A CBL Estimation Method for Industrial Customers Based on K-Means and LSTM

In pursuit of attaining the objective of “carbon peaking and carbon neutrality”, the structure of energy in China has undergone a great transformation. With the rapid development of renewable energy generation, the power grid has been confronted with new challenges. The reliability and flexibility of the grid have been affected. Demand response (DR), as an effective management method, has become a hotspot. Customer baseline load (CBL) serves as the standard for calculating economic compensation in DR. Hence, accurate estimation of CBL is a crucial aspect in ensuring the effective implementation of DR strategies. This paper proposes a CBL estimation method for industrial customers based on the K-means and long short-term memory (LSTM). First, the K-means method is used to cluster the daily load curves of various industries to yield the typical patterns of electricity consumption behaviors. Second, the LSTM neural network is established to estimate CBL. It is trained with historical load data of different industries and clustering results. Finally, the effectiveness of the novel method is validated via a real data set.