Enhancing Fault Detection in Power Load Management Systems: A Neural Network Approach with Seq2Seq-CNN-LSTM Models

This paper presents a groundbreaking approach for fault detection in electric power load management systems using an advanced neural network. This network uniquely integrates Seq2Seq, CNN, and LSTM models to enhance predictive accuracy and fault classification. The Seq2Seq model, by learning from historical data, facilitates the prediction of future sensor data, thus providing the ability to forecast system states. Concurrently, the CNN model is utilized for extracting features from the predicted data, capturing potential anomalous patterns. Meanwhile, the LSTM model is responsible for classifying these features and detecting potential faults. Through extensive experimentation, the proposed network demonstrates its capability to detect system faults preemptively by combining prediction and classification methods. This approach significantly contributes to improving the reliability and efficiency of electric power load management systems.