第43卷第2期 Vol. 43No. 2
叙压技木
FORGING & STAMPING TECHNOLOGY
2018年2月 Feb.2018
基于小波神经网络控制的伺服直驱泵控系统压力与位置特性分析
韩贺永，乔永杰，刘少龙，王
凯
（太原科技大学重型机械教育部工程研究中心，山西太原030024）
摘要：由于服直驱泵控缸压力与位置双闭环控制系统的结构复杂，存在非线性、时变性等特点，单纯通过数学模型进行分析比较紫项，结果也不精确。通过对泵控缸压力与位置双闭环控制工作原理的分析，在AMEsim中搭建出系统控制回路的物理模型，在Simulink中搭建小波神经网络控制部分模型，然后进行联合仿真分析，并进行实验验证。伺服电机采用三相交流永磁同步电机，控制芯片采用DSP28335，驱动器采用台达ASDA-A2伺服驱动器，实验采用2种控制方式，即普通PID控制和小波神经网络控制，通过对额定负载情况下压力与位置的响应特性曲线进行分析，结果表明采用小波神经网络控制可以显著提高伺服直驱票控缸压力与位置控制系统的控制精度和稳定性，
关键词：票控缸系统；小波神经网络；AMEsim；Simulink；双闭环控制；联合仿真 DOI: 10. 13330/j. issn. 1000-3940. 2018. 02. 021
中图分类号：TH137
文献标识码：A
文章编号：1000-3940（2018）02-0122-06
Analysis onpressureand position characteristics of servo direct drive pump-control-cylinder system based on wavelet neural network control
Han Heyong, Qiao Yongjie, Liu Shaolong, Wang Kai
( Heavy Machinery Engineering Research Center of the Ministry of Education, Taiyuan University of Science and
Technology, Taiyuan 030024, China)
Abstract : Because the double closed-loop control system structure of pressure and position for servo direct drive pump-control-cylinder arc complex, nonlinear and time-invariance, the analysis solely based on mathematical model will be very complicate with inaccurate result. By analyzing the double closed-loop control principle of pressure and position for pump-control-cylinder, a physical model of the system control loop was set up by AMEsim, and the wavelet neural network control model was constructed by Simulink. Then, the co-simulation analysis was conducted, and the experimentation and verification were also perfomed. Therefore, three-phase AC permanent magnet syn-chronous motor was used as the servo motor, DSP28335 was used as the control chip, and the telda ASDA-A2 servo drive was used as the drive. The experiment was controlled by both common PID control and wavelet neural network control, and the response characteristics curves of pressure and position under rated load were analyzed. The results show that the control precision and stability of control system for servo direct drive pump-control-cylinder system can be improved significantly by the wavelet neural network control.
Key words : pump-control-cylinder system; wavelet neural network; AMEsim; Simulink ; double closed-loop conrol ; co-simulation
司服直驱泵控缸系统属于直驱式容积控制电液伺服系统，该系统通过交流伺服电机直接驱动定量泵来实现液压缸的速度、1
位置等闭环控制。因此，
系统的关键是对伺服电机的控制，国内外学者已做了大量研究工作，包括先进的PID控制、模糊控制、神经网络控制等，而小波神经网络控制是将小波分析与神经网络相结合的产物，是解决非线性问题的
收稿日期：20170821；修订日期：20171118 基金项目：国家自然科学基金资助项目（51505315）作者简介：韩贺水（1982-），男，博士，副教投
万方整据il：277921887@qy:com
有力工具，在电机控制领域也不断发展壮大。文献[1］针对低速下定子磁链畸变、定子电流谐波大，电磁转矩脉动大的缺点，提出一种基于小波神经网络的非线性自回归移动平均模型来改善直接转矩控制的性能缺陷。文献【2］提出了基于小波神经网络补偿器和偏差耦合控制结构相结合的协同控制策略，解决了多伺服电机系统协同控制方案难以满足高精度控制的要求。文献［3］在复合前馈PID控制方法的基础上，利用小波神经网络实现了对永磁同步直线电机干扰的在线估计补偿，通过所构建的小波神经网络能解决跟踪性能易受推力波动、摩擦力等影响的问题。文献【4］采用基于小波神经网