Fault Detection And Fault Tolerant Control Based On Mixed Logic Dynamic Model

As the size, automation and complexity of modern industrial process are improved continuously, production processes not only contain the continuous physical reaction, chemical reaction and biological reaction, but also contain a large number of human instructions, logical constraints, switching between different operating modes, ect. Mixed characteristics of system are more and more obvious, and the computational complexity significantly increased. According to the characteristics of hybrid system, modeling and predictive control problem based on mixed logical model are researched in this paper. At the same time, fault diagnosis issues based on mixed logical model fault and active fault tolerant are studied. Take the typical hybrid system continuous stirred tank reactor for example simulation, the simulation results proved the validity of this algorithm in the paper. Main research works and achievements are summarized as following:The modeling approach based on mixed logical model is introduced in detail. Logic proposition and mixed logic proposition are converted into integer linear inequality constraints, and the mixed logical model hybrid system CSTR is set up. Predictive control strategy of hybrid system based on MLD is discussed. By the optimization control simulation of CSTR, the simulation results show the effectiveness of the predictive control, and the strategy can satisfy needs of the control, besides it can deal with the nonlinear of the system effectively and avoid shock when switching models.The fault diagnosis problem of hybrid system is studied. The fault models of actuator and sensor fault for hybrid system are analyzed and established, and these faults can be unified into the system MLD model framework. According to the principle of moving horizon estimation, the cost function of moving horizon estimation is constructed then it is used to detect hybrid system failure by estimating the states of the system. Based on the above failure detection and in view of the actuator faults, the active fault tolerance predictive control is researched. The strategy can achieve the fault-tolerance goals, and maintain stability and acceptable performance of the fault system. Active fault tolerant control scheme is then applied to a continuous stirred tank reactor(CSTR). Simulation results show the effectiveness of this fault tolerant MPC control.In the paper, the optimization problems of predictive control, fault detection and fault tolerant control can be converted into MIQP problems. The optimization control respectively based on 1 norm, 2 norm and the infinite norm indicators are discussed. Through linear programming and quadratic programming performance simulation, compare the control performance of the two methods.