Study On Fault Diagnosis Technology Based On The Theory Of Set-membership Filtering

In this paper, we use set-membership filter method for the study of fault diagnosis technology, this method does not need to know the statistical properties of the estimated, and it only requires the noise unknown but bounded, then it will be able to guarantee the estimated value is between it upper bounds and lower bounds. This method not only can estimate the size of the fault, but also can detect the type of fault signal. The main research work and contributions of this dissertation are as follows:1. Summary of current development of the fault diagnosis at home and abroad, the significance of research of fault diagnosis technology is put forward. Describe several common fault diagnosis methods, and its advantages and disadvantages of the methods are introduced, and then describe the theory of set-membership filtering. Eventually, work out which method to be used for fault diagnosis.2. For systems with unknown but bounded faults, simply assuming the process noise, measurement noise and fault signal are unknown but bounded, then S-procedure and LMI methods are applied to design the set-membership filter. Finally, a recursive algorithm is developed for computing the set membership filter. The simulation result shows that the method presented is available and effective.3. In order to overcome the difficulties of the uncertainty caused by the process and the noise, we introduce the fuzzy logic when deal with complex and nonlinear time-varying systems, then put the whole control of nonlinear system as fuzzy approximation of many parts of linear system to design set membership filter, and achieve fuzzy diagnosis. Finally the numerical simulation proved its effectiveness.4. For systems with state constraint and unknown but bounded faults, the nonlinear state constraint and inequality constraint are considered. For the nonlinear state constraint, it is first linearized. Finsler’s Lemma is employed to project the unco- nstrained set membership filter onto the constrained surface, and then design set membership filter that with nonlinear state constrain. Then the method proposed in this paper is applied to moving vehicle with state constraints, and modeling for the velocity and position of the vehicle. Finally the results prove its feasibility and effectiveness.5. Finally, summarizes the contents of this paper, and pointed out the existing problems and directions for future research of fault diagnosis.