Fault Diagnosis And Fault Tolerant Control For Stochastic Distribution Collaborative Control Systems

With the development of modern science and technology,the modern control system has become more and more complicated and intelligent.Complex systems mostly consist of multiple interacting subsystems and the control objectives of the system can not be completed by a single system or a single control algorithm.How to ensure the stability of complex system becomes an important research content.Moreover,fault diagnosis and fault-tolerant control of a single system can not meet the actual needs.Therefore,the fault diagnosis and fault-tolerant control algorithms for collaborative systems are urgently needed.Collaborative systems refer to multiple subsystems working together to accomplish some common control objectives.At present,few research results on fault-tolerant control of collaborative systems are reported.The new idea of the collaborative fault tolerance control is that when fault happens in one of the subsystems the whole collaborative system,the faulty subsystem itself can not compensate the fault or compensate the fault with higher costs.Then the fault-tolerant control algorithm will be designed in other non-fault subsystems in the collaborative systems,so that the entire collaborative system can still achieve the original control objectives after the failure occurs.In this thesis,fault diagnosis and fault-tolerant control algorithms are designed for a class of stochastic distribution collaborative system.The specific contents are shown as follows:(1)A fault diagnosis and fault-tolerant control scheme is present for a class of stochastic distribution collaborative control systems,which are composed of two subsystems connected in series to complete the control target.The output of the whole system is the output PDF of the second subsystem.To diagnose the fault in the first subsystem,an adaptive diagnostic observer and linear matrix inequality(LMI)technique are used to obtain the adaptive tuning law to estimate the fault.When fault occurs,the fault itself can not be compensated in the first subsystem and a model predictive controller is designed in the second subsystem to compensate the fault,making the post-fault output PDF still track the desired PDF as close as possible.A simulated example is given and desired results have been obtained.(2)The model predictive fault-tolerant control algorithm is conducted for a class of stochastic distribution collaborative control systems,which are composed of three subsystems connected in series to complete the target.For the considered collaborative system,the system parameters of the second subsystem are affected by the output PDF of the first subsystem and the system parameters of the third subsystem are affected by the output PDF of the second subsystem.Thus,the output of the whole collaborative system is the output PDF of the third subsystem.The RBF neural network is used to approximate the output PDF.To diagnose the fault in the first subsystem,an adaptive diagnostic observer and the adaptive tuning law are used to estimate the fault.When fault occurs,the fault itself can not be compensated in the first subsystem and the desired output PDF is unknown,so the minimum entropy controller is designed in the second and third subsystems to compensate the influence of the fault,minimizing the entropy of the system output.A simulated example is given and desired results have been obtained.(3)A fault-tolerant control algorithm is designed for a class of nonlinear stochastic distribution collaborative control systems,which are composed of two nonlinear subsystems connected in series to complete the target.The Takagi-Sugeno(T-S)fuzzy model is applied to approximate the nonlinear dynamics of subsystems.The output of the whole system is the output PDF of the second subsystem.The fuzzy logic systems(FLS)is used to approximate the output PDF.To diagnose the fault in the first subsystem,an adaptive diagnostic observer is designed to estimate the fault.When fault occurs,the fault itself can not be compensated in the first subsystem and a model predictive controller is designed in the second subsystem to compensate the fault,making the post-fault output PDF still track the desired PDF as close as possible.A simulated example is given and desired results have been obtained.