One-class Classification Fault Detection Method Based on Manhattan Distance and Bagging

In response to the difficulty of obtaining fault samples in avionics equipment fault detection, a bagging ensemble classification algorithm based on Manhattan distance is proposed by combining the Bagging ensemble learning framework with similarity score. This algorithm achieves avionics component fault detection under only normal sample training conditions. Based on normal data samples, an algorithm flow for bagging ensemble classification based on Manhattan distance was proposed, and classification experiments were conducted on avionics equipment detection data. The final results demonstrate that the proposed method has better detection performance.