| Summary: | 碩士 === 國立中興大學 === 資訊管理學系所 === 103 === The traditional electric power market focuses more on the development on the supply side and has long neglected the information of price on the demand side, such as users’ outage cost, preference and willingness to pay (WTP). It was not until the two oil crises in the 1970s and the increasingly serious “not in my backyard (NIMBY)” effect that have the power companies around the world start realizing that the costs of power development increase dramatically. On top of that, the environmental consciousness that advocates energy saving and CO2 reduction makes people realize that the demand-side management (DSM) is something that needs considerations.
Thanks to the rapid progress in energy information communication technology (EICT) and the popularization of advanced metering infrastructure (AMI), users are now given the access to power consumption information through smart meter, and spontaneously able to start with power saving, power suspension and restoration management and the asset management of metering equipment. In addition, Taiwan Power Company (TPC) completed the establishment of AMI in 24,000 high voltage power users and 10,000 low voltage users at the end of Oct 2013. From 2016 onwards, the plan is to install single-phase smart meters at 5,000,000 low voltage users’ homes. Big data is coming as a result of this installation plan. It is estimated that it is necessary to read: 1 meter (9 indices) x 96 (4 times (once/15min) x 24hr) x 30 days x 24,000 users ＝ 620,000,000 indices per months for high voltage users alone. These readouts will allow for more delicate and refined analysis on the demands of users.
The study was focused on the power consumption data of high voltage users with a contract capacity of 500kW or more. High voltage users have been provided with smart metering systems and their consumption of power accounts for 60% of the total power generated in Taiwan. The study on the attributes of these users and their preferences in power consumption behaviors may provide a solution for the reduction of power loads at peak hours together with appropriate power price discount packages.
The 351 valid questionnaires of high voltage users’ power consumption behaviors were collected using questionnaire survey in Oct 2013.First, data pre-processing and conversion were conducted on the data of 153 users who chose to participate in one of the discount packages. Then, k-means, which is a technique of cluster analysis, was used to separate attributes of user groups and the tendency in the power consumption behaviors. The model obtained from data mining was used to determine how the results were presented and explain and evaluate the results. Finally, for the 198 high voltage users who have not yet chosen any of the discount packages, excluding the invalid data of 4 users who have not filled in the fifth and sixth part of questionnaire, parameters that were related to this study was used to identify to which groups these users belonged, and appropriate power DSM solutions were suggested to them in order to achieve effective use of power.
|
|---|
