Isomap and deep belief network-based machine health combined assessment model/Kombinirani model za ocenjevanje stanja strojev na osnovi tehnike Isomap in globoke verjetnostne mreze

This paper presents a novel combined assessment model (CAM) for machine health assessment, in which 38 original features of the vibration signal were extracted from time domain analysis, frequency domain analysis, and wavelet packet transform (WPT), following which the nonlinear global algorithm Isomap was adopted for dimensionality reduction and extraction of the more representative features. Next, the acquired low-dimensional features array is input into the well trained deep belief network (DBN) model to evaluate the performance status of the bearing. Finally, after the bearing accelerated degradation data from Cincinnati University were investigated for further research, through the comparison experiments with two other popular dimensionality reduction methods (principal component analysis (PCA) and Laplacian Eigenmaps) and two other intelligent assessment algorithms (hidden Markov model (HMM) and back-propagation neural network (BPNN)), the proposed CAM has been proved to be more sensitive to the incipient fault and more effective for the evaluation of bearing performance degradation. Keywords: Isomap, dimensionality reduction, deep belief network (DBN), machine health, combined assessment model (CAM) Nadzor in vrednotenje trendov degradacije nekaterih kljucnih strojnih delov kot so lezaji omogoca odpravo degradirane zmogljivosti ali napak se pred okvaro stroja. Ker pa kolicine zbranih podatkov o strojih postajajo vse obilnejse in ker so vse strozje tudi zahteve glede hitrosti in tocnosti vrednotenja stanja strojev, tradicionalne metode ne jamcijo vec za ucinkovito delo. Isomap kot tehnika za nelinearno globalno transformacijo dimenzionalnosti poisce resitev preslikave z vrsto pretvorb, ki omogocijo predstavitev geodezicne razdalje med vhodnimi tockami v izvirnem prostoru z evklidsko razdaljo v prostoru projekcije. Globoka verjetnostna mreza (DBN) kot probabilisticni generativni model, ki uspesno zajema znacilne informacije v surovih podatkih z raznimi nelinearnimi transformacijami in kompleksnimi aproksimativnimi nelinearnimi funkcijami, je primerno orodje za vrednotenje stanja strojev. Po primerjavi in analizi pomanjkljivosti obstojecih metod je v clanku predstavljen novi kombinirani model vrednotenja (CAM), ki zdruzuje WPT, Isomap in DBN za vrednotenje stanja obravnavanega stroja (oz. njegovih kotalnih lezajev). Podatki o pospeseni degradaciji lezajev so bili zbrani s preizkusanjem lezajev pod stalno obremenitvijo do odpovedi na posebnem preizkusev