DEVELOPMENT OF NEW TOOLS TO PROMOTE A MORE EFFECTIVE CONSUMER PARTICIPATION IN SHORT-TERM ELECTRICITY MARKETS

摘要

This paper summarizes the research work performed to show the capability of a combination of tools based on Self-Organizing Maps (SOM) and Physically Based Load Models (PBLM) to classify and extract patterns from distributor, aggregator and customer electrical demand databases (the objective known as data mining). This approach basically uses low cost information available for almost all supply side agents: historic load curves of several kinds of customers. The first objective is to find a correlation between demand and the evolution of energy prices in short-term energy markets. A SOM was trained that should allow to select the most suitable customer clusters whose demand modification would benefit customer and supply-side agents through, for example, energy efficiency, distributed generation or demand response. After a previous evaluation through PBLM of different possible strategies to reduce demand during consumption peaks, a SOM was trained to detect opportunities among users with high reduction capabilities during periods when day-ahead prices are lower than shorter-term prices. The results obtained clearly show the suitability of SOM approach to find easily coherent clusters between electrical users with high demand or available response capacity, and therefore a possible way to promote customer participation in electrical energy markets is opened.