Research Of Fault Diagnosis Based On Adaptive Fuzzy Logic System

In modern industrial production process, there is a lot of complicatednonlinear system, this kind of system is extremely complex due to its structure andoperating environment, there are many unknown time-varying parameters, proneto failure and cause personnel and property losses. Due to the fault diagnosistechnology for nonlinear system is still in an early stage of research，nonlinear alsobelongs to the unsolved problems in the field of mathematics, therefore, in thispaper, In view of the fault signal acquisition problems of nonlinear system faultdiagnosis, the adaptive fuzzy logic system combine with the parity space methodand the strong tracking filter method. Proposed two improved fuzzy faultdiagnosis algorithm to make up some of the deficiencies of the original algorithm.This paper first introduces the definition and classification of system failure,then analysis of several methods of research in fault diagnosis system. Thenintroduces the basic principle of the analysis model, and the residual generationmethod, parity space method is used for a class of discrete linear system to thesystem design and evaluation of mathematical deduction, the theoretical derivationof strong tracking filter algorithm based on the extended kalman filter. Completedtwo parts based on this study, first of all, A parity space approach to fault detectionbased on fuzzy tree model has presents. The single input and single output affinenonlinear system are approximated by a T-S fuzzy tree model to realize piecewiselinearization around different operating points. Discretization of continuous-timeT-S fuzzy systems and using parity space method for fault detection, the uses ofresidual generators make fault detection transformed into solving a minimizationproblem. This algorithm is used to minimize the impact of the rule explosionproblem of fault detection system performance. The numerical simulation of aninverted pendulum with abrupt fault verifies the efficiency of the proposed method.Secondly, strong tracking filter tracking small mutations in the state is not ideal,and prone to over-regulate, the paper puts forward a fuzzy adaptive strongtracking filter, introduced the fuzzy controller to online adjust the factor, inimproving approximation accuracy while making a smoother curve approximation.Finally, using adaptive fuzzy strong tracking filter method for fault diagnosis of aclass of simple nonlinear system, prove the superiority of the algorithm.