The dynamic classification fusion algorithm using triangular fuzzy clustering based on multi-sensor information

Due to various complex environmental impact, coupled with unpredictable equipment failure, the data collected by a single sensor may be untrue inaccurate. Simple multisensor data fusion does also not accurately reflect the true state of the monitored object. An algorithm using triangular fuzzy variables as the distance of cluster analysis is developed to classify all data to some clusters according strong correlation after sensor data are cleanup on Grubbs rule. Then data fusions are done inside each of clusters using weighted data fusion. The algorithm takes advantage of integrations of data from different time and space to reduce the rate of false positives and rate of missing situations. The algorithm makes it possible to capture data classes with the strong correlation matching to the event dynamically and accurately, and improve the level of intelligence of the sensor network.