Research On Fault Detection For Networked Control Systems

Networked control systems (NCSs) are systems with feedback loops closed via real-time shared network transmission media. The integration of the computer technology, network technology, control technology and information technology is pregnant with the development of the NCSs. As an interdisciplinary discipline, NCSs involve a wide range of content. The analysis object of it is no longer an independent control process, but the entire NCS evolving the complicated network. Due to the complexity of the networked information and the large scale of the NCSs, in practical engineering, if the NCSs fail, the economic and social losses they bring will be difficult to estimate. Therefore, the require-ment of system security and reliability is relatively higher for NCSs, which also makes the fault detection problem for NCSs become a challenging research topic with great theoret-ical and practical significance. In recent years, the research on this problem has received extensive attentions, however, there are still many problems to be resolved. For example, for the NCSs with stochastic features, there is no valid fault detection method considering the frequency range of the fault, etc.On the basis of the previous work, this dissertation is concerned on the research of fault detection method for NCSs. New definitions of finite frequency performance in-dex and the corresponding fault detection method based on these definitions are given for NCSs with stochastic features. A multi-model fault detection scheme is proposed. Mean-while, the relation between the packet dropout rate and the system performance is estab-lished. For the NCSs with finite frequency servo inputs, a new fault detection method, which is dependent on the servo input, is proposed. A new modeling approach and a valid fault detection method are given for multi-packet transmission NCSs with time-varying transmission intervals, network-induced delay and data losses. Further, under the frame-work of the geometrical method, a fault detection and isolation scheme is proposed for NCSs with packet dropout and communication constraint. Parts of the developed theories are applied to the fault detection for VTOL aircraft model, F404aircraft engine model by simulations. Simulation examples illustrate the advantages and effectiveness of our approaches. The main contents are outlined as follows:Chapters1-2summarize the development and main research methods on fault detec-tion for NCSs, give some preliminaries about the considered problem.Chapter3investigates the fault detection problem for NCSs with delay and quan-tization, and NCSs with packet dropout and quantization, respectively. Considering a logarithmic quantizer, the NCSs are first modeled as time-delay systems with quantiza-tion error and Markov jump linear systems with quantization error, respectively. Further, a new definition of finite-frequency stochastic H_index is given, which permits a direct treatment of the finite frequency performance indexes, completely avoiding approxima-tions associated with frequency weights. Subsequently, sufficient conditions are derived to guarantee that the corresponding system can achieve such a performance. The simula-tions illustrate the advantages of our approaches compared with the existing techniques.Based on the average dwell time method, chapter4studies the fault detection prob-lem for NCSs with respect to sensor failure and random packet dropout. The NCSs are firstly modeled as discrete time switched systems. Subsequently, a novel multi-model based fault detection scheme, which is valid to detect some types of sensor fault, is pro-posed. Furthermore, new sufficient conditions, which are less conservative, for solving the fault detection filters are presented. Meanwhile, the relation between the packet dropout rate and the system performance is established. Finally, the simulation is given to demon-strate the effectiveness of the proposed results.Chapter5investigates the fault detection problem for NCSs with finite frequency servo inputs and random packet dropouts in the framework of Markov jump systems and stochastic parameter systems, respectively. The sufficient conditions are first given to guarantee the corresponding systems being with a prescribed finite frequency stochastic H-index. And then a new fault detection scheme is proposed depending on the time-varying finite frequency servo inputs. Also, the valid fault detection filter is designed, and new convex sufficient conditions, which are less conservative and have less amount of computation than the existing results, are given in terms of LMIs. Finally, the simulation results show the effectiveness of the proposed methods.Chapter6is concerned on the fault detection problem for NCSs with communica-tion constraints under multi-packets transmission. With simultaneous consideration of time-varying network-induced delay, time-varying transmission interval and data packet loss, the NCSs are first modeled into stochastic switched systems with multiple stochastic parameters. Subsequently, the valid fault detection method is given. And then the convex LMI conditions are given based on the knowledge of probability statistics, which can be reduced to the existing results. Finally, an example is given to show the effectiveness of the proposed method.Chapter7studies the fault detection and isolation problem for NCSs with commu-nication constraint and data missing in the framework of geometric method. Firstly, the NCSs are modeled into Markov jump systems. And then by virtues of the common un-observable subspace and the finite frequency stochastic H_index, a novel fault detection and isolation scheme is developed in which a set of FDI filters that perform the FDI task with only partially available measurements are designed such that each residual is only sensitive to one fault in certain frequency domain and decoupled from the others. Further, less conservative conditions including some previous existing results have been presented to construct the FDI filters. Finally, an example is given to illustrate the effectiveness of the proposed method.Finally, the results of the dissertation are summarized and further research topics are pointed out.