滞环非线性时滞系统的输出反馈自适应控制方法的研究

发布时间：2018-04-22 06:06

本文选题：滞环 + 时滞　；参考：《东北电力大学》2017年硕士论文

【摘要】：随着智能材料的不断发现和愈加广泛的应用,对于严重影响系统控制精度的智能材料的固有特性—滞环现象越来越受到重视;与滞环现象类似,时滞现象也广泛存在于各种装置和实际系统中,它会降低系统控制的准确性甚至使系统变得不稳定;在实际系统中,通常输出量是可测的,并且通常情况下具有明确的物理意义。因此,研究滞环非线性时滞系统的输出反馈不但具有丰富的实际应用背景,更具有理论价值。针对同时具有滞环、时滞等非线性因素的复杂非线性系统,仅仅依靠线性系统理论或者传统的PID等控制方法已经不能很好的满足控制目标,随着对自适应控制方法研究的深入,反推法在对含有未知参数的控制、改善过渡品质等方法优势明显,以反推法为基础发展而来的动态面方法继承反推法的优势,克服了反推法的缺点,目前受到广泛关注。本文将滞环非线性时滞系统作为主要研究对象,以动态面算法为主要控制器设计算法,取得了以下成果:(1)针对一类状态变量全部可测的滞环非线性时滞系统,提出了一种基于RBF神经网络的动态面控制方法。该方法的主要特点为:采用动态面的控制方法,大大简化了控制器的设计过程;通过RBF神经网络估计未知时滞函数,避免了建立一般用来处理时滞的复杂的Krasovskii函数,放宽了对时滞函数的假设,可以在时滞函数完全未知的情况下设计控制器,改进了现有的动态面算法;稳定性分析表明,控制方案可以保证由控制对象、控制律、调参律等组成的闭环系统稳定并且通过参数的选取,可以使跟踪误差任意小;最后,通过仿真证明了方法的有效性。(2)针对一类只有输出可测的滞环非线性时滞系统,提出了一种基于RBF神经网络的动态面控制方法。该方法的主要特点为:在第三章采用动态面的控制方法以及通过RBF神经网络估计未知时滞函数的基础上,针对状态未知的情况,采用K-滤波器估计未知状态变量,实现了在状态变量未知情况下控制器的设计;稳定性分析表明,控制方案可以保证闭环系统的稳定并且通过参数的选取,可以使跟踪误差任意小;最后,通过仿真证明了方法的有效性。
[Abstract]:With the continuous discovery of intelligent materials and their wider application, more and more attention has been paid to the inherent characteristics of intelligent materials, which seriously affect the control accuracy of the system, such as hysteresis, which is similar to the hysteresis phenomenon. The phenomenon of time-delay also exists widely in various devices and practical systems. It can reduce the accuracy of the control of the system and even make the system unstable. In the actual system, the output is usually measurable. And usually has a clear physical meaning. Therefore, the study of output feedback for hysteretic nonlinear time-delay systems not only has a rich background of practical application, but also has theoretical value. For complex nonlinear systems with nonlinear factors such as hysteresis, delay and so on, the theory of linear systems or the traditional control methods such as PID can not satisfy the control objectives well. The backstepping method has obvious advantages in controlling the unknown parameters and improving the quality of transition. The dynamic plane method developed on the basis of the backstepping method inherits the advantage of the backstepping method and overcomes the shortcomings of the backstepping method. At present, it has received extensive attention. In this paper, the hysteresis nonlinear time-delay system is taken as the main research object, and the dynamic plane algorithm is taken as the main controller design algorithm. The following results are obtained: 1) for a class of hysteretic nonlinear time-delay systems with all state variables measurable, A dynamic surface control method based on RBF neural network is proposed. The main features of this method are as follows: the controller design process is greatly simplified by using dynamic surface control method, and the unknown time-delay function is estimated by RBF neural network, which avoids the establishment of complex Krasovskii functions, which are generally used to deal with delays. The assumption of time-delay function is relaxed, the controller can be designed under the condition that the time-delay function is completely unknown, and the existing dynamic plane algorithm is improved, the stability analysis shows that the control scheme can be guaranteed by the control object, the control law and the control law. The closed-loop system composed of tuning parameter law and so on is stable and the tracking error can be reduced arbitrarily by selecting parameters. Finally, the simulation results show that the method is effective for a class of hysteretic nonlinear time-delay systems with only measurable output. A dynamic surface control method based on RBF neural network is proposed. The main features of this method are as follows: in Chapter 3, the unknown state variables are estimated by using K- filter based on the control method of dynamic plane and the estimation of unknown delay function by RBF neural network. The stability analysis shows that the stability of the closed-loop system can be guaranteed by the control scheme and the tracking error can be reduced arbitrarily by selecting the parameters. The effectiveness of the method is proved by simulation.
【学位授予单位】：东北电力大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP13

【参考文献】