Hybrid Computational Intelligence Model for Short-Term Bus Load Forecasting

Distribution Generation (DG) technologies correspond to a technical field of increased interest since their application aid on system security and reliability. In order to bring forth the fully potential of DG, a robust forecasting tool is especially designed for small sized loads at the various buses. Bus load exhibit low correlation compared to the total system's load; the presence of outlier loads is more regular and the load pattern presents high degree of stochasticity. Thus a load forecasting model designed for the system's load is likely to show poor performance. This work proposes a hybrid bus load forecasting tool. The hybridization refers to the combined use of a clustering process with a feed-forward Artificial Neural Network (ANN). The proposed model is tested at four buses within the Greek interconnected system and simulation results highlight the efficiency of the model.