Novel decision-making technique for damage diagnosis

A study was conducted to implement the proposed method for damage diagnosis, validate the proposed method using experimental results for bearing damage diagnosis, and compare the proposed method with the known method. The proposed method was a supervised classification method based on the k-nearest neighbor (kNN) approach, clustering, anomaly detection and application of the novel weighted majority rule. The proposed method consisted of a number of stages, such as data clustering, calculating the novelty scores, anomaly detection, and damage detection and diagnosis. Data clustering allowed to consider the localization of data groups, which were representatives of the different aspects of normal behavior of the training data. The classical k-means algorithm was chosen for data clustering, and this algorithm created compact clusters at a low computational cost.