Aiming at the prominent indoor safety hazards and the difficulty in timely detecting abnormal behaviors of elderly living alone against the backdrop of accelerating population aging, an abnormal behavior monitoring system for elderly living alone indoors, integrating multi-sensor technology and intelligent algorithms, is designed. The system adopts a four-layer architecture of “perception layer–processing layer–communication layer–application layer,” integrating hardware modules for visual collection, environmental sensing, physiological monitoring, and posture perception, and realizing collaborative data processing through embedded processors and edge computing devices. At the algorithm level, it optimizes target detection, behavior classification, posture fusion, and multi-modal discrimination models to achieve real-time identification of risks such as falls, posture abnormalities, and physiological abnormalities. The communication link combines short-range and long-range technologies, coupled with a multi-terminal early warning mechanism to ensure efficient transmission of abnormal information. The system balances detection accuracy, privacy protection, and user acceptance, providing technical support for the home safety of elderly living alone. The relevant design ideas can serve as a reference for the development of similar intelligent elderly care equipment.