Integrated Design Of Fault Diagnosis And Nonlinear Robust Fault-Tolerant Control For Aero-engines

The aero-engine is a typical multivariable,nonlinear,and strongly uncertain control system with high safety requirements.The currently linear single-input single-output control method based on the classical control theory is difficult to meet the requirements of control performance and high security for aero-engines.Consequently,the multivariable nonlinear robust control method is required to improve the performance of aero-engines.Due to the complex operating environment and long flight cycle,the actuator system of aero-engines need frequent operations,which can easily lead to actuator faults.The corresponding active fault-tolerant control method is studied for aero-engines to solve the actuator fault problem.Therefore,based on the high-precision component-level simulation platform T-MATS provided by NASA,the design method of the nonlinear robust control and active fault-tolerant control systems is proposed for aero-engines.The design method is suitable for aero-engines operating within large ranges near the equilibrium point,including multivariable nonlinear robust control,nonlinear robust fault-tolerant control,actuator fault diagnosis,and intelligent faulttolerant control system.The main research contents and innovative achievements of this paper can be summarized as follows.Firstly,a multivariable polynomial nonlinear H∞ control method is proposed for aero-engines.First of all,according to the input and output data,a multivariable nonlinear model in the form of the polynomial state-space equation is established to expand the controllable range of aero-engines.Then,based on the nonlinear model,the dynamic model of the aero-engine tracking error is constructed considering the reference signal.Hence,the problem of aero-engine control is transformed into the problem of stability control of tracking errors.Next,on the basis of the Lyapunov stability theory,the nonlinear H∞ robust controller and a sufficient condition are given to ensure the stability and robustness of the closed-loop control system for aero-engines.In addition,the multivariable nonlinear robust controller is obtained by solving the optimization problem in the form of linear matrix inequality.The controller with the clear analytical expression can be directly solved and calculated off-line,so the control design method has strong practicability.Finally,simulation results show that the design method can improve transient responses,disturbance rejection,surge margins,and fuel consumption of the aero-engine.Secondly,the design method of multivariable nonlinear H∞ active fault-tolerant controllers is provided for aero-engines.First of all,the multivariable nonlinear model of aero-engines is established considering the characteristics of actuator faults.Then,the structure and logic of the active fault-tolerant control system are proposed for aeroengines with actuator faults.Next,based on the theory of nonlinear H∞ control and L2 gain-like,a nonlinear active fault-tolerant controller is designed for aero-engines with actuator faults and the stability and robustness of the fault-tolerant control system are proved.The design of the fault-tolerant controller considering disturbances,actuator faults,and model errors can effectively reduce the accuracy requirement of fault diagnosis.Finally,simulation results show that this design method can reduce the impact of actuator faults on the aero-engine,and improve the transient performance and disturbance rejection ability before and after fault occurrence.Thirdly,the diagnosis method of convolutional neural networks(CNN)is proposed for aero-engines with actuator faults.First of all,the significant measurable characteristics of the gas path are selected with the operating features of actuator faults.Then,these characteristic data are preprocessed to construct sample datasets of different fault modes.Next,based on the datasets,the algorithm,process,and essential parameters of convolutional neural networks are designed and selected to optimize diagnosis results,and then the convolutional neural network diagnosis model of aeroengines is obtained.This design method can avoid the influence of the aero-engine mathematical model on the diagnosis results.Finally,simulation results show that the diagnosis model can detect the operation status of the actuators in real-time and provide reliable fault information for the aero-engine control system.Finally,the design method of the intelligent fault-tolerant control system is presented for aero-engines.First of all,on the basis of the convolutional neural network diagnosis model,the stack decision model is introduced to establish the CNN-based fault diagnosis mechanism.Then,based on the diagnosis mechanism,the nonlinear active fault-tolerant controller is designed.Furthermore,the structure and working process of the intelligent fault-tolerant control system was constructed for aero-engines.Theories prove that the control system can meet intelligent fault-tolerant control performance specifications for aero-engines.The design of the intelligent fault-tolerant control system takes account of the characteristics of fault diagnosis and fault-tolerant control respectively,which can ensure the coordinated operation of all parts of the system and complete the fault-tolerant control task for aero-engines.Finally,simulation results show that the design method can ensure the stability of the closed-loop control system for aero-engines,and has good ability in disturbance rejection and fault tolerance.In summary,the integrated design methods of fault diagnosis and nonlinear robust fault-tolerant control are systematically established and validated on the T-MATS platform for aero-engines.These methods provide strong support for the design of the next-generation aero-engine control system with higher control performance,fault tolerance,and safety.