Research on Distributed Intrusion Detection Model for Nuclear Facility Spatial Fields Based ...

Research on Distributed Intrusion Detection Model for Nuclear Facility Spatial Fields Based on Synesthesia Integration Technology

To address challenges in nuclear facility perimeter security, such as environmental interference, high false/missed alarm rates, and insufficient multi-target positioning...

To address challenges in nuclear facility perimeter security, such as environmental interference, high false/missed alarm rates, and insufficient multi-target positioning accuracy, this paper proposes a distributed intrusion detection model based on integrated sensing technology. The model combines intelligent detection cables and microwave sensing via a hardware layer to form a dynamic spatial perception network. At the algorithm level, it incorporates multi-point signal joint detection and end-toend AI recognition, utilizing spatiotemporal feature fusion to accurately classify and locate intrusions. Key innovations include a distributed hardware architecture compatible with domestic platforms (e.g., Haiguang CPU, Kirin OS), a Dynamic Field Strength Compensation Algorithm (DFCA) to mitigate environmental interference, and a dual-stream network for feature extraction-one analyzing cable signal disturbances, the other integrating device status and environmental parameters. Experimental results show a positioning accuracy of

, a false alarm rate below 3% in heavy rain, and 98.7% accuracy in multi-target recognition. The system also improves resource utilization by 42 % and reduces response delay to 200 ms, complying with nuclear security standards (EJ/T 20180.1-2018) and enhancing perimeter protection reliability and intelligence.

2025 IEEE 7th Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC)

With the continuous improvement of the safety level requirements of nuclear facilities, the front-end perimeter intrusion detection system needs to achieve high-precision and low delay real-time monitoring in extreme environments , strong electromagnetic interference, rainstorm/dust). The current nuclear security standards (such as EJ/T 20180.1-2018) require the system to have multi-target concurrent recognition and anti-interference capabilities, but the traditional scheme faces severe challenges: a single sensing mode (such as laser or vibrating optical fiber) is significantly affected by environmental mutations, the false alarm rate of rainstorm scene exceeds 15 %, and the false alarm rate caused by equipment aging increases by 3.5 % annually; Multi target positioning relies on sparse sensor deployment, and the positioning error is more than 5 m, which is difficult to meet the high-precision protection requirements of nuclear facilities. In addition, the timespace asynchrony of heterogeneous sensor data (microwave signals, environmental parameters) leads to the analysis delay of more than 500 ms, which seriously restricts the timeliness of early warning[1] [2].

References is not available for this document.