A meta-learning based distribution system load forecasting model selection framework

• We introduce meta-learning concept to achieve distribution system load forecasting model selection. make the framework rigorously formulated, applicable and extendable. propose a scoring-voting mechanism improve selection accuracy. create comprehensive test cases validate proposed methodology. This paper presents based, automatic framework. The includes following processes: feature extraction, candidate preparation labeling, offline training, online recommendation. Using needs data characteristics as input features, multiple metalearners are used rank forecast models based on their Then, is weights recommendations from each meta-leaner final recommendations. Heterogeneous tasks with different temporal technical requirements at aggregation levels set up train, validate, performance of Simulation results demonstrate that approach satisfactory in both seen unseen tasks.