统液压缸位移误差在-0. 3～0. 3 mm之间,变流量液压系统液压缸位移误差在-0. 5～0. 5 mm之间。

Control Strategy of High-speed Switch Valve under Multistage Adaptive Voltage Based on BP Neural Network

In order to improve the control precision of hydraulic system, the following two aspects were done. Firstly, in terms of driving strategy, the driving strategy was considered and the dynamic characteristics of the valve core under several common driving strategies were analyzed. After considering the following aspects, a multi-stage adaptive voltage excitation driving strategy was put forward accordingly. The strategy had better dynamic characteristics than the commonly used dual voltage excitation strategy. Under this multi-stage adaptive voltage excitation driving strategy, the valve core and the closing time were reduced to 2.2ms and 1.7ms, respectively. At the same time, the coil thermal power was reduced by 68.5%. Moreover, a driving circuit which can output any voltage between 0V and 60V through PWM modulation was designed. Secondly, in terms of control strategy, the BP neural network was used to adjust the PID parameters to achieve precise control of hydraulic cylinder displacement. The network PID controller of BP neural had the characteristics of short response time, small overshoot and good robustness and so on. Under the combined control of adaptive voltage excitation and BP neural network, the hydraulic cylinder displacement error of constant flow hydraulic system was controlled within -0.3~0.3mm. Meanwhile, thanks to the combined control of adaptive voltage excitation and BP neural network, the hydraulic cylinder displacement error of variable flow hydraulic system was controlled within -0.5~0.5mm. The research had a great promotion to the study of this field.