Hybrid clustering-time series-bayesian neural network short-term load forecasting method

摘要

This paper offers a hybrid short-term load forecasting (STLF) model using a Bayesian neural network (BNN) with a pre-processing stage consisting of a k-means clustering algorithm and time series analysis. The data clusters are time series analyzed to provide the most accurate data sets for each hour of the day. The final forecast is provided from the BNN output. California load data is used to determine the accuracy and processing speed of the proposed method. Additionally a comparison between BNN and other intelligent algorithms is provided using the same pre-processing stage to further gauge performance benchmarks.