Aiming at the problem of dangerous behavior monitoring, a dangerous behavior intelligent monitoring system based on UWB wireless carrier communication technology was designed and developed. The system uses UWB positioning technology to complete real-time positioning of target locations by arranging UWB positioning base stations at home and wearing UWB positioning tags for users. It uses the YOLOv5s target detection algorithm combined with open source flame and smoke image data sets, MSCOCO data sets, and Google Open Image data set, etc., complete the monitoring of dangerous goods in safe areas. Using the AlphaPose human posture estimation algorithm, the Home Action Genome Dataset is selected as the training and test data set, which can monitor dangerous behaviors such as users climbing through windows or falling down. The system can issue alarms in a timely manner and transmit monitoring data in real time through connections with relevant equipment to ensure safety in all aspects.