基于RBF神经网络的自抗扰控制器的三电机同步控制系统

发布时间：2018-01-15 09:36

本文关键词：基于RBF神经网络的自抗扰控制器的三电机同步控制系统　出处：《江苏大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着科技的飞速发展,工业自动化已逐渐渗透到各行各业,速度、效率、性能以及经济高的多电机控制成为研究和讨论的热点,本文针对三电机同步控制的控制器参数优化问题以及解耦问题进行了一系列探讨。在分析三电机同步控制模型的基础上,因传统比例-积分-微分(Prorportion Integration Differentiation,PID)控制方案的不足,根据自抗扰控制器(Active Disturbance Rejection Controller,ADRC)不依赖于系统具体模型且可将外部扰动和内部扰动作为一个总扰动进行观测与补偿的优势,提出了基于一阶ADRC的三电机同步控制策略解决张力与速度的解耦问题,并将其与PID控制性能进行仿真实验对比,验证了ADRC在调节速度、超调量和抗干扰等方面的优异性。其次针对ADRC算法复杂、参数多且调节耗时耗力的问题,对ADRC结构进行优化的基础上,提出基于径向基函数神经网络(Radical Basis Function Neural Network,RBFNN)的自适应ADRC来解决参数优化问题。该控制器利用RBFNN能以任意精度逼近任意非线性函数能力以及结构简单等特点,实现对被控对象的在线跟踪,获得ADRC参数的在线调整信息,从而实现ADRC参数的在线调节功能。本文控制系统以可编程逻辑控制器(Programmable Logic Controller,PLC)S7-300作为控制器,通过PROFINET通讯方式上连台式机,完成硬件组态、程序的下载、调试等功能,同时在视窗控制中心(Windows Control Center,WinCC)组态软件中设计远程监控系统,实时监控控制过程变化。通过PROFIBUS-DP通讯方式下连变频器,完成主从方式通讯网络的构建。ADRC、参数采集、通讯、RBFNN等程序的编写采用结构化编程的方法在上位机博途V13软件中完成。在搭建的实验平台上,进行基于RBFNN的自适应ADRC的阶跃响应实验、解耦实验以及抗负载实验,验证了RBFNN在非线性映射方面的优势,表明该控制器不仅能实现部分参数的自调节功能,而且可降低系统超调量,提高系统的动态性能和稳态精度。实验结果证明该方法具有实际应用性。
[Abstract]:With the rapid development of science and technology, industrial automation has gradually penetrated into all walks of life, speed, efficiency, performance and high economic hot multi motor control can be studied and discussed in this paper, three motor synchronous control of controller parameters optimization and decoupling problem undertook a series of study. Based on synchronization control model in the analysis of three of the motor. Because of the traditional proportional integral differential (Prorportion Integration Differentiation, PID) lack of control scheme, based on ADRC (Active Disturbance Rejection Controller, ADRC) do not depend on the specific model and the external disturbance and the internal disturbance as a general disturbance observation and compensation advantages, proposed synchronous decoupling control problem strategies to solve the tension and speed of the three motor order based on ADRC, and the PID control performance simulation comparison, verification The ADRC in the regulation of speed, excellent overshoot and anti-jamming. Secondly, ADRC algorithm is complex, multiple parameters and the adjusting time-consuming problem, optimization based on ADRC structure, is proposed based on radial basis function neural network (Radical Basis Function Neural Network, RBFNN ADRC) to solve the adaptive a parameter optimization problem. The controller uses RBFNN can approximate any nonlinear function ability and simple structure with arbitrary precision, realize the online tracking of the controlled object, on-line tuning information of ADRC parameters obtained online, so as to realize the ADRC parameter adjustment function. The control system based on programmable logic controller (Programmable Logic, Controller, PLC) S7-300 as a controller, through the PROFINET communication way connected on the desktop, complete the hardware configuration, program download, debugging and other functions, at the same time in the windows control center (Wind Ows Control Center, WinCC) remote monitoring system design of configuration software, real-time monitoring and control of process changes. Through PROFIBUS-DP communication even under the inverter, completes the construction of master-slave communication networks.ADRC, data acquisition, communication, preparation method adopts the structure of programming RBFNN program in PC botu V13 software. The experimental platform, adaptive ADRC RBFNN step response experiment based on decoupling experiment and anti load experiment, verify the advantages of RBFNN in nonlinear mapping, show that the self regulating function of the controller can not only achieve some parameters, but also can reduce overshoot and improve the system dynamic performance and steady-state precision experimental results show that this method has practical applications.

【学位授予单位】：江苏大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;TP273;TM301.2

【参考文献】