【作者】 李亚洁；

【导师】 李炜；

【作者基本信息】 兰州理工大学 ， 控制理论与控制工程， 2008， 硕士

【摘要】 网络化控制系统是通过一个实时网络构成的闭环控制系统,其突出特点是反馈控制系统中的控制回路是通过网络信道连接而成的闭环。与传统意义上的控制系统相比,网络化控制系统具有连线少,成本低,资源共享,便于系统安装、维护、扩展和故障诊断等优点。然而,在网络化控制系统通过共享网络信息资源实现控制而带来了诸多优势的同时,受网络带宽的限制,以及自身存在的不同采样率、丢包、拥塞、延迟等问题也给传统的控制理论带来了新的挑战。随着计算机与信息技术的迅速发展,现代网络化控制系统已广泛应用于各个行业中,此类系统不仅规模庞大、复杂度高,不确定因素众多,而且对安全性和可靠性的要求甚高,容错能力已成为现代网络化控制系统的本征性能要求。因而对不确定网络化控制系统容错控制的研究就成为一个极具理论和现实意义的研究课题。本文针对不确定网络化控制系统的容错控制问题,以受网络时延和不同步影响的网络化控制系统为受控对象,综合应用Lyapunov稳定性理论和线性矩阵不等式方法,采用状态反馈与时滞状态反馈控制策略,主要进行了以下几方面的工作:1)研究了不确定网络化控制系统的鲁棒完整性问题。分别针对仅存在时不确定性、同时存在时延及参数不确定性、存在不确定扰动的三类网络化控制系统,基于时延模型,采用状态反馈或时滞状态反馈控制策略,分别讨论了此类系统对传感器或执行器失效故障下具有鲁棒完整性需满足的充分条件,并给出了控制器的设计方法。2)研究了不确定网络化控制系统的鲁棒保性能容错控制问题。分别针对仅存在时延不确定性和同时存在时延及参数不确定性的网络化控制系统,基于时延模型,采用状态反馈或时滞状态反馈控制策略,分别讨论了此类系统对传感器、执行器失效故障需满足的鲁棒保性能容错的充分条件,并给出了控制器的设计方法。3)针对文中推证出的理论结果进行大量仿真验证和对比分析,证实了文中所推证定理的正确性和方法的有效性。通过文中定理的推证,在理论上得到了对同一网络化控制系统当传感器个数和执行器个数相同时,对于传感器和执行器失效故障其鲁棒容错充分条件具有数学意义上的等价性的结论,同时用仿真实例验证了在实际应用中,两种容错控制器的通用性和可替代性。4)当具有控制时延的网络化控制系统采用时滞状态反馈策略时,在传感器或执行器发生失效故障时,系统均具有良好的控制性能,大量的仿真结果还揭示出,随着时滞状态反馈时延τ与系统控制时延偏离度的增大,系统的控制性能会随之变差,系统所付出的性能代价也随之变大,当τ值与系统控制时延接近时,其控制效果会变好。对于同一网络化控制系统当时滞状态反馈的控制时延τ与系统控制时延相匹配时,时滞状态反馈的控制效果要明显优于状态反馈,但前者求解控制器的约束条件较后者的增多,控制器的可解性有所下降,因而在实际应用中需要在性能的优劣和解空间的大小二者之间寻找一个折中。5)对仅考虑网络诱导时延不确定性和同时考虑参数不确定性与时延不确定性的网络化控制系统,在传感器或执行器发生失效故障时,其鲁棒容错控制器的约束条件会随系统复杂度的增加而变多,也会导致控制器的求解空间变小。更多还原

【Abstract】 Networked control system was a closed control system by a real time network,and the feature of the system was that the control loop of the feedback control system was built by the networked channel.Compared with the conventional control systems,the NCS had some primary advantages,such as:reducing system wiring,cutting the cost,sharing the information, ease of the system maintenance,extending,diagnosis.However,while the NCS brought lots of conveniences to us through exchanging data among devices connected to the shared medium,because of the limiting of the network band,the issues of unequal-distance sampling, packet dropout,and networked induced-delay brought new challenge to the conventional control theory.With the development of the computer and information technology,modern networked control system which had wide application in the industry had large scale,and it must have high reliability and safty,so fault-tolerant ability of the NCS was a intrinsic condition during the system operating.In a word,the study of fault-tolerant control in networked control system not only had theoretical value but also had strong practical value.The problem of faut-toleant control in the NCS was studied in this paper,which based on the time-delay model of the networked control system with networked induced-delay and adopted the Lyapunov stable theory,Linear Matrix Inequalities.Some work was done as following by adopting state feedback control law or time-delay state feedback control law.Firstly,the problem of robust integrity in the NCS was studied,and the problem studied three systems which are the NCS with uncertainty of time-delay,the NCS with uncertainties of time-delay and parameter,the NCS with uncertainty of disturb.Based on the time-delay model,the sufficient conditions for the NCS against actuar or sensor failures possessing robust integrity were given by adopting state feedback control law or time-delay feedback control law,and the corresponding design method of robust fault-tolerant controller was presented.Secondly,the problem of robust guaranteeing fault-tolerant control in the NCS was studied,the problem studied two systems which are the NCS with uncertainty of time-delay, the NCS with uncertainties of time-delay and parameter.Based on the time-delay model,the sufficient conditions for the NCS against actuar or sensor failures possessing robust guaranteeing fault-tolerant were given by adopting state feedback control law or time-delay feedback control law,and the corresponding design method of robust fault-tolerant controller was presented. Thirdly,the results of simulation demonstrated the effectiveness and feasibility of the presented method in this paper.Further more,when the number of actuator was equal to the number of the sensor in the same closed-loop systems,it showed that the robust fault tolerant controllers are equivalent in mathematics and exchangeable in practise.Lastly,on the one hand,when the NCS with control time-delay adopted the time-delay feedback control law,the system against the sensor or actuar failures had good performance. The results of simulation showed that with the time-delay becoming large the performance of the system would turn bad and the system cost would turn large.When the control time-delayτis similar to the system time-delay,the performance of the system would be better.When the control time-delay is similar to the system time-delay for the same NCS,the performance of the NCS by adopting the time-delay feedback control law was better than the performance of the NCS by adopting the feedback control law,however the constrains of the controller in the former was more than the later,which would make the space of solution become small.On the other hand,comparing the NCS with uncertainty of time-delay to the NCS with the uncertainty of time-delay and parameter,the robust fault tolerant controller in the later against sensor or actuator failure would have more constrains which would lead to the solver space turning small.更多还原