An adaptive load forecasting model in microgrids: A cloud-edge orchestrated approach tailored for accuracy, real-time response, and privacy needs

Abstract

The load forecasting tasks in different types of microgrids offer diversified requirements on application, such as forecasting accuracy, model complexity restrictions, and hardware environment. In our paper, typical load forecasting tasks in microgrids are classified into accuracy-oriented, real-time response and privacy-preserving type. An adaptive load forecasting model is proposed considering the trade-off between accuracy and efficiency by utilizing the customized AI algorithm and real cloud-edge orchestrated architecture. The decoupled module of forecasting model is considerably analyzed from accuracy impact and computing resource occupation, which arranges in different hardware environments to meet needs of different microgrid. Finally, the adaptive forecasting model is verified by the actual dataset from the MiRIS microgrid in Belgium. The proposed model can achieve satisfactory trade-off between accuracy and computation resource consumption, which meets the requirement for different types of microgrid load forecasting tasks.

Keywords

• Load forecasting
• Different types of microgrids
• Customized AI algorithm
• Cloud-edge orchestrated architecture