网络化控制系统平台研发与数据驱动控制算法研究

发布时间：2018-07-16 09:07

【摘要】：在开发控制系统的过程中,通过软件实现控制算法的常规手段是通过手工编程实现,由于编程过程中需要多次调试且完成后需要多次的仿真验证,开发成本高、效率不高、可靠性不好,而且手工开发往往达不到产品级别的应用标准[1]。本文针对实时控制的需求,以及半实物仿真平台的技术特点,利用Simulink Real-Time和LabVIEW搭建实时控制平台,编写了一系列自定义s函数实现实时控制功能,作为实例进行了伺服电机的位置控制,同时实现了在此平台上的网络通信,并针对网络化控制中时延、丢包等问题提出了两种数据驱动预测控制算法。首先,提出了网络化控制系统平台研发与数据驱动算法研究设计方案,确定总体思路和各部分工作,包括平台环境设置、硬件选型、电机控制、网络通信、延迟补偿算法以及仿真等等。其次,进行了基于Simulink Real-Time的实时控制平台的搭建,控制电机的位置和实现与LabVIEW的通信的关键技术和实现方法。最大化利用电机Modbus通信协议的通讯间隔,设计了一系列自定义s函数实现电机的实时位置控制;通过自封装DLL文件和调用已有DLL文件实现了从LabVIEW调用Simulink编译文件的功能;最后实现两个目标机之间通过以太网进行通信,并实现了PID算法下对伺服电机的位置控制。再次,对于网络化控制中不可避免的网络延迟与数据包丢失、乱序等问题提出了一种数据驱动预测控制方法,可以有效地缓解网络环境对控制带来的不确定性和干扰。之后通过仿真和平台实验验证了该数据驱动预测控制算法在弥补网络诱导时延和数据包丢失、乱序问题方面的可靠性和有效性。最后,对于上面提出的基于数据预测控制算法提出了改进型;改进的具有时延与丢包的网络的数据驱动预测控制方法利用滑动窗口内选取的系统之前连续多个时刻的输入,可以分散捕获系统的复杂动态行为,有更强的跟踪时变参数的能力和更好的动态性能。之后通过仿真和本文中搭建的网络实施控制平台的实验验证了有效性。
[Abstract]:In the process of developing the control system, the conventional way to realize the control algorithm by software is to realize it by manual programming. Because of the need of many debugging and many simulation verification after the completion of the programming process, the development cost is high and the efficiency is not high. Poor reliability and manual development often fail to meet product-level application standards [1]. Aiming at the demand of real-time control and the technical characteristics of hardware-in-the-loop simulation platform, the real-time control platform is built by Simulink Real-Time and LabVIEW, and a series of self-defined s functions are programmed to realize the real-time control function. The position control of servo motor is carried out as an example, and the network communication on this platform is realized, and two data-driven predictive control algorithms are proposed to solve the problems of delay and packet loss in networked control. First of all, the research and design scheme of networked control system platform and data-driven algorithm are put forward, and the overall idea and work are determined, including platform environment setting, hardware selection, motor control, network communication, etc. Delay compensation algorithm and simulation etc. Secondly, the real-time control platform based on Simulink Real-Time is built, the position of motor is controlled and the key technology and realization method of communication with LabVIEW are presented. In order to maximize the communication interval of motor Modbus communication protocol, a series of self-defined s functions are designed to realize the real-time position control of motor, and the function of transferring Simulink file from LabVIEW is realized by self-encapsulating DLL file and calling existing DLL file. Finally, the communication between the two target machines via Ethernet is realized, and the position control of servo motor is realized under pid algorithm. Thirdly, a data-driven predictive control method is proposed for the inevitable network delay, packet loss, disorder and other problems in networked control, which can effectively alleviate the uncertainty and interference brought by the network environment to the control. Then the reliability and effectiveness of the data-driven predictive control algorithm in compensating network induced delay packet loss and order problem are verified by simulation and platform experiments. Finally, an improved data-driven predictive control algorithm is proposed for the data-based predictive control algorithm proposed above. The improved data-driven predictive control method for networks with delay and packet loss uses the input of several consecutive times before the selected system in the sliding window. It can disperse the complex dynamic behavior of the capture system, and has better ability to track time-varying parameters and better dynamic performance. Then the effectiveness is verified by simulation and the experiment of the network implementation control platform in this paper.
【学位授予单位】：北方工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP273

【参考文献】