Market-clearing Price Forecasting In Day-ahead Market Based On Decision-tree Approach

Electricity price is the key factor for resource flow and benefit allocation in the power market. Electricity price issues are also the key problems and how to price the special commodity-electricity is essential for the smooth market operation.Forecasting the future electricity price is a very meaningful work and people have put forward some forecasting methods such as time series method, neural network, chaos theory etc. But electricity price forecasting needs considering not only the characteristics of time series of electric price but also the non-price factors' influences such as load command, anomalous days, supply and demand etc., especially for artificial elements. These factors result in the more hardness and deepness.This paper presents a successful application of a decision-tree to the market-clearing price forecasting for Queensland day-ahead electricity market in Australia (Pool trades model) . Taking Load and historical MCP into full account, principles of the decision-tree technique and its involvement in electricity price forecasting are summarized in detail. We construct MCP predict models using data mining function module of Microsoft Analysis Service. Our models will present different decision tree forms for different time ranges and display prediction processes and results in an easy understanding tree-like representation. Furthermore, user could easily refine the model introducing their own expert knowledge and practical experiences. As a conclusion, our model has the following advantages: relatively simpler predict process, visualized and understandable output, high accuracy and broad application fields.