Prognostics And Health Management Research Articles

Remaining useful life with self-attention assisted physics-informed neural network

Remaining useful life (RUL) prediction as the key technique of prognostics and health management (PHM) has been extensively investigated. The application of data-driven methods in RUL prediction has advanced greatly in recent years. However, a large number of model parameters, low prediction accuracy, and lack of interpretability of prediction results are common problems of current data-driven methods. In this paper, we propose a Physics-Informed Neural Networks (PINNs) with Self-Attention mechanism-based hybrid framework for aircraft engine RUL prognostics. Specifically, the self-attention mechanism is employed to learn the differences and interactions between features, and reasonably map high-dimensional features to low-dimensional spaces. Subsequently, PINN is utilized to regularize the end-to-end prediction network, which maps features to RUL. The RUL prediction framework termed AttnPINN has verified its superiority on the Commercial Modular AeroPropulsion System Simulation (C-MAPSS) dataset. It achieves state-of-the-art prediction performance with a small number of parameters, resulting in computation-light features. Furthermore, its prediction results are highly interpretable and can accurately predict failure modes, thereby enabling precise predictive maintenance.

Spherical-dynamic time warping – A new method for similarity-based remaining useful life prediction

Machinery prognostics and health management (PHM) plays a key role in the reliable and efficient operation of industrial processes. With the emerging big data era, data-driven prognostic methods which avoid considering complicated system models have attracted growing research interest. Among many data-driven models, similarity-based prediction methods have been popular due to their strong interpretability and relatively simple implementation process. Nevertheless, when quantifying the similarity between two trajectories, most existing similarity measures neglect the nonlinearity of the distance measurement at different degradation stages and degradation alignments with timing difference, which may not be sufficient to retrieve the most suitable trajectories for remaining useful life (RUL) prediction. To overcome these limitations, a spherical-Dynamic Time Warping (spherical-DTW) algorithm is put forward to find an optimal match between the test and training trajectories at the retrieval step. Dynamic Time Warping allows degradation alignments with timing difference through stretching or compressing the trajectories with regard to time, thereby the data in similar degradation levels can be well aligned across different units. Moreover, a newly defined nonlinear spherical distance method is introduced and incorporated into the retrieval process to account for the nonlinearity of the damage propagation process. The significance of this study is that the newly proposed spherical-DTW algorithm goes one step further to consider the nonlinearity of fault evolutions and allow degradation pattern alignments with timing difference when performing similarity-based prognostics. Two run-to-failure cases, involving a real-world industrial compressor failure case and a gas turbine engine failure dataset, are investigated to demonstrate the effectiveness and superiority of the proposed algorithm.

Prognostic and Health Management of Critical Aircraft Systems and Components: An Overview.

Prognostic and health management (PHM) plays a vital role in ensuring the safety and reliability of aircraft systems. The process entails the proactive surveillance and evaluation of the state and functional effectiveness of crucial subsystems. The principal aim of PHM is to predict the remaining useful life (RUL) of subsystems and proactively mitigate future breakdowns in order to minimize consequences. The achievement of this objective is helped by employing predictive modeling techniques and doing real-time data analysis. The incorporation of prognostic methodologies is of utmost importance in the execution of condition-based maintenance (CBM), a strategic approach that emphasizes the prioritization of repairing components that have experienced quantifiable damage. Multiple methodologies are employed to support the advancement of prognostics for aviation systems, encompassing physics-based modeling, data-driven techniques, and hybrid prognosis. These methodologies enable the prediction and mitigation of failures by identifying relevant health indicators. Despite the promising outcomes in the aviation sector pertaining to the implementation of PHM, there exists a deficiency in the research concerning the efficient integration of hybrid PHM applications. The primary aim of this paper is to provide a thorough analysis of the current state of research advancements in prognostics for aircraft systems, with a specific focus on prominent algorithms and their practical applications and challenges. The paper concludes by providing a detailed analysis of prospective directions for future research within the field.

Physics-informed machine learning and its structural integrity applications: state of the art.

The development of machine learning (ML) provides a promising solution to guarantee the structural integrity of critical components during service period. However, considering the lack of respect for the underlying physical laws, the data hungry nature and poor extrapolation performance, the further application of pure data-driven methods in structural integrity is challenged. An emerging ML paradigm, physics-informed machine learning (PIML), attempts to overcome these limitations by embedding physical information into ML models. This paper discusses different ways of embedding physical information into ML and reviews the developments of PIML in structural integrity including failure mechanism modelling and prognostic and health management (PHM). The exploration of the application of PIML to structural integrity demonstrates the potential of PIML for improving consistency with prior knowledge, extrapolation performance, prediction accuracy, interpretability and computational efficiency and reducing dependence on training data. The analysis and findings of this work outline the limitations at this stage and provide some potential research direction of PIML to develop advanced PIML for ensuring structural integrity of engineering systems/facilities. This article is part of the theme issue 'Physics-informed machine learning and its structural integrity applications (Part 1)'.

Multiscale global and local self-attention-based network for remaining useful life prediction

Remaining useful life (RUL) prediction plays an important role in prognostics and health management (PHM) and can significantly enhance equipment reliability and safety in various engineering applications. Accurate RUL prediction enables proactive maintenance planning, helping prevent potential hazards and economic losses caused by equipment failures. Recently, while deep learning-based methods have swept the RUL prediction field, most existing methods still have difficulties in simultaneously extracting multiscale global and local degradation feature information from raw multi-sensor monitoring data. To address these issues, we propose a novel multiscale global and local self-attention-based network (MGLSN) for RUL prediction. MGLSN consists of global and local feature extraction subnetworks based on self-attention, which work in parallel to simultaneously extract the global and local degradation features of equipment and can adaptively focus on more important parts. While the global network captures long-term dependencies between time steps, the local network focuses on modeling local temporal dynamics. The design of parallel feature extraction can avoid the mutual influence of information from global and local aspects. Moreover, MGLSN adopts a multiscale feature extraction design (multiscale self-attention and convolution) to capture the global and local degradation patterns at different scales, which can be combined to better reflect the degradation trend. Experiments on the widely used Commercial Modular Aero-Propulsion System Simulation (CMAPSS), New CMAPSS (N-CMAPSS), and International Conference on Prognostics and Health Management 2008 challenge datasets provided by NASA show that MGLSN significantly outperforms state-of-the-art RUL prediction methods and has great application prospects in the field of PHM.

Review on prognostics and health management in smart factory: From conventional to deep learning perspectives

At present, the fourth industrial revolution is pushing factories toward an intelligent, interconnected grid of machinery, communication systems, and computational resources. Smart factories (SF) and smart manufacturing (SM) incorporate a cyber-physical system that employs advanced technologies such as artificial intelligence (AI) for data analysis, automated process driving, and continuous data handling. Smart factories operate by combining machines, humans, and massive amounts of data into a single, digitally interconnected ecosystem. Prognostics and health management (PHM) has become a critical requirement of smart factories to meet production needs. PHM of components/machines in the smart factory is crucial for securing uninterrupted operation and ensuring safety standards. The growing availability of computational capacity has increased the use of deep learning in PHM strategies. Deep learning supports comprehensive PHM solutions, thus reducing the need for manual feature development. This review presents an extensive study of the PHM strategies employed in the smart factory ranging from the conventional perspective to the deep learning perspective. This includes consideration of the conventional methodologies used for health management along with latest trends in the PHM domain in the smart factory.

Artificial intelligence-based data-driven prognostics in industry: A survey

In the age of Industry 5.0, prognostics and health management (PHM) is very important for proactive and scheduled maintenance in industrial processes. The target of prognosis is the health state prediction of the system or machine under consideration, hence its Remaining Useful Life RUL. The life of a tool, a part, or a component of the system must be tracked to increase its productivity, reduce human effort and save lives. Data driven prognostics is highly relying on statistical or artificial intelligence AI methods including machine learning (ML) and deep learning (DL) models. AI is a massive enlarging field with encouraging outcomes in prognostics for modelling of data with complex representations and temporal dependencies. A sample of latest research in prognostics especially in industry applications has been collected during this research. About 76% of the collected research papers used data-driven prognostics in their model including 48% applied DL different architectures for prognostic purposes in industrial systems in the last few years. Therefore, this survey concentrate on presenting AI-based data-driven prognostics in industrial systems especially DL-based architectures. The study also puts spot on the main challenges with opportunities of future work in the DL-based PHM applications in the age of Industry 5.0.

Survey of Prognostics and Health Management for Transformers: From Statistical Analysis to Condition-Based Diagnostics

Power transformers are one of the key network components for reliable and efficient operation of power grids. Over the past few decades, there have been growing research efforts in improving the prognostics and health management (PHM) for transformers, including failure analysis using time-to-event data and condition-based diagnostics for both single and multiple components. In this paper, we survey recent literature and relevant works, focusing on widely used statistical models and advanced diagnostic techniques that leverage on condition data and maintenance history. Additionally, we examine the role of artificial intelligence (AI) applications in PHM for power transformers. Finally, we summarize the current limitations and future opportunities to support new research efforts for improving the monitoring of power transformers.

Explainable multimodal learning for predictive maintenance of steam generators

Prognostics and Health Management (PHM) is identified as an important lever for enhancing the development of predictive maintenance to ensure the reliability, availability, and safety of industrial systems. However, the efficiency of data- driven PHM approaches is dependent on the quality and quantity of data. Therefore, exploiting multiple data sources can provide additional, useful information than single-modal data. For instance, by incorporating multiple data sources, including condition monitoring data, images from cameras, and texts from maintenance technicians’ reports, multi-modal learning can provide a more comprehensive and accurate understanding of the system’s health. However, multi-modal deep learning is complex to understand. To address this complexity, it is crucial to incorporate explainable artificial intelligent techniques to provide clear and interpretable insights into how the model makes decisions. In this light, this paper proposes the application of the model-agnostic-explanation approach, i.e., SHAP, to explain the working mechanism of multimodal learning for the prediction of industrial steam generator degradation. Particularly, we determine the important features of each data modality and investigate how multimodal learning can overcome the issues of low-quality data from a single modality due to the additional information from other data modalities.

Building The Health Monitoring and Fault Diagnosis Models For Stamping Press

A stamping press is widely used for the metal forming process. To achieve continuous automation and high precision forming, monitoring the press's health and diagnosing faults during stamping is necessary. The three primary types of faults that may occur in the stamping press are lack of lubrication oil, quality variation of lubrication oil, and clearance variation, which can lead to a decline in workpiece quality and reduced lifespan of the dies and presses. This study adopted the Prognostics and Health Management (PHM) technique to implement a predictive maintenance system for the stamping press. To extract relevant data, the National Instrument (NI) DAQ was used to acquire the three-phase currents and X, Y, and Z vibration signals. Six signals provided a total of seventy-two features, and the top three key features were selected for building a health assessment model using the Logistic regression and PCA algorithms. An early warning is triggered when the health indicator drops below the threshold, alerting the operators. Additionally, fault diagnosis was achieved using classification algorithms such as Support Vector Machine (SVM), K Nearest Neighbors (K-NN), and eXtreme Gradient Boosting (XGBoost). The fault diagnosis model achieved high accuracies of up to 99%.

Development of demonstration system for fault diagnosis of rotating equipments using RK4 test rig

Rotating machinery is an essential equipment in the manufacturing industry, of which the failure can lead to the interruption of whole production line. To prevent such failure, there have been a large number of studies for Prognostics and Health Management (PHM) technology. However, most studies have been focused on a specific algorithm or component, making it difficult to apply them in the field. In this study, a demonstration system for integrated fault diagnosis is developed for critical failure modes in the rotating equipment such as the mass imbalance, shaft misalignment, and the bearing fault using the vibration signals. To this end, Bently Nevada's RK4 rotor kit is revised to impose the failure modes easily. The solution is developed to detect anomalies, identify failure modes, and diagnose them in real time. Coupling ISO-based anomaly detection, rotational failure diagnosis and bearing failure diagnosis, the system is configured to enable comprehensive condition monitoring. The results are demonstrated by real-time simulation of each fault on the test-rig.

PHM for Spacecraft Propulsion Systems: Similarity-Based Model and Physics-Inspired Features

This paper presents a methodology designed for the Prognostics and Health Management (PHM) Asia-Pacific 2023 Conference Data Challenge. In particular, this study targets the health assessment of spacecraft propulsion systems. The challenge involved analyzing and categorizing a simulation-generated dataset that included four unique spacecraft and multiple health conditions, such as normal operation, bubble anomalies, and solenoid valve faults in various system locations. The proposed approach uses a two- step process. First, a model based on similarity measures is employed to classify the data into one of four health states. Then, a model incorporating physics-inspired features is utilized in solenoid valve faults to identify the fault location and estimate the valve opening ratio. The validity of the model is confirmed through cross-validation with the training dataset, which achieved a flawless total score across all permutations. Our method effectively categorizes the test data, including cases from a spacecraft not covered in the training, thereby securing a top position in the competition. The findings highlight the strength of our proposed model, which uses physics-inspired features to predict valve opening ratios, proving useful in managing and interpreting complex, unfamiliar spacecraft health data.

Simple Hybrid Model for Estimating Remaining Useful Life of SiC MOSFETs in Power Cycling Experiments

Recording and prediction of the accumulated damage, which will eventually lead to the failure of power electronic modules, is an aspect of high importance for power electronic systems design and, in particular, for development of Prognostic and Health Management (PHM) schemes for in-field applications. To this end, this paper presents a simple and cost-effective prognostic method for predicting the remaining useful life (RUL) of TO-247 packaged silicon carbide (SiC) metal-oxide semiconductor field-effect transistors (MOSFETs) subjected to power cycling experiments. The model assumes that the major failure mode is bond-wire lift-off and uses a damage accumulation scheme based on Paris’ crack law. The only inputs to the model are historical data on the average junction temperature swing and the temperature-compensated drain-source ON-state resistance at the peak temperature of the current cycle. Using only these two input values, the model is shown to predict RUL with surprising accuracy for the range of constant current loads determining cycling conditions under which the test data series have been acquired. This work is a first step in an ongoing project towards building more elaborate prognostic schemes for RUL-determination of SiC power MOSFETs in actual working conditions, using physics-informed neural networks (PINNs).

Expert-Informed Hierarchical Diagnostics of Multiple Fault Modes of a Spacecraft Propulsion System

This paper presents a comprehensive study on diagnosing a spacecraft propulsion system utilizing data provided by the Prognostics and Health Management (PHM) society, specifically obtained as part of the Asia-Pacific PHM conference’s data challenge 2023. The objective of the challenge is to identify and diagnose known faults as well as unknown anomalies in the spacecraft’s propulsion system, which is critical for ensuring the spacecraft’s proper functionality and safety. To address this challenge, the proposed method follows a systematic approach of feature extraction, feature selection, and model development. The models employed in this study are kMeans clustering and decision trees combined to ensembles, enriched with expert knowledge. With the method presented, our team was capable of reaching high accuracy in identifying anomalies as well as diagnosing faults, resulting in attaining the seventh place with a score of 93.08 %.

To Trust or Not: Towards Efficient Uncertainty Quantification for Stochastic Shapley Explanations

Recently, explainable AI (XAI) techniques have gained traction in the field of prognostics and health management (PHM) to enhance the credibility and trustworthiness of data-driven nonlinear models. Post-hoc model explanations have been popularized via algorithms such as SHapley Additive exPlanations (SHAP), but remain impractical for real-time prognostics applications due to the curse of dimensionality. As an alternative to deterministic approaches, stochastically sampled Shapley-based approximations have computational benefits for explaining model predictions. This paper will introduce and examine a new concept of explanation uncertainty through the lens of uncertainty quantification of stochastic Shapley attribution estimates. The proposed algorithm for estimating Shapley explanation uncertainty is efficiently applied for the 2021 PHM Data Challenge problem. The uncertainty in the derived explanation for a single prediction is also illustrated through personalized prediction recipe plots, improving post-hoc model visualization. Finally, important practical considerations for the implementation of Shapley-based XAI for industrial prognostics are provided.

Virtual sample generation for model-based prognostics and health management of on-board high-speed train control system

In view of class imbalance in data-driven modeling for Prognostics and Health Management (PHM), existing classification methods may fail in generating effective fault prediction models for the on-board high-speed train control equipment. A virtual sample generation solution based on Generative Adversarial Network (GAN) is proposed to overcome this shortcoming. Aiming at augmenting the sample classes with the imbalanced data problem, the GAN-based virtual sample generation strategy is embedded into the establishment of fault prediction models. Under the PHM framework of the on-board train control system, the virtual sample generation principle and the detailed procedures are presented. With the enhanced class-balancing mechanism and the designed sample augmentation logic, the PHM scheme of the on-board train control equipment has powerful data condition adaptability and can effectively predict the fault probability and life cycle status. Practical data from a specific type of on-board train control system is employed for the validation of the presented solution. The comparative results indicate that GAN-based sample augmentation is capable of achieving a desirable sample balancing level and enhancing the performance of correspondingly derived fault prediction models for the Condition-based Maintenance (CBM) operations.

Open Heterogeneous Data for Condition Monitoring of Multi Faults in Rotating Machines Used in Different Operating Conditions

Rotating machines are widely used in several fields such as railways, renewable energies, robotics, etc. This diversity of application implies a large variety of faults of critical components susceptible to fail. For this purpose, prognostics and health management (PHM) is deployed to effectively monitor these components through the detection, diagnostics as well as prognostics of faults. In the literature, there exist numerous methods to ensure the above monitoring activities. However, few of them consider different failure types using heterogeneous data and various operating conditions. Also, there are no dominant methods that can be generalized for monitoring. For this reason, the genericity of these methods and their applicability in several systems is a crucial issue. To help researchers to achieve the above challenges, this paper presents a detailed description of data sources from experimental test benches. These data-sets correspond to different case studies that monitor the health states of multiple critical components in various operating conditions using numerous sensors.

A novel IoT-based framework with Prognostics and Health Management and short term fire risk assessment in smart firefighting system

In recent years, “smart firefighting” becomes a popular research topic which aims at improving the efficiency of fire protection and safety. However, in existing smart firefighting systems: (1) state and analog firefighting IoT data is not fully utilized; (2) fire risk assessment requires a long time to update; (3) few existing works can recognize faults and forecast potential risks with both temporal and spatial information from multiple IoT sensors. Therefore, a novel framework is proposed to solve the above problems as follows: (1) Short-term Fire Risk Assessment (SFRA) is proposed by considering the working status, maintenance and rectification performance of firefighting facilities based on IoT state data; (2) Prognostics and Health Management (PHM) algorithms are applied in the fire safety field for the first time to detect faults and forecast potential risks with temporal and spatial information with IoT analog data from single or multiple sensors. A case study is carried out in a building in Shanghai and a smart firefighting system is deployed with the edge computing gateway. The experimental results of the SFRA and PHM are presented and analyzed in details. The performance evaluation and comparison results are presented with multiple metrics. The reliability of the proposed framework is discussed. The case study concluded that the proposed framework extends the utilization of IoT data, improves the efficiency of fire risk assessment and enhances the accuracy of faults detection and potential risk forecast. Moreover, the deployed firefighting system improves the data privacy and computational efficiency.

A Reinforcing-Learning-Driven Artificial Bee Colony Algorithm for Scheduling Jobs and Flexible Maintenance under Learning and Deteriorating Effects

In the last decades, the availability constraint as well as learning and deteriorating effects were introduced into the production scheduling theory to simulate real-world case studies and to overcome the limitation of the classical models. To the best of our knowledge, this paper is the first in the literature to address the permutation flowshop scheduling problem (PFSP) with flexible maintenance under learning and deterioration effects to minimize the makespan. Firstly, we address the PFSP with flexible maintenance and learning effects. Then, the deteriorating effect is also considered. Adaptive artificial bee colony algorithms (ABC) enhanced with Q-learning are proposed, in which the Nawaz–Enscore–Ham (NEH) heuristic and modified NEH heuristics are hybridized with a maintenance insertion heuristic to construct potential integrated initial solutions. Furthermore, a Q-learning (QL)-based neighborhood selection is applied in the employed bees phase to improve the quality of the search space solutions. Computational experiments performed on Taillard’s well-known benchmarks, augmented with both prognostic and health management (PHM) and maintenance data, demonstrate the effectiveness of the proposed QL-driven ABC algorithms.

Remaining useful life prediction of mechanical system based on improved adaptive fractional Lévy stable motion with statistical dependence measurement error

Prediction of remaining useful life (RUL) is a critical component of prognostics and health management (PHM) strategies for mechanical systems. Although some degradation models based on stochastic processes have played a pivotal role in the field of RUL prediction, the diffusion coefficients associated with these models are often fixed values that remain independent of the drift coefficient. Additionally, the measurement error is typically assumed to follow a Gaussian distribution that is independently and identically distributed with respect to the level of degradation. To address these constraints, an RUL prediction framework is proposed based on improved adaptive fractional Lévy stable motion (IAFLSM) with statistical dependence measurement error. The established IAFLSM model is capable of effectively characterizing the individual differences and time-varying uncertainties inherent in equipment degradation processes, with the nonlinear drift and diffusion coefficients exhibiting positive correlation in their variability. Moreover, the state space model is employed to realize the synchronous adaptive dynamic update of the drift coefficient and diffusion coefficient, thus accommodating the mechanical equipment degradation trajectory. Furthermore, a statistical dependence measurement error based on fractional Lévy stable motion is constructed, and the corresponding scale parameter incorporates a dynamic update mechanism with statistical correlation of degradation incremental behavior. The hidden variables related to performance degradation model are estimated through parameter estimation method and characteristic function. Expanding upon the proposed framework for RUL prediction, the quantification of the uncertainty intrinsic in the forecasting results is accomplished by employing the stability theorem of stable distribution and Monte Carlo technique. The RUL prediction framework is validated through the use of authentic truck rear axle full-life data and benchmark rolling bearing data. The comparative analysis results demonstrate the effectiveness and superiority of the proposed methodology.