A two-stage centralized energy scheduling by mining demand response potential in a renewable energy-based Industrial Park

Renewable Energy-based Industrial Parks (REIP), characterized by their substantial electricity consumption and predictable usage patterns, stand out as prime flexible resources, offering considerable demand response (DR) potential. By thoroughly mining DR potential based on REIP flexible resources, the utilization of renewable energy and flexible resources within the REIP can be optimized. Accurate mining of DR potential serves as the data foundation for the REIP economic scheduling. Therefore, a meticulously designed multi-time scale scheduling strategy for the park becomes pivotal in achieving a harmonious balance between energy supply and demand, reducing energy costs, and enhancing renewable energy utilization efficiency. Based on this premise, this study proposes a two-stage centralized energy scheduling approach that integrates the DR potential through load profile analysis. Initially, the Density-Based Spatial Clustering of Applications with Noise Method is employed to preprocess load and photovoltaic (PV) data. Following this, the primary types of electricity consumption plan within the REIP are identified using an enhanced ordered clustering algorithm, and load curve profiling is conducted based on the maximum correlation minimum redundancy criterion. Subsequently, an Improved Seasonal Autoregressive Integrated Moving Average model is utilized to forecast load demand using typical electricity consumption datasets, thus determining the adjustable capacity of typical users during the scheduling period. Finally, a Nonlinear Affine Adjustable Robust Optimization method and a Neural Network-based Model Predictive Control approach are applied to achieve a two-stage centralized energy scheduling process that addresses the uncertainty in load demand and PV output. The effectiveness of the proposed energy scheduling framework is verified by an actual REIP in Chengdu, Sichuan, China.