No abstract

To extract strong correlations between different energy loads and improve the interpretability and accuracy for load forecasting of a regional integrated energy system (RIES), an explainable framework for load forecasting of an RIES is proposed. This includes the load forecasting model of RIES and its interpretation. A coupled feature extracting strategy is adopted to construct coupled features between loads as the input variables of the model. It is designed based on multi-task learning (MTL) with a long short-term memory (LSTM) model as the sharing layer. Based on SHapley Additive exPlanations (SHAP), this explainable framework combines global and local interpretations to improve the interpretability of load forecasting of the RIES. In addition, an input variable selection strategy based on the global SHAP value is proposed to select input feature variables of the model. A case study is given to verify the effectiveness of the proposed model, constructed coupled features, and input variable selection strategy. The results show that the explainable framework intuitively improves the interpretability of the prediction model.

No abstract

No abstract

This paper investigates network partition and edge server placement problem to exploit the benefit of edge computing for distributed state estimation. A constrained many-objective optimization problem is formulated to minimize the cost of edge server deployment, operation, and maintenance, avoid the difference in the partition sizes, reduce the level of coupling between connected partitions, and maximize the inner cohesion of each partition. Capacities of edge server are constrained against underload and overload. To efficiently solve the problem, an improved non-dominated sorting genetic algorithm III (NSGA-III) is developed, with a specifically designed directed mutation operator based on topological characteristics of the partitions to accelerate convergence. Case study validates that the proposed formulations effectively characterize the practical concerns and reveal their trade-offs, and the improved algorithm outperforms existing representative ones for large-scale networks in converging to a near-optimal solution. The optimized result contributes significantly to real-time distributed state estimation.