ISOMAP Algorithm-Based Feature Extraction for Electromechanical Equipment Fault Prediction

As for the difficult problem of sensitive feature extraction during fault prediction for nonlinear electromechanical equipment, nonlinear dimensionality reduction ISOMAP (isometric feature mapping) algorithm is introduced based on the comprehensive analysis of the operation state data to reduce the dimensionality of high dimensional operation data and acquire sensitive fault features, furthermore, the ISOMAP dimensionality reduction result is verified by compiling algorithm programs based on MATLAB platform. The suitability of applying ISOMAP algorithm to dimensionality reduction of high dimensional data is discussed in terms of algorithm principle in this paper, and calculation steps of the algorithm are provided. According to MATLAB tests, ISOMAP algorithm is able to reduce dimensionality greatly and find out the essential data structure so as to provide a new method for extracting sensitive features of electromechanical equipment fault from another point of view.