A Comparative Analysis of Supervised/unsupervised Algorithms in PHM Application

Intelligent learning algorithms, as one of the most currently remarkable data-driven approaches, have achieved great success in many applications, such as fault diagnosis, state evaluation and life prognosis of airborne equipment. This paper mainly introduces the classification and typical principles of supervised and unsupervised learning algorithms, and compares their differences in the application of system health status assessment. Through the establishment of K-means clustering model and artificial neural network model, the data obtained from Monte Carlo analysis are processed and classified, and then the health state of the system is estimated. This paper verifies the superiority of the supervised algorithm in evaluation accuracy through the experimental data, and analyzes its essential reason. The conclusion can provide guidance for the engineering application of supervised and unsupervised learning algorithm.