Hybrid algorithm for detecting intrusion with optical fiber sensor in rainy weather.

Fiber optic intrusion detection is recognized as one of the most effective perimeter defense strategies against intruders and terrorists. Most current infiltration detection systems gather information via optical cables and trace invasions using classification techniques. This research introduces two intrusion detection algorithms and a hybrid approach to address the challenges of unstable and hard-to-extract fiber optic infiltration features, especially during rainy weather. The first method, known as the phase parameter-matching algorithm, calculates the system's disturbance covariance at any position and uses the Frobenius norm to identify intrusions by measuring the difference between disturbance covariances at distinct sites. The second method, the cross-rebuilding algorithm, employs the Kalman filter to reconstruct the output of each point based on the non-infiltration state parameters, detecting intrusions through cross-reconstruction error. Experimental tests were conducted to evaluate both methods. The results indicate that both methods are viable, with the phase parameter-matching algorithm being more robust but slower, and the cross-rebuilding algorithm being faster but less reliable. The hybrid algorithm combines the strengths of both approaches, offering a balanced solution with moderate computation time and enhanced resistance to environmental interference, leading to improved overall detection accuracy and reduced false positive rates. [ABSTRACT FROM AUTHOR]