Prognostics and health management (PHM) increasingly play a constructive role throughout the entire lifetime of industrial equipment, significantly benefiting from extensive research in physical modelling and machine learning techniques. This has led to the development of hybrid approaches that seamlessly integrate both domains through physics-informed machine learning (PIML). PIML ensures the generation of cohesive solutions encompassing various aspects of physics knowledge across different stages of the machine-learning pipeline, substantially contributing to detection, diagnostics, and prognostics. However, PIML’s design relies heavily on expert experience and demands rigorous interdisciplinary expertise, requiring a profound understanding of machine learning and physical principles. Inadequate design of PIML often leads to suboptimal outcomes, where the combined effect is less than the sum of its parts. Currently, PIML lacks a scalable and engineered application architecture to effectively utilise its embedded results. To address this challenge, our paper introduces a novel parallel architectural approach that employs pre-training and fine-tuning strategies for optimising the different model parts. Its data-driven branch is first trained with zero output in the PI branch, then fine-tuning the physics-informed branch. It takes the frozen data-driven model as a fixed feature extractor to get physics-consistency prediction. This approach proposes a generic solution for embedding physics knowledge into ML that guarantees performance improvement. The effectiveness of our approach is validated in the context of Remaining Useful Life (RUL) prediction using MIT-Stanford battery data.

Rotating machines play a vital role in many industrial applications, ranging from power generation to manufacturing. The early detection of mechanical faults in industrial rotating machines is crucial for enhancing the reliability and safety of industrial systems. This research focuses on rotating machines that operate under varying working conditions, which exhibit specific characteristics that make the development of practical data-driven fault diagnostics methods challenging. The paper proposes a methodology that aims to develop robust techniques that can operate effectively in real industrial environments. To simulate rotating machinery operating under varying working conditions, an experimental protocol is conducted, and the resulting vibration signals are analyzed. Finally, the study describes an approach to train data-driven methods based on realistic data availability scenarios.

In the prognostics and health management (PHM) of industrial systems, prediction of remaining useful life (RUL) is a crucial task. RUL prediction is based on data collected from the industrial system, and involves learning underlying health indicator trends. As industrial systems are complex and can be monitored by different sensors, time alignment of multiple temporal data streams and extraction of their underlying characteristics are essential to perform an accurate prognostics. Hence, this paper aims to develop an efficient method to address the above issue. The proposed method is based on the attention and convolution mechanisms of deep neural networks. Its performance is highlighted when compared to other state of the art models such as RNN and LSTM using the C-MAPSS datasets. Numerous experiments demonstrate that our model provides better results in some situations, as well as an ability to capture both local short term contexts and long term associations.

Fault detection and diagnostics are important steps in the predictive maintenance of industrial systems, especially faults in the mechanical parts most susceptible to fail, such as bearings and gears in rotating machines. These two components represent more than 50% of causes of the operational downtime. Therefore, the detection of their appearance allows anticipating the total failure of the machine and schedule in advance maintenance actions. However, in the presence of a combined gear and bearing faults, it is difficult to isolate their states. To remedy this situation, this paper proposes a data processing methodology that exploits the three-phase current signals of the rotating machine and build a health indicator (HI) from each current phase that reveals the different health states. This indicator is constructed by extracting features from the collected raw data in frequency and time domains, and then they properly combined with a physical significance. After that, all health indicators (HIs) of the three phase current data are fed to a machine learning model for an online pattern recognition of the bearing and gear states, including the combined faults. The proposed approach is demonstrated through a test bench that studies bearing and gear defects of a gearbox under different operating conditions.

During the two last decades, the failure prognostics has been given an increasing importance by the researchers and industrials thanks to the benefits its results can induce in terms of availability, security, and maintenance costs. However, a detailed study of the literature in the domain shows that most of the published works deal with prognostics at component level, where the failure of the system is linked to the failure of its critical components. Even if this assumption can be tolerated in some cases, it cannot be generalized. This is because, in practice, the components are interconnected and, consequently, a degradation of one component can affect the others leading to a modification in the system's health state and then in its remaining useful life (RUL). To take into account this reality, a new model for prognostics is proposed in this article, namely inoperability input-output model (IIM). This model allows representing multicomponent systems while considering the interactions between their components. More than that, the IIM enables to integrate the mission profile effects on the system degradations to make its RUL predictions more accurate. To illustrate the effectiveness of the proposed modeling approach, a mechatronic system is considered as a case study.

Democratization of sensing devices in industrial systems has made it possible to collect a large amount of data of different types, which has led to the necessity of handling complex analyses for knowledge extraction . The field of water resources is of those areas which has drawn the attention of decision-makers seeking to preserve human health and safety. Recent advances in Artificial Intelligence, particularly in the domain of Machine Learning, have opened the potential to leverage massive data to better address the issue related to the relationship between water quality and human activities. However, high rate of missing data and heterogeneity of the measurements are scientific issues that cannot be solved by standard methods, especially when no prior knowledge on the label of each observation is provided. In this article, Prognostics and Health Management was implemented to detect and diagnose anomalies in water quality datasets, taking into account the uncertainties induced by the above-mentioned issues. Fuzzy c-means was used to identify the different water quality classes, while Random Forest was applied to determine the most influencing parameters, with respect to potential contamination of water resources in the southwest of France. The results suggest that multiple imputation methods can handle the missingness issue, while the use of decision rules based on well-known water quality standards can solve the problem regarding the lack of labelled observations. In addition, two potential sources of contamination (atrazine and nitrate) were identified and then validated by hydrogeology experts, prior to further online deployment of the proposed model.

Nowadays, companies producing goods use production systems that are equipped by different sensors in order to monitor efficiently their behavior. Most of the time, the information collected by these sensors is mainly used for production monitoring rather than to analyzing the state of health of the production system. By so doing, these companies have a large and growing amount of data at their disposal. These data make it possible to extract information and knowledge for a better control of the system in order to improve its efficiency and reliability. With the emergence of Prognostics and Health Management (PHM) paradigm few years ago, it has become possible to study the state of health of an equipment and predict its future evolution. Globally, the principle of PHM is to transform a set of raw data gathered on the monitored equipment into one or more health indicators. In this framework, the present paper addresses issues related to raw data. A generic approach is proposed for obtaining monitoring data that are reliable and exploitable in a PHM application. The proposed approach is based on 2 steps: collecting data and preprocessing data. This approach will be applied to a real world case in broadcast industry to show its feasibility.

This paper addresses the problem of uncertainty quantification in system-level prognostics. To this purpose, a three-step methodology, based on the inoperability input-output model, is presented. The first step concerns the estimation of the system inoperability, using a new adapted particle filtering method, while considering the interactions between its components. The second step focuses on the long-term prediction of the system inoperability in order to determine its evolution. Finally, in the third step, a method for calculating the remaining useful life of the system, based on its configuration, is formulated. The proposed methodology is applied on data obtained from the Tennessee Eastman Process simulations to predict the shutdown due to violation of process constraints.

This paper presents the data acquisition step of a Prognostics and Health Management (PHM) of Micro-Electro-Mechanical Systems (MEMS) application. The targeted MEMS device is an electro-thermally actuated MEMS valve. The data acquisition is performed during the accelerated lifetime tests. To perform tests, an experimental test bed is designed and built. Several test campaigns are performed where MEMS valves operated continuously and data acquired regularly. The obtained experimental results show that MEMS fabricated with the same micro-fabrication process and tested in the same conditions do not have the same behavior and the same evolution of degradation in time. Therefore, this supports the importance of applying PHM of MEMS rather than the predictive reliability.

In condition-based maintenance, real-time observations are crucial for on-line health assessment. When the monitoring system is a wireless sensor network (WSN), data loss becomes highly probable and this affects the quality of the remaining useful life prediction. In this paper, we present a fully distributed algorithm that ensures fault tolerance and recovers data loss in WSNs. We first theoretically analyze the algorithm and give correctness proofs, then provide simulation results and show that the algorithm is (i) able to ensure data recovery with a low failure rate and (ii) preserves the overall energy for dense networks.

Prognostics activity deals with the estimation of the Remaining Useful Life (RUL) of physical systems based on their current health state and their future operating conditions. RUL estimation can be done by using two main approaches, namely model-based and data-driven approaches. The first approach is based on the utilization of physics of failure models of the degradation, while the second approach is based on the transformation of the data provided by the sensors into models that represent the behavior of the degradation. This paper deals with a data-driven prognostics method, where the RUL of the physical system is assessed depending on its critical component. Once the critical component is identified, and the appropriate sensors installed, the data provided by these sensors are exploited to model the degradation's behavior. For this purpose, Mixture of Gaussians Hidden Markov Models (MoG-HMMs), represented by Dynamic Bayesian Networks (DBNs), are used as a modeling tool. MoG-HMMs allow us to represent the evolution of the component's health condition by hidden states by using temporal or frequency features extracted from the raw signals provided by the sensors. The prognostics process is then done in two phases: a learning phase to generate the behavior model, and an exploitation phase to estimate the current health state and calculate the RUL. Furthermore, the performance of the proposed method is verified by implementing prognostics performance metrics, such as accuracy, precision, and prediction horizon. Finally, the proposed method is applied to real data corresponding to the accelerated life of bearings, and experimental results are discussed.

This paper addresses a data-driven prognostics method for the estimation of the Remaining Useful Life (RUL) and the associated confidence value of bearings. The proposed method is based on the utilization of the Wavelet Packet Decomposition (WPD) technique, and the Mixture of Gaussians Hidden Markov Models (MoG-HMM). The method relies on two phases: an off-line phase, and an on-line phase. During the first phase, the raw data provided by the sensors are first processed to extract features in the form of WPD coefficients. The extracted features are then fed to dedicated learning algorithms to estimate the parameters of a corresponding MoG-HMM, which best fits the degradation phenomenon. The generated model is exploited during the second phase to continuously assess the current health state of the physical component, and to estimate its RUL value with the associated confidence. The developed method is tested on benchmark data taken from the “NASA prognostics data repository” related to several experiments of failures on bearings done under different operating conditions. Furthermore, the method is compared to traditional time-feature prognostics and simulation results are given at the end of the paper. The results of the developed prognostics method, particularly the estimation of the RUL, can help improving the availability, reliability, and security while reducing the maintenance costs. Indeed, the RUL and associated confidence value are relevant information which can be used to take appropriate maintenance and exploitation decisions. In practice, this information may help the maintainers to prepare the necessary material and human resources before the occurrence of a failure. Thus, the traditional maintenance policies involving corrective and preventive maintenance can be replaced by condition based maintenance.