Cost-oriented load forecasting

Accurate load prediction is an effective way to reduce power system operation costs. Traditionally, the Mean Square Error (MSE) a common-used loss function guide training of accurate forecasting model. However, MSE unable precisely reflect real costs associated with errors because cost caused by in probably neither symmetric nor quadratic. To tackle this issue, paper proposes generalized cost-oriented framework. Specifically, how obtain differentiable that reflects and integrate regression models are studied. The economy effectiveness proposed method verified case studies optimal dispatch problem built on IEEE 30-bus open dataset from Global Energy Forecasting Competition 2012(GEFCom2012).