Modeling and Forecasting of Industrial Consumers’ Load Using Fuzzy Clustering and Advanced Neural Networks
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Effective electrical load management is essential for enhancing power system stability and minimizing energy costs. This study presents a methodology for analyzing and forecasting electricity consumption patterns. Initially, consumption data were examined using the Fuzzy C-Means (FCM) clustering algorithm. The fuzzy approach was chosen for its ability to handle uncertainties and overlapping data, allowing for the assignment of membership degrees to each cluster and providing a more precise analysis of complex consumption patterns. Cluster centers, representing typical consumption behaviors, were subsequently used for future load forecasting employing two advanced deep learning models: Long Short-Term Memory (LSTM) and Bidirectional Temporal Convolutional Network (BTCN). Evaluation based on RMSE and MAE metrics indicated that BTCN, leveraging its bidirectional structure and enhanced temporal learning capabilities, outperformed LSTM in prediction accuracy. This methodology offers a powerful tool for load management and the design of optimal electricity consumption strategies. Finally, suggestions for further research and potential improvements are discussed.