METHODS FOR LOAD ANA PRICE FORECASTING IN COMPETITIVE MARKET

Abstract

Forecasting of electricity has always been the essential part of an efficient power system planning and operation, especially short term forecasts, as they are becoming increasingly important since the rise of the competitive energy markets. The aim of short term load forecasting is to predict future electricity demands based on historical data and other information such as temperature and humidity etc. It has many applications including energy generation, load switching, contract evaluation, and infrastructure development. The forecast of energy requirements i.e., load is a complex exercise. Accurate models for electric power load forecasting are essential to the operation and planning of a utility company. This thesis discusses the development of two load forecasting models, one using feed forward neural networks and the other one using radial basis networks. Both the models use wavelet decomposition techniques to further enhance the accuracy. In an electricity market, price forecasting is required by producers and consumers. Both producers and consumers use day-ahead price forecasts to derive their respective bidding strategies to the electricity market. Therefore, accurate price estimates are crucial for producers to maximize their profits and for consumers to maximize their utilizations. Forecasting electricity prices is difficult because, unlike demand series, price series possess characteristics like nonconstant mean and variance and sudden significant changes. The wavelet transform convert a price series in a set of constitutive series. These series presents better behaviors i.e., more stable variance and no outliers than the original price series, and therefore, they can be predicted more accurately. In this report, we have discussed a total of three developed price models, one using feed forward neural networks, one with self organizing maps and the other is a seasonal auto regressive moving average technique.