Optimization method for load aggregation scheduling in industrial parks considering multiple interests and adjustable load classification

With the development of new power systems, the load scheduling and management of high-energy-consuming users in industrial parks have become increasingly complex. This paper proposes an optimization method of load aggregation scheduling for industrial parks. First, an industrial load characteristic classification model is developed, analyzing and categorizing the load characteristics of five energy-intensive industries. An optimization model for load aggregation scheduling is then constructed, considering factors such as spot markets, time-of-use pricing, and the interests of both the load aggregator and the plants. To accurately predict the baseline load of plants, an improved forecasting model, an improved temporal convolutional network based on the variational mode decomposition optimized by the improved whale optimization algorithm is proposed, which outperforms four other models, improving the average MAPE by 21.64 %, 30.25 %, 83.42 %, and 97.18 % across three test datasets. Finally, simulations based on real scenario and data from a case city were conducted using Gurobi for model solving and sensitivity analysis. Results show that the optimized solutions improved the objective function by 0.93 %, 1.03 %, and 0.72 %, demonstrating that the proposed method meets demand response requirements and significantly boosts the economic benefits for both the Load Aggregator and the plants.