- 學位論文
以萬用序列匯流排為傳輸介面之非侵入式虛擬電力監測儀表系統開發
The Development of Non-Intrusive Virtual-Instrument Power Monitoring System Based On Universal Serial Bus Interface
摘要
隨著工業上的發展，各種負載設備的使用大量增加，因此大型電力系統日趨複雜化，而使得負載監測的成本大幅增加。在虛擬儀表的應用上，將資料收集、資料處理與資料顯示皆整合在同一個儀表系統上，可以更方便、有效、有彈性而經濟地方式執行電力監測的任務。在電力監測上，由於電力系統的複雜度提高，如何確保電力波形資料快速而正確的傳輸，傳輸介面的選擇也日益重要。然而傳統虛擬儀表系統採用RS-232等傳輸介面，因其頻寬的限制而使得整體效能降低，且不具擴充性。而分散式的虛擬儀表電力監測系統，可以透過萬列序阜匯流排（Universal Serial Bus, USB）介面的使用，具體地增加並改善其即時傳輸的功能。 在工業與商業電力系統監測的應用上，低成本的電力監測儀表的訴求是必需的。傳統之侵入式負載監測系統需要於每一個負載加裝電力監測儀表，其缺點除了維修不易外，也大幅增加硬體設備成本。而非侵入式負載監測系統，只需於電力入口處加裝電力監測儀表，並與虛擬儀表搭配使用，將可以大幅減少硬體設備成本。因此本文擬以開發一以萬列序阜匯流排為傳輸介面的非侵入式負載監測系統。 而在負載分析的演算法上，非侵入式負載監測系統也比傳統之監測系統來得複雜。因而在負載分析上，負載特徵的選取將直接影響電力監測能力的好壞。傳統的特徵擷取方法，以統計學上的主成分分析法（Principal Component Analysis, PCA）為最普遍的方法。然而主成分分析法只適用於線性問題的處理，因此若要處理非線性的資料問題，則可以使用非線性成分分析法（Nonlinear Principal Component Analysis, NLPCA），用來解決非線性的問題。而決策邊界之資料萃取法（Decision Boundary Feature Extraction, DBFE），亦可處理非線性的資料問題。本文將比較這三種特徵擷取方法的效能，並以負載辨識率最高、演算法所需之運算時間最短與所需之電腦運算資源這三項指標加以比較，以期找到一種最適合的特徵擷取方法。
並列摘要
Following the grown of the industry, it is increase in use of the any kinds of the load equipment, so the large power system will become more complication, and it will be increase in cost of the load monitoring. The virtual instrument (VI) application, which integrates all data acquisition, processing, display, and database management in one metering system, enhances the convenience, efficiency, flexibility, and economy of the power monitoring task. Because the power systems become more complication, how to ensure that the data of power waveform transfer is fast and correct, the chosen of the transfer interface is more important. However, the overall performance of the VI system is specifically limited by the bandwidth of the communication interface between the power meters and the host computer, when conventional serial communications are applied. The distributed VI power monitoring system will significantly enhance its real-time performance by interfacing the VI host computer via the universal serial bus (USB). LOW-COST power monitors are in great demand for industrial and commercial applications to assess power quality and manage demand and energy. Observing the energy consumption of electric loads, convectional load monitoring system needs to install hardware circuit on each load to be monitored. However, non-intrusive load monitoring system (NILMS) only needs to install a monitoring device on the electric power entrance point to collect the data for energy consumption of the loads by analyzing the signal waveforms collected and identifying the loads accordingly. And non-intrusive load monitoring system can be combined with the virtual instrument power monitoring system, so the cost is less than those of the conventional one. The thesis aims at developing the virtual instrument power monitoring meter, which transferred the data or command via universal serial bus (USB) interface. However, the techniques needed for load identification are more sophisticated as compared with the convectional one. So, how to choose the load feature is very important. In the convectional feature extraction method, one of the statistics method is principal component analysis（PCA）, the PCA method can solve the linear problem. Nevertheless, to solve the nonlinear problem, we should use nonlinear method, like nonlinear principal analysis（NLPCA）. The decision boundary feature extraction（DBFE） method is available for the nonlinear problem. In the thesis, we will compare with three kinds of feature extraction method, and which method can identify the load combination the highest, and the computer time and floats is the minimization, the method we will be proposed.