Remaining Useful Life Prediction With Partial Sensor Malfunctions Using Deep Adversarial Networks

In recent years, intelligent data-driven prognostic methods have been successfully developed, and good machinery health assessment performance has been achieved through explorations of data from multiple sensors. However, existing data-fusion prognostic approaches generally rely on the data availability of all sensors, and are vulnerable to potential sensor malfunctions, which are likely to occur in real industries especially for machines in harsh operating environments. In this paper, a deep learning-based remaining useful life (RUL) prediction method is proposed to address the sensor malfunction problem. A global feature extraction scheme is adopted to fully exploit information of different sensors. Adversarial learning is further introduced to extract generalized sensor-invariant features. Through explorations of both global and shared features, promising and robust RUL prediction performance can be achieved by the proposed method in the testing scenarios with sensor malfunctions. The experimental results suggest the proposed approach is well suited for real industrial applications.

I. Introduction

IN past years, due to the great merits of increasing machinery reliability, enhancing operational safety, and reducing maintenance cost, the field of prognostics and health management (PHM) has been attracting growing attention in both academic researches and real productions. A number of industrial scenarios have largely benefited from effective PHM applications [1]-[3], such as intelligent manufacturing, the automotive industry, the aero-space industry, and so on. Generally, existing popular PHM methods include physics-of-failure-based and data-driven approaches. In the recent years, with the rapid development of computing power and algorithms, the intelligent data-driven PHM techniques have been popularly applied in system predictive maintenance tasks, which require little prior expertise in advance and facilitate industrial applications. Through exploration of condition monitoring signals by using artificial intelligence, data-driven methods have achieved promising PHM performance, and have been more effective to meet the rising industrial demands of high reliability and efficiency.