An Incremental ELM Method for Hourly Load Forecasting
Hanmin Sheng, Kai Chen, Kuang Hongjun, Ye Linhai, Yuanyuan Li · 2019 IEEE PES GTD Grand International Conference and Exposition Asia (GTD Asia) · 2019
Electrical load is affected by multiple factors. Machine learning has the ability to approximate complicate nonlinear relations, thus its application in load forecasting has received much attention. However, common machine learning technologies put high requirement for data consistency. Load varies periodically and non-periodically over time, to cover the influence factors is obvious unrealistic. To solve this problem, a load forecasting model based on ensemble learning is proposed. In this framework, knowledge is updated continuously to keep track with the latest data distribution. Information entropy is applied to measure the changes in data distribution. The evolutionary process is done through genetic algorithm. Considering the randomness of sub-model, extreme learning machine (ELM) is adopted. In the end, this method is verified through ISO-New England actual load data.