“Cost-Accuracy” Hedging Based Load Forecasting Technique on Two-Stage Electricity Market

Abstract

Electricity plays an irreplaceable role in the national economy as a fundamental industry. Electricity contributes to the stable operation and scheduling of the power grid, promotes the efficient consumption of energy and avoids waste of resources. In most of electricity markets, the Load Serving Entities (LSEs) would submit the load scheduling by adopting model of load forecasting, which can be provided as a basis for trading in day ahead market. At present, most of load forecasting model focus on predicting accuracy instead of the fluctuation of market prices and LSEs’ benefits. This paper proposes a load forecasting strategy which balances accuracy and economic efficiency for two-stage electricity markets and establishes a “Cost-Accuracy” hedging based load forecasting technique (CAHFT). This technique is based on the traditional load forecasting technique, the term cost is introduced into the objective function, the improved backpropagation is used as the neural network for training. Case studies uses load data in New York area, and the verification results show that CAHFT has obvious effects on quantifying the benefits of the LSEs and contributing to the comprehensive improvement of its economic efficiency and accuracy.

Keywords

• load forecasting
• electricity market
• neural network
• lses' benefits
• improved backpropagation