Interval Valued Intuitionistic Fuzzy Logic for Anomaly Detection in Public Health Monitoring Signals

Abstract

Interval valued fuzzy logic can more flexibly and comprehensively characterize and handle uncertaintyand fuzziness in information by extending the membership degree of fuzzy sets from a single numericalvalue to a closed interval form. This provides a new technological path to solve the complex processingproblems of public health monitoring signals. This article focuses on the application of interval valuedfuzzy logic in public health monitoring signal processing. Firstly, the sources and characteristics ofuncertainty in public health monitoring signals were analyzed, and the adaptability advantage ofinterval valued fuzzy logic was clarified. Subsequently, a processing model based on interval valuedfuzzy logic was constructed from three key steps: signal denoising, feature extraction, and anomalydetection. This study used monitoring data of influenza like cases in a certain region from 2023 to 2024,collecting a total of 1200 continuous monitoring signals. The data types included normal signals (60%),abnormal warning signals (30%), and noise interference signals (10%). In the anomaly detection stage,establish interval value fuzzy anomaly judgment rules, match the fluctuation range of monitoring signalswith the public health safety threshold in intervals, and achieve early warning of public healthemergencies. The experimental group adopts interval valued fuzzy logic algorithm, which enhancesuncertainty handling ability by expanding the membership degree of fuzzy sets into closed intervals; Thecontrol group used traditional fuzzy logic algorithm and BP neural network algorithm. Compared withtraditional fuzzy logic and BP neural network algorithms, the processing model based on intervalvalued fuzzy logic improves the accuracy of uncertainty information processing by 15% -20%. In termsof false positive rate and noise processing, it maintains high stability in complex scenarios with lowsignal-to-noise ratio and incomplete data. At the same time, it performs well in alarm response time,and its signal recognition accuracy and noise filtering efficiency are better than the control groupalgorithm.