Application of Artificial Intelligence Technology in Electromechanical Information Security Situation Awareness System

Abstract

The information security situational awareness system is proposed in this paper to leverage big data and artificial intelligence (AI) to enhance information security situation prediction. Deep learning techniques, specifically the long short-term memory recurrent neural network (LSTM-RNN), predict security situations using complex non-linear and autocorrelation time series data from current and past system conditions. Additionally, the study incorporates the variant gated recurrent unit (GRU) within the LSTM-RNN framework. A comprehensive experimental analysis is conducted, comparing various methods, including LSTM, GRU, and others, to assess and compare their predictive performance. The experimental results reveal that LSTM-RNN demonstrates a commendable level of predictive accuracy on the test dataset, with a mean absolute percentage error (MAPE) of 8.79%, a root mean square error (RMSE) of 0.1107, and a relative root mean square error (RRMSE) of 8.47%. Both LSTM and GRU exhibit exceptional predictive accuracy, with GRU offering a slightly faster training speed due to its simplified architecture and fewer trainable parameters. Overall, this research highlights the potential of AI-based methodologies in constructing robust information security situational awareness systems.