Fault Diagnosis Method And Application Research Based On Clustering

Industrialized process monitoring and fault diagnosis, which are the intensive researchsubjects for assuring the stability and security of production process in modem industrializedprocess. However, fault diagnosis is the intensive research problem. In order to solve theproduction process of stability and security issues, this thesis presents the fault diagnosisbased on clustering algorithm, clustering algorithm is the improved fuzzy C means algorithm,which is proposed based on the classical fuzzy C means clustering algorithm. Fault diagnosisability was validated by an example and Tennessee Eastman (TE) process. The experimentalresults show that the proposed algorithm has a very high reliability and validity.At the same time, another algorithm is proposed based on probabilistic neural networkand improved fuzzy C means algorithm. The main research work is as follows:1. Elaborated the fault diagnosis method, summarized and analysis of the fault diagnosistechnology of meaning and purpose.2. Clustering methods were reviewed, and the developments of it were analyzed.3. Tennessee Eastman process is introduced to validate the effectiveness of faultdiagnosis method, and Haar wavelet is detailed to deal noise.4. Described fuzzy C means algorithm and subtractive clustering algorithm, proposed theimproved fuzzy C means algorithm, fault diagnosis ability was validated by an example andTennessee Eastman (TE) process.5. Proposed another algorithm based on probabilistic neural network and improved fuzzyC means algorithm through the Tennessee Eastman (TE) process as an example for testingfault ability.