A Neural Network Anomaly Detection Model Based on Archimedes Optimization Algorithm for Intelligent Operation and Maintenance

The operation and maintenance data abnormal detection technology is currently hampered by difficult detection procedures and low accuracy. The primary purpose of this paper is to investigate and address the problem of low accuracy in the abnormal detection of operation and maintenance data by combining artificial neural network models with deep learning technology. We choose long and short term memory (LSTM) and Encoder-Decoder LSTM model as the object for further optimization because they perform better in data anomaly identification than other neural network models in many measures. Also, by altering model parameters like loss rate, learning rate, training epoch, and batch size, artificial neural network model efficiency can be increased. Therefore, we propose an artificial neural network model that combines the recently developed Archimedes optimization algorithm (AOA). By using the AOA algorithm to optimize the model parameters, we can make the LSTM and Encoder-Decoder LSTM models reach a better state. The optimized model performs better than the non-optimized model in several measures of anomaly detection, with accuracy and precision showing the most pronounced improvement. In addition, the recall rate, classification effect, and error between the predicted value and real value all show significant improvements in the optimized model. By applying the optimized model, the anomaly detection accuracy of the ordinary computer room operation and maintenance data can achieve good effects, and provide the possibility for the model to be widely used in the actual intelligent operation and maintenance data anomaly detection.