Research On Rotor Rub-Impact Fault Detection Method Based On Interval Observer

As the core part of large-scale rotating machinery and equipment,the rotor system is prone to multiple failures under high-speed operating conditions.Rotor-stator rub-impact fault is one of the common typical faults.When it occurs,it will reduce the mechanical operation efficiency,lose effective energy,and even cause serious safety accidents and economic losses.Therefore,rapid and accurate fault diagnosis of rub-impact faults ensures the safe operation of rotating machinery.The research on fault detection and estimation of rotor-stator rub-impact fault have certain value both in theory and application.Aiming at the problem of rotor rub-impact fault detection,this dissertation studies the fault detection and fault estimation methods based on interval observer.For the time-invariant system containing disturbance and measurement noise,a design method based on the coordinate transformation of interval observer is proposed.Hurwitz and Metzler matrices are designed for continuous systems,Schur and Metzler matrices are designed for discrete systems to ensure the stability and nonnegative of the error dynamic system.On this basis,a design method of interval observer based on positive realization is proposed,which can better ensure the existence of a gain matrix.The effectiveness of the state estimation and fault detection of the interval observer is illustrated by examples of fault-free and actuator faults,respectively.By comparing the real state of the system with the state observed by the interval observer,the detection of actuator faults is completed.The simulation results show that the interval observer can quickly detect actuator faults.The interval observer method is introduced into the rotor-stator rub-impact fault diagnosis,and a fault detection strategy based on output error is proposed.Based on the force analysis of the rotor rub-impact fault model,the system dynamic model is established.The interval observer of the normal rotor is designed by using the method proposed above,and the numerical solution is performed using Matlab.The upper and lower bounds of the interval observer are used as the fault diagnosis threshold interval,the fault is detected according to the fault detection strategy.The results show that the method can detect the fault in time and satisfy the requirements of fault diagnosis.Based on this method,the structure of interval observer is improved,and the optimal gain matrix optimization based on H∞ method is proposed,which tighten the estimation interval,improves the detection accuracy,and reduces underreporting.Through simulation verification,the comparison result with the aforementioned method shows that the interval width is significantly reduced,which can effectively increase the fault detection,and has certain optimality.