Computer Sciences and Information Technology
Title : | Gaussian Process based Probabilistic Machine Learning method for Predictive Maintenance in Industry 4.0 |
Area of research : | Computer Sciences and Information Technology |
Focus area : | AI and Machine Learning |
Principal Investigator : | Dr. Rajnish Mallick, Thapar Institute Of Engineering & Technology, Punjab |
Timeline Start Year : | 2023 |
Timeline End Year : | 2026 |
Contact info : | drrajnish.mallick@gmail.com |
Details
Executive Summary : | Gaussian process regression is a powerful, non-parametric Bayesian method. It is a class of regression problems that can be utilized in predictive analytics for remaining useful life estimation in aerospace and mechanical sciences industry. The task of remaining useful life (RUL) estimation is a major challenge within the field of prognostics and health management (PHM). The quality of the RUL estimates determines the economical feasibility of the application of predictive maintenance strategies, that rely on accurate predictions. Therefore, myriad effective methods for RUL estimation have been developed in the recent years. Especially deep learning methods have evolved as one of the best performing RUL estimation approaches for potential Industry 4.0 applications. These machine intelligence algorithms have also demonstrated new record accuracies on bench mark data sets. However, these deep learning approaches need intensive data-inputs and often rely on large volume of run-to-failure sequences of the components under investigation. The run-to-failure sequences and RUL labels are highly scarce, either missing or extremely expensive to capture in real-world use cases. In this investigation and proposal, the objective(s) are to develop a new, data-efficient method, which is based on Gaussian Process regression for RUL estimation utilizing the Bayesian machine learning principles. The proposed approach does neither rely on entire run-to-failure sequences nor on any RUL labels and will be tested on the benchmark NASA C-MAPSS turbo fan data set. The predictive analytics results will be compared with the state-of the art machine learning methods while using only sparse available training dataset. Therefore, the proposed approach allows RUL estimation in various Industry 4.0 use cases, in which gathering enough component or system level failure data is intractable. |
Total Budget (INR): | 6,60,000 |
Organizations involved