- 學位論文
主成分分析法的應用：以分析直流馬達訊號為例
Application of Principal Component Analysis: Analysis of DC Motor Signals as an Example
摘要
本篇論文包含下列的兩大主題，分別是：(a)主題一：主成分分析法(Principal Component Analysis, PCA)的研究，及(b)主題二：將主成分分析法應用於分析直流馬達的電流訊號。其中主題一的主成分分析法(PCA)，它的主要功能是將資料簡化(data reduction)或維度化簡(dimension reduction)。PCA的分析過程簡述如下：(1)分析變數之間的相關性，目的是要找出這些變數之間的相關性；(2)依據分析的結果來減少變數個數或是產生另一組數量較原變數個數少的新變數。簡言之，PCA首先是從原始變數間找到它們的相依關係，而後是保留影響程度較大的變數，而去除影響程度較小的變數。如此，PCA不但達到減少變數個數的目的，也同時保留大部分的訊息。其中的主題二，是將PCA法應用於分析直流馬達(DC motor)的電流訊號。本主題二分成如下的四個部份，分別是：(a)定義馬達電流訊號之測試點；(b)尋找與儲存馬達電流訊號的測試點；(c)以PCA法選取各種馬達品質類別的主要成分；(d)各種故障馬達品質類別的辨識。本文能辨識五種的馬達品質類別，包含好的馬達品質類別(Type-good)及四種故障馬達品質類別(Type error-1~Type error-4)。經多次實驗，正確辨識率(TCA)Type error-1為91.66%，Type error-2為86.66%，Type error-3為84.61%，Type error-4為81.81%，Type-good為94.73%，平均正確辨識率為88.57%。
並列摘要
This study consists of two main topics: (a) Topic 1: The research on Principal Component Analysis (PCA), and (b) Topic 2: Using PCA for analyzing DC motors current signal. The principal component analysis is utilized for data reduction or dimension reduction. The process of PCA is summarized as follows: (1) To find the correlation between these variables by analyzing the correlation between variables and ; (2) According to the analyzed results, one can reduce the number of variables or generate another group which the number of new variables is smaller than the number of original variables. In other words, the PCA firstly finds dependencies of these original variables, and then retains the variables that have a greater impact and removes the less-affected variables. Therefore, PCA not only achieves the purpose of reducing the number of variables, but also retains most of the information. The second topic is to apply the PCA method to analyze the current signals of DC motors. This topic is divided into two parts: (a) Defining test points for motor current signals; (b) Finding and storing test points for motor current signals; (c) Selecting principal components of various motor quality classes using the PCA method; (d) Determing motor quality types. This study identifies five types of motor quality, including Type-good and Type error-1 to Type error-4. After lots of experiments, the results show that the total classification accuracies (TCAs) are 91.66%，86.66%，84.61%，81.81%，and 94.73% for Type error-1, Type error-2, Type error-3, Type error-4, and Type-good, respectively. The average TCA is 88.57%.