Equilibrium Optimizer-Based Joint Time-Frequency Entropy Feature Selection Method for Electric Loads in Industrial Scenario

A prerequisite for refined load management, crucial for intelligent energy management, is the precise classification of electric loads. However, the high dimensionality of electric load samples and poor identification accuracy of industrial scenarios make it difficult to be used in actual production. As such, this research presents a selection approach equilibrium optimizer-based joint time-frequency entropy feature selection method for electric loads in industrial scenarios to address these issues. The method first introduces entropy value features based on extracting time-frequency domain features and then uses an equilibrium optimizer (EO) to screen the joint feature set. A Chinese cement plant was chosen as the acquisition site for the experiments, and the low-frequency data from power equipment were gathered to form an original dataset for power analysis. The features screened by the EO were used as model inputs to verify the effectiveness of the EO on the joint feature set under K-nearest neighbor (KNN), support vector machine (SVM), decision tree (DT), random forest (RF), and discriminant analysis (DA) models. Experimental results show that introducing entropy value features for the joint feature set can significantly improve the classification performance. The average accuracy of the features screened by the EO was as high as 95.58% on SVM, while the computation time was 0.75 s. Therefore, for industrial electricity scenarios, the approach suggested in this research can enhance the identification accuracy of electric loads and significantly reduce the computation time of the model to a great extent. This has essential research significance for intelligent energy management in real industrial scenarios.