Design of Anomaly Detection and Prediction Algorithm for Real-time Personnel Positioning System Based on Machine Learning

In dangerous or complex environments such as underground, real-time positioning and behavior monitoring of personnel is an important means to ensure safety and improve efficiency. Machine learning (ML) can process a large amount of data, extract useful features from it, and predict the position and behavior of people by learning these features. This article mainly studies the application of received signal strength indicator (RSSI) algorithm and back propagation neural network (BPNN) algorithm in underground personnel positioning system. Firstly, the signal strength data from different locations are received by RSSI algorithm, and the location information of personnel is calculated by using these data. Then, the BPNN algorithm is used to model and analyze personnel behavior, so as to realize abnormal detection and prediction of personnel behavior. The algorithm in this article is obviously superior to the traditional RSSI algorithm in accuracy performance, and the accuracy value is stable at about 95%. In addition, the algorithm also has a low running time, which shows that it has good real-time and scalability. The algorithm has good performance in real-time personnel positioning system, and can effectively improve the detection accuracy and efficiency.