회전기계의 결함진단을 위한 결정레벨 융합기술 : Decision-level Fusion Techniques for Fault Diagnosis in Rotation Machinery

다국어 초록 (Multilingual Abstract)

Improvement of recognition rate is ultimate aim for fault diagnosis researchers using pattern recognition techniques. However, unique recognition method can only supports a limited classification capability, which is insufficient for real-world application. An ongoing strategy is the decision fusion techniques. In order to avoid the shortage of single information source coupled with unique decision method, the new approach is required to generate better results. This paper proposes a decision fusion system for fault diagnosis, which integrates data sources of different types of sensors and decisions of multiple classifiers. First, non-commensurate sensors data sets are combined using an improved sensor fusion method at decision-level by using relativity theory. The generated labels vectors are then selected based on correlation measure of classifiers in order to find an optimal sequence of classifiers fusion, which can lead to the best fusion performance. Finally, multi-agent classifiers fusion algorithm is employed as the core of the whole fault diagnosis system. Also different fusion methods at decision-level are compared. The efficiency of the proposed system was demonstrated through two experiments. The first one is fault diagnosis of induction motors using test rig designed by our intelligent mechanics lab. In the second experiment, faults data of elevator motor received from Korea Elevator Safety Center (KESC) were employed and regarded as a particular diagnosis example. The results of the two experiments show that the proposed system can take super performance when compared with the best individual classifier with single source data.

목차 (Table of Contents)

• I. Introduction = 1
• 1. Background of Machinery Fault Diagnosis = 1
• 2. Developments and Challenges of Condition Monitoring and Fault Diagnosis Techniques = 2
• 2.1 Traditional Diagnosis Techniques = 3
• 2.2 Artificial Intelligence (AI) Techniques = 3
• 2.3 Information and Decision Fusion = 4
• 3. Objective of the Study = 5
• 4. The Structure of the Thesis = 6
• 5. Science Contribution of the Thesis = 8
• References = 8
• II. Background Knowledge = 10
• 1. Multiple Sensors Fusion (MSF) = 10
• 1.1 Data-level Fusion = 11
• 1.2 Feature-level Fusion = 12
• 1.3 Decision-level Fusion = 13
• 2. Introduction of Classifiers = 14
• 2.1 The concept of Classifiers = 14
• 2.2 k-Nearest-Neighbor classifier (k-NN)[8] = 15
• 2.3 Linear Discriminant Analysis (LDA)[9] = 16
• 2.4 Decision Trees [7] = 18
• 2.5 Support Vector Machines(SVM)[11] = 20
• 2.6 Fuzzy c means (FCM)[7] = 21
• 2.7 ART-KNN[12] = 22
• 3. Classifier Selection = 25
• 3.1 Background = 25
• 3.2 Agreement Measurement[16] = 26
• 3.3 Correlation Measure[17] = 26
• 3.3.1 2-Classifier correlation analysis = 26
• 3.3.2 Multi-classifier correlation analysis = 27
• 4. Multiple Classifiers Fusion (MCF) = 28
• 4.1 Background = 28
• 4.2 Majority-Voting[19] = 30
• 4.3 Bayesian Belief Method[20] = 30
• 4.4 Behavior-Knowledge Space(BKS)[22] = 31
• 4.5 Dempster-Shafer Theory[24] = 34
• 4.6 Decision Templates[25] = 35
• 4.7 Borda count method[26] = 38
• 4.8 Architectures for MCF = 39
• 4.8.1 Stacked generalization = 39
• 4.8.2 Other architectures = 40
• References = 41
• Ⅲ. Fusion Diagnosis System Based on Multi-agent Algorithm = 44
• 1. Level 1 to Level 3 - Preparation for Decision Fusion = 45
• 2. Level 4 - Sensors Data Fusion at Decision Level = 46
• 3. Level 5 - Classifier Selection = 47
• 4. Level 6 - Decision Fusions using Multi-agent Algorithm = 47
• References = 51
• Ⅳ. Common Induction Motor Faults = 52
• 1. Introduction = 52
• 2. Main Faults of Induction Motor = 52
• 2.1 Bearing Faults = 52
• 2.2 Broken Rotor Bar = 53
• 2.3 Stator Winding Faults = 55
• 2.4 Eccentricity Faults = 56
• 2.5 Misalignment Faults = 57
• 2.6 Mass Unbalance = 59
• 2.7 Bent Shaft or Rotor Bow = 60
• V. Experiments and Methods Comparison = 62
• 1. Self-designed Test Rig = 62
• 1.1 Sensors Data Fusion = 67
• 1.2 Selection of Classifiers = 68
• 1.3 Multi-agent Fusion Algorithm and Comparison = 70
• 2. Real-world Application to Induction Motors in Elevator Faults Diagnosis = 73
• 2.1 Background of Elevator Faults Diagnosis = 73
• 2.2 Decision Fusion System used in Elevator Motor Fault Diagnosis = 73
• 2.2.1 Experiment apparatus = 74
• 2.2.2 Description of experiment data = 76
• 2.2.3 Description of features calculated = 97
• 2.2.4 Description of classifier used = 97
• 2.3 Results and Discussion = 99
• 2.3.1 Individual classification accuracy rate = 99
• 2.3.2 Selection of classifiers = 101
• 2.3.3 Classifiers fusion = 103
• 2.3.4 Comments for experiment results = 105
• References = 106
• VI. Conclusions and Future Developments = 108
• 1 Conclusions = 108
• 2 Future Developments = 109