Deep Boltzmann machine based condition prediction for smart manufacturing

• Published in: Journal of ambient intelligence and humanized computing Vol. 10; no. 3; pp. 851 - 861
• Main Authors: Wang, Jinjiang, Wang, Kebo, Wang, Yangshen, Huang, Zuguang, Xue, Ruijuan
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.03.2019; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Artificial neural networks; Computational Intelligence; Deep learning; Engineering; Machinery; Manufacturing; Neurons; Optimization; Original Research; Particle swarm optimization; Probability distribution; Prognosis; Robotics and Automation; Time series; User Interfaces and Human Computer Interaction; Deep Boltzmann machines; Gaussian preprocessing; Predictive analytics; Smart manufacturing
• ISSN: 1868-5137; 1868-5145
• DOI: 10.1007/s12652-018-0794-3

Modern manufacturing systems are increasingly equipped with sensors and communication capabilities, and data-driven intelligence is gaining more popularity to analyze big manufacturing data. This paper presents a new deep neural network model based on Gaussian–Bernoulli deep Boltzmann machine (GDBM) for optimized condition prognosis. GDBM firstly uses Gaussian neurons to normalize the sequential input. Then, Extremum Disturbed and Simple Particle Swarm Optimization (tsPSO) method is introduced to optimize the model hyperparameters. Finally, a hybrid modified Liu–Storey conjugate gradient (MLSCG) algorithm is utilized to get a better rate of convergence, which makes the prognosis process being more computational efficient. Experimental study is conducted on condition prediction of a compressor in field, and the experimental results have shown that the presented model is able to obtain better performance over conventional data driven approaches.