Multi-user Power Load Forecasting Based on K-means and Deep Neural Network

• Published in: 2023 IEEE 3rd International Conference on Electronic Technology, Communication and Information (ICETCI) pp. 508 - 512
• Main Authors: Zhang, Haitao, Zhu, Chuanjing, Yang, Shaochun
• Format: Conference Proceeding
• Language: English
• Published: IEEE 26.05.2023
• Subjects: Artificial neural networks; Clustering algorithms; Computational modeling; deep neural networks; Distributed databases; Feature extraction; forecasting; K-means clustering; Load forecasting; multi-user power load data; Power system stability; spark distributed computing engine
• DOI: 10.1109/ICETCI57876.2023.10176656

Short-term power load forecasting is an important issue in power system management and planning, and is of great importance for enhancing the efficiency and stability of the power system. Traditional short-term power load forecasting systems are mostly based on historical load data of a single region or user which cannot meet the demand for load forecasting of multiple users with different electricity consumption characteristics. This paper proposes a method that integrates big data distributed computing, clustering algorithms, and deep neural networks(DNN) for preprocessing, feature extraction, classification and prediction of multi-user power load data. Firstly, we use the Spark distributed computing engine to preprocess and extract features from historical load data. Then, we use the K-means clustering algorithm to classify users into different classes based on their characteristics. Finally, we design deep neural network models for different categories, then we train, validate, and test the models with a real user load dataset from a region of China. The results show that the proposed method has high accuracy and can meet the demand for multi-user power load forecasting.