A Hybrid Edge-Cloud Computing Method for Short-Term Electric Load Forecasting Based on Smart Metering Terminal

As the edge node in electric Internet of Things, the application of smart metering terminal (SMT) enables the massive electric big data to be widely collected and processed on the edge. This creates a positive condition for short-term electric load forecasting, which is very important for electricity sales company under the back ground of electricity spot market. In this study, the structure of a novel hybrid edge-cloud computing framework for electric load forecasting is proposed, and the forecasting model based on extreme learning machine (ELM) is also developed. In the proposed framework, the distributed SMTs are regarded as the edge nodes and widely collect electric data inner electricity customers. Then, these original data are preprocessed in the regional SMT, and then sent to the cloud server as standard time series data. Finally, the proposed ELM forecasting model runs in the cloud server, and outputs the forecasting load demand for all the customers of electricity sales company. Experimental results show the efficiency of the proposed framework and ELM forecasting model.