以簡單有效的權重主成份分析法選取主要特徵點 及其應用於辨識馬達的品質類別

Feature Selection Algorithm for Motor Quality Types using Weighted Principal Component Analysis

劉美珍(Mei-Chen Liu)

指導教授 ： 葉雲奇

健行科技大學/電資學院/電子工程系碩士班/碩士(2016年)

摘要

本篇論文提出以簡單有效的權重主成份分析法(WPCA)選取主要特徵點及其應用於辨識馬達的品質類別。本篇論文，包含兩大主題：主題一是以WPCA演算法選取主要特徵點，它包含二個重要的程序，程序Procedure-FFV (功能是尋找最終之權重值)，與程序Procedure-DPC (功能是尋找最終的主要成份)。主題二是以群聚分析法(cluster analysis)辨識馬達的品質類別。實驗過程如下：首先，輸入原始的馬達電流波形訊號，該訊號有9個原始特徵點，經WPCA處理後，選取了其中的6個成為主要特徵點。接著再以群聚分析法辨識馬達的品質類別。經過多次的實驗，其正確辨識率平均可達99.62%。依據實驗結果，可說明本篇論文所提出的是一個有效、簡單、及快速的方法。

關鍵字

權重主成份分析法(WPCA) ； 群聚分析法(cluster analysis) ； 特徵點選取(feature selection)

並列摘要

This paper proposes a qualitative feature selection for motor quality types using Weighted Principal Component Analysis (WPCA) method. In this dissertation, some novel and efficient algorithms in two related research topics about current waveform of motor will be presented and discussed. In the first research topic, qualitative feature selection using WPCA。The WPCA includes two processes, one is the Procedure-FFV (find the final weights) process and the other is the Procedure-DPC (determine the principal components) process. In the second research topic, the classification stage for determining motor’s quality types using the cluster analysis method. In the experiment, the input variables of the WPCA are nine original features and the output variables are six qualitative features, the total classification accuracy was approximately 99.62% by cluster analysis method. Experimental results indicate that the proposed WPCA provides an efficient, simple and fast method for feature selection on motor’s current waveforms.

並列關鍵字

Weighted Principal Component Analysis (WPCA) ； fuzzy logic ； feature selection.

參考文獻

1. 宋有維,“使用簡單有效的歐式距離量測法辨識馬達的品質類別”, 健行科技大學碩士論文, 2014.

2. 林綠綺,葉雲奇,楚萃瑤, “使用群聚分析法辨識馬達的品質類別,” Proceedings of 2013 National Symposium on Systems Science and Engineering, pp. 18-20, 2013

3. 黃煌翔, “介面技術與週邊設備,”全華圖書公司, 2011.

4. 網站 http://www.alldatasheet.com/view.jsp?Searchword=MP7523.

5. A.Kapun, M. Curkovic, A. Hace, K. Jezernik, “Identifying dynamic model parameters of a BLDC motor,” Simul. Model. Pract. and Theory, vol. 16, pp. 1254-1265, 2008.