Modeling of district load forecasting for distributed energy system

Distributed energy system (DES) has successfully aroused increasing interests among energy policy makers and system designers, as its potential of replacing conventional energy system. The optimal modeling of district load forecasting is essential to guarantee the best design and operation of DES. This paper presents a comprehensive review of district load forecasting (DLF) models to support the application of DES. The main factors affecting district load are discussed from inside to outside, including building indoor condition, building design characteristics, district layout, local microclimate, and social & economic factors. Through classifying and comparing top-down and bottom-up methods in terms of their key features and applications, it is found that the existing methods are either lack of forecasting accuracy or burdened with forecasting workload. Previous literatures reviewed in this paper show that the hybrid forecasting models including scenario analysis, physical-statistical numerical simulation and least square support vector machine based intelligent approaches have a superior ability to balance these two contradictions under different conditions. Based on the comparison results and current trend, a framework of district load forecasting, as well as corresponding future research work, is proposed for DES planning, design and service.