원문정보
초록
영어
In traditional fault diagnosis method, a large number of experiments are needed to get the optimal performance classifier which diagnoses type of fault. Because of classifier algorithm limit, there is no one classifier can be applied to all kinds of fault diagnosis. In order to avoid the disadvantages caused by single classifier approach, decision level fusion method based on multiple classifiers fusion is introduced in the field of fault diagnosis. The fusion method with fuzzy comprehensive evaluation is put forward and the basic evaluation model is set up. The reasonable distribution of classifiers weight that affects diagnosis result directly is vital. Firstly, the evaluation function which measures member classifier’s diagnostic accuracy and correctness is constructed based on the theory of information entropy. Then, weights are distributed to each classifier with entropy coefficient according to the value of evaluation function. Experiments are carried out to demonstrate the effectiveness of the proposed method and results show that fault recognition rate after fusion is higher compared with the single classifier method.
목차
1. Introduction
2. The Fuzzy Model for Comprehensive Evaluation of Fault Diagnosis
2.1. Factor Set
2.2. Judgment Set
2.3. Single Factor Judgment
3. Method of Deciding the Weight of Each Member Classifier
3.1. Accuracy of Fault Diagnosis
3.2. Correctness of Fault Diagnosis
3.3. Member Classifiers’ Weights
4. Fusion of Multiple Classifiers
5. Experiment
6. Conclusions
References