Multivariate process monitoring with improved real-time contrasts control chart
- 주제(키워드) Real-time contrasts (RTC) , Fault detection , Fault isolation , Random forests , Weighted voting , Class imbalance
- 발행기관 고려대학교 대학원
- 지도교수 백준걸
- 발행년도 2017
- 학위수여년월 2017. 2
- 학위구분 석사
- 학과 대학원 산업경영공학과
- 원문페이지 65 p
- 실제URI http://www.dcollection.net/handler/korea/000000071610
- 본문언어 영어
- 제출원본 000045897390
초록/요약
Real-time fault detection and isolation are important tasks in process monitoring. A real-time contrasts (RTC) control chart converts the process monitoring problem to a real-time classification problem and outperforms existing methods. However, the monitoring statistics of the original RTC chart are discrete, and this could make the fault detection ability less efficient. To make monitoring statistics continuous, distance-based RTC control charts using support vector machines (SVM) and kernel linear discriminant analysis (KLDA) were proposed. Although the distance-based RTC charts outperformed the original RTC chart, the distance-based RTC charts have a disadvantage in that it is difficult to analyze the causes of faults using these charts. Therefore, we propose improved RTC control charts using random forests with weighted voting. These improved RTC control charts not only detect changes more rapidly by making monitoring statistics continuous, but they can also analyze the causes of faults, in a similar manner to the original RTC chart. In addition, the improved RTC control charts alleviate a class imbalance problem by using F-measure, G-mean, and Matthews correlation coefficient (MCC) as performance measures to assign proper weights to individual classifiers. Experiments show that the proposed methods outperform the original RTC chart and are more effective than the distance-based RTC charts using SVM and KLDA.
more목차
TABLE OF CONTENTS
1. Introduction 1
2. Real-time contrasts (RTC) control chart 6
2.1 Real-time contrasts (RTC) method 6
2.2 Random forests 9
3. RTC control charts using random forests with weighted voting 12
3.1 Weighted voting using F-measure, G-mean and Matthews correlation coefficient 12
3.2 RTC control charts using weighted voting 16
3.3 Fault isolation via variable importance 18
4. Experiments 22
4.1 Normal distribution 23
4.2 The effects of the moving window sizes 30
4.3 Non-normal distribution 33
4.4 A real example 39
5. Concluding remarks 49
[References] 50
