The Classification of Customer Load Pattern

• Main Authors: Guo-Sin Huang, 黃國欣
• Other Authors: Tai-Ken Lu
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/90776151450684710120

碩士 === 國立臺灣海洋大學 === 電機工程學系 === 96 === Market restruction and deregulation for electric utilities are the global trend in the past years, which penetrate the open market competition mechanism and provide the good electricity service, as well as promote the living level of the people. However, the electrical industry liberalization brings the impact which deregulates the traditional electrical industry to the power generation company, the power transmission company, the power distribution company, and the selling electricity industry. As a result of the deregulation of power industry, the unit commitment, the load forecast, the spot price of electricity, the generation planning schedule and so on, have all faced the new innovation. Therefore, from the customer’s aspect, suggests the appropriate electric power products to the customer based on the forecast of the future electric power supply and demand, will be the important research issue. This thesis studies the load pattern, which is the reference data for the electricity price formulation, the load forecast, and the management strategy planning and so on. But the user’s data are too many to the load patterns are also many, so the representative load pattern finding is the key point of this research. The discussion of the load pattern has two parts in this thesis: one part is to use the Autocorrelation Function(ACF) analysis analysis method to find the load pattern of the original series; then use Cross-Correlation Function(CCF) analysis method to find the similar load patterns; finally, adjustment the threshold of Euclidean distance to decide the load pattern cluster, and average these load patterns as the representative load pattern of this cluster. The other part is to find the attribute materials of load pattern, and then to find the characteristic attribute material classification criterion using Data mining technology. The representative load patterns can be used as reference for the unit commitment, the load forecast, the spot price of electricity, the generation planning schedule and so on.