Data-driven Remaining Useful Life Prediction Research Articles

New Health Indicator Construction and Fault Detection Network for Rolling Bearings via Convolutional Auto-Encoder and Contrast Learning

As one of the most important components in rotating machinery, if bearings fail, serious disasters may occur. Therefore, the remaining useful life (RUL) prediction of bearings is of great significance. Health indicator (HI) construction and early fault detection play a crucial role in data-driven RUL prediction. Unfortunately, most existing HI construction methods require prior knowledge and preset trends, making it difficult to reflect the actual degradation trend of bearings. And the existing early fault detection methods rely on massive historical data, yet manual annotation is time-consuming and laborious. To address the above issues, a novel deep convolutional auto-encoder (CAE) based on envelope spectral feature extraction is developed in this work. A sliding value window is defined in the envelope spectrum to obtain initial health indicators, which are used as preliminary labels for model training. Subsequently, CAE is trained by minimizing the composite loss function. The proposed construction method can reflect the actual degradation trend of bearings. Afterwards, the autoencoder is pre-trained through contrast learning (CL) to improve its discriminative ability. The model that has undergone offline pre-training is more sensitive to early faults. Finally, the HI construction method is combined with the early fault detection method to obtain a comprehensive network for online health assessment and fault detection, thus laying a solid foundation for subsequent RUL prediction. The superiority of the proposed method has been verified through experiments.

A Data-Model Interactive Remaining Useful Life Prediction Approach of Lithium-Ion Batteries Based on PF-BiGRU-TSAM

Accurate remaining useful life (RUL) prediction of lithium-ion batteries is critical for energy supply systems. In conventional data-driven RUL prediction approaches, the battery's degradation mechanism is difficult into incorporate in the RUL prediction. Furthermore, there are notable limitations in reflecting the significance of different time instances, and the uncertainty in the degradation process. Consequently, a novel data-model interactive remaining useful life prediction approach based on particle filter-temporal attention mechanism-bidirectional gated recurrent unit (PF-BiGRU-TSAM) is proposed. Specifically, BiGRU-TSAM is trained offline through historical data, which assigns corresponding significance to battery capacities at different time instances. Moreover, regarding the interactive data-model for the online prediction phase based on PF-BiGRU-TSAM, the advantages of data-driven and model-based approaches are integrated, which accomplishes the purpose of modifying each other. The proposed PF-BiGRU-TSAM approach is validated with a real-world battery dataset. Experimental results demonstrate the proposed approach is better than some published approaches. Taking the 50th operational cycle of the four batteries B0005, B0006, B0007, and B0018 in the dataset as an instance, the absolute errors of the proposed PF-BiGRU-TSAM are 0, 1, 3, 3, respectively, which represents the proposed approach has an excellent performance.

A review on physics-informed data-driven remaining useful life prediction: Challenges and opportunities

Remaining useful life (RUL) prediction, known as ‘prognostics’, has long been recognized as one of the key technologies in prognostics and health management (PHM) to maintain the safety and reliability of the system, and reduce the operating and management costs. Particularly, thanks to great advances in sensing and condition monitoring techniques, data-driven RUL prediction has attracted much attention and various data-driven RUL prediction methods have been reported. Despite the extensive studies on data-driven RUL prediction methods, the successful applications of such methods depend heavily on the volume and quality of the data, and purely data-driven methods possibly generate physically infeasible/inconsistent RUL prediction results and have the limited generalizability and interpretability. It is noted that there is an increasing consensus that embedding the physics or the domain knowledge into the data-driven methods and developing physics-informed data-driven methods will hold promise to improve the interpretability and efficiency of the RUL prediction results and lower the requirement of the volume and quality of the data. In this context, physics-informed data-driven RUL prediction has become an emerging topic in the prognostics field. However, there has not been a systematic review particularly focused on this emerging topic. To fill this gap, this paper reviews recent developments of physics-informed data-driven RUL prediction methods. In this review, current methods fallen into this type are broadly divided into three categories, i.e. physical model and data fusion methods, stochastic degradation model based methods, and physics-informed machine learning (PIML) based methods. Particularly, this review is centered on the PIML based methods since the fast development of such methods have been witnessed in the past five years. Through discussing the pros and cons of existing methods, we provide discussions on challenges and possible opportunities to steer the future development of physics-informed data-driven RUL prediction methods.

Domain adaptation via alignment of operation profile for Remaining Useful Lifetime prediction

Effective Prognostics and Health Management (PHM) relies on accurate prediction of the Remaining Useful Life (RUL). Data-driven RUL prediction techniques rely heavily on the representativeness of the available time-to-failure trajectories. Therefore, these methods may not perform well when applied to data from new units of a fleet that follow different operating conditions than those they were trained on. This is also known as domain shifts. Domain adaptation (DA) methods aim to address the domain shift problem by extracting domain invariant features. However, DA methods do not distinguish between the different phases of operation, such as steady states or transient phases. This can result in misalignment due to under- or over-representation of different operation phases. This paper proposes two novel DA approaches for RUL prediction based on an adversarial domain adaptation framework that considers the different phases of the operation profiles separately. The proposed methodologies align the marginal distributions of each phase of the operation profile in the source domain with its counterpart in the target domain. The effectiveness of the proposed methods is evaluated using the New Commercial Modular Aero-Propulsion System (N-CMAPSS) dataset, where sub-fleets of turbofan engines operating in one of the three different flight classes (short, medium, and long) are treated as separate domains. The experimental results show that the proposed methods improve the accuracy of RUL predictions compared to current state-of-the-art DA methods.

Dynamic Model-Assisted Bearing Remaining Useful Life Prediction Using the Cross-Domain Transformer Network

Remaining useful life (RUL) prediction of rolling bearings is of paramount importance to various industrial applications. Recently, intelligent data-driven RUL prediction methods have achieved fruitful results. However, the existing methods heavily rely on the quality and quantity of the available data. For some critical bearings in industrial scenarios, the real run-to-failure data are insufficient, which impair the applicability of data-based methods for industrial practices. To address these issues, this article proposes a novel dynamic model-assisted RUL prediction approach for rolling bearing, in which sufficient simulation data are applied as the training data to solve the problem caused by insufficient real data. More specifically, a dynamic rolling bearing model is introduced for simulating the degradation process of physical structures. Then, a multilayer cross-domain transformer network is developed to implement RUL prediction and adapt the learned prediction knowledge from simulation to the actual measurements. Furthermore, a mutual information loss is utilized to preserve the generalized prediction knowledge of the measured data. The proposed approach can achieve a high RUL prediction accuracy with only limited measured data, which tackles the drawbacks of the existing data-driven methods. The experimental results of the rolling bearing degradation datasets demonstrate the effectiveness and superiority of the proposed RUL prediction approach.

Data-Driven Approach for Lithium-Ion Battery Remaining Useful Life Prediction: A Literature Review

Nowadays, lithium-ion battery has become more popular around the world. Knowing when batteries reach their end of life (EOL) is crucial. Accurately predicting the remaining useful life (RUL) of lithium-ion batteries is needed for battery health management systems and to avoid unexpected accidents. It gives information about the battery status and when we should replace the battery. With the rapid growth of machine learning and deep learning, data-driven approaches are proposed to address this problem. Extracting aging information from battery charge/discharge records, including voltage, current, and temperature, can determine the battery state and predict battery RUL. In this work, we first outlined the charging and discharging processes of lithium-ion batteries. We then summarize the proposed techniques and achievements in all published data-driven RUL prediction studies. From that, we give a discussion about the accomplishments and remaining works with the corresponding challenges in order to provide a direction for further research in this area.

Augmented model-based framework for battery remaining useful life prediction

Traditional, model-based approaches for predicting the remaining useful life (RUL) of a rechargeable battery cell simply update and extrapolate a mathematical model which describes the evolution of the cell’s capacity fade trend. These approaches are straightforward but tend to break down when the capacity fade trend changes over the cell’s lifetime. To retain the desirable properties of model-based prediction approaches (uncertainty quantification, long-term accuracy, limited physical meaning) and improve their overall accuracy in RUL prediction, we augment empirical model-based prediction with data-driven error correction. Our approach decomposes the task of RUL prediction into two steps: 1) Offline training of data-driven models for RUL error correction and 2) Online data-driven correction of model-based RUL prediction. The approach is evaluated on five datasets consisting of 237 cells: 1) three open-source datasets, 2) one proprietary dataset, and 3) a simulated out-of-distribution dataset. Results show that data-driven error correction effectively reduces root-mean-square-error by 40% and mean uncertainty calibration error by 34% compared to a model-based approach alone. The proposed approach is also shown to be more conservative in its uncertainty estimates than a purely data-driven RUL prediction approach. Special attention is given to ensure the initial model-based uncertainty estimates are propagated through the data-driven error correction model and considered in the final RUL prediction. The enhanced uncertainty quantification of our approach makes it suitable for deployment in an online predictive maintenance scheduling framework.

Rolling Bearing Health Indicator Extraction and RUL Prediction Based on Multi-Scale Convolutional Autoencoder

Rolling bearings are some of the most crucial components in rotating machinery systems. Rolling bearing failure may cause substantial economic losses and even endanger operator lives. Therefore, the accurate remaining useful life (RUL) prediction of rolling bearings is of tremendous research importance. Health indicator (HI) construction is the critical step in the data-driven RUL prediction approach. However, existing HI construction methods often require extraction of time-frequency domain features using prior knowledge while artificially determining the failure threshold and do not make full use of sensor information. To address the above issues, this paper proposes an end-to-end HI construction method called a multi-scale convolutional autoencoder (MSCAE) and uses LSTM neural networks for RUL prediction. MSCAE consists of three convolutional autoencoders with different convolutional kernel sizes in parallel, which can fully exploit the global and local information of the vibration signals. First, the raw vibration data and labels are input into MSCAE, and then, MSCAE is trained by minimizing the composite loss function. After that, the vibration data of the test bearings are fed into the trained MSCAE to extract HI. Finally, RUL prediction is performed using the LSTM neural network. The superiority of the HI extracted by MSCAE was verified using the PHM2012 challenge dataset. Compared to state-of-the-art HI construction methods, RUL prediction using MSCAE-extracted HI has the highest prediction accuracy.

Accurate Remaining Useful Life Prediction With Uncertainty Quantification: A Deep Learning and Nonstationary Gaussian Process Approach

Remaining useful life (RUL) refers to the expected remaining lifespan of a component or system. Accurate RUL prediction is critical for prognostic and health management and for maintenance planning. In this article, we address three prevalent challenges in data-driven RUL prediction, namely the handling of high-dimensional input features, the robustness to noise in sensor data and prognostic datasets, and the capturing of the time-dependency between system degradation and RUL prediction. We devise a highly accurate RUL prediction model with uncertainty quantification, which integrates and leverages the advantages of deep learning and nonstationary Gaussian process regression (DL–NSGPR). We examine and benchmark our model against other advanced data-driven RUL prediction models using the turbofan engine dataset from the NASA prognostic repository. Our computational experiments show that the DL-NSGPR predictions are highly accurate with root mean square error 1.7 to 6.2 times smaller than those of competing RUL models. Furthermore, the results demonstrate that RUL uncertainty bounds with the proposed DL-NSGPR are both valid and significantly tighter than other stochastic RUL prediction models. We unpack and discuss the reasons for this excellent performance of the DL-NSGPR.

A novel remaining useful life prediction method based on gated attention mechanism capsule neural network

High-accuracy remaining useful life (RUL) prediction is helpful to make in-time maintenance scheduling, reduce the waste of resources, and prevent the occurrence of serious accidents. Currently, data-driven RUL prediction methods are widely used in engineering fields due to their simplicity, efficiency, and robustness. In data-driven methods, the RUL is predicted by learning the mapping from the sensor data to the RUL of machinery. However, the sensor data are often disturbed by noises, and the existence of noise can negatively affect the follow-up RUL prediction. Moreover, the uncertainty of the predicted RUL is often ignored. To address the issues, this paper proposes a novel gated attention mechanism capsule neural network (GAM-CapsNet). A gated attention mechanism (GAM) is developed to increase the anti-interference ability of the model against noises and assign large weights to the most important features. In order to improve the feature extraction ability of the model and quantify the uncertainty of the RUL prediction, the primary capsule, digital capsule, and Bayesian layer are implemented in the proposed GAM-CapsNet. The effectiveness and superiority of the GAM-CapsNet model are verified on turbine engine and cutter wear datasets. Compared to state-of-the-art methods, the prediction ability of the GAM-CapsNet on four engine datasets is improved by 1.11%, 13.44%, 3.06%, and 12.49%, respectively. In addition, the prediction ability of the GAM-CapsNet on three cutter wear datasets is improved by 33.82%, 58.95%, and 36.33%, respectively. The experimental results indicate that the GAM-CapsNet model has better prediction performance.

Remaining useful life estimation of HMPE rope during CBOS testing through machine learning

Fibre rope used in cranes for offshore deployment and recovery has significant potential to perform lifts with smaller cranes and vessels to reach depths limited by weight of steel wire rope. Current condition monitoring methods based on manual inspection and time-based and reactive maintenance have significant potential for improvement coupled with more accurate remaining useful life (RUL) prediction. Machine learning has found use as a condition monitoring approach, coupled with vast improvements in data acquisition methods.This paper details data-driven RUL prediction methods based on machine learning algorithms applied on cyclic-bend-over-sheave (CBOS) tests performed on two fibre rope types until failure. Data extracted through computer vision and thermal monitoring is used to predict RUL through neural networks, support vector machines and random forest. Random forest and neural networks methods are shown to be particularly adept at predicting RUL compared to support vector machines . Additionally, improved RUL predictions can be achieved by combining data from distinct rope types subject to different test conditions.

Early and robust remaining useful life prediction of supercapacitors using BOHB optimized Deep Belief Network

The lifespan, power density, and transient response make supercapacitors a component of choice for the electric vehicle and renewable energy industry. Supercapacitors’ long lifecycle often makes it difficult for designers to assess the system’s reliability over the complete product cycle. In the existing literature, the remaining useful life (RUL) estimations utilize up to 50% state of health (SOH) degradation data to successfully predict the RUL of the supercapacitors with reasonable accuracy, making them impractical in terms of time and resources required to collect the data. The time to acquire data imposes restrictions on developing a data-driven RUL prediction model for the supercapacitors. The objective of this study is to reliably predict the SOH degradation curve of the supercapacitors with the availability of less than 10% degradation data to avoid time and cost-consuming lifecycle testing. This study presents a novel combination of deep learning algorithm-Deep Belief Network (DBN) with Bayesian Optimization and HyperBand (BOHB) to predict the RUL of the supercapacitors in the early phases of degradation. The proposed method successfully predicts the degradation curve using the data of the initial 15 thousand cycles (less than 6% data for training in most of the cases), which is very promising since the supercapacitor has yet to show much degradation at this stage, thus reducing up to 54% time for the development of the RUL prediction model. The proposed model shows good accuracy with percent error and root mean squared error (RMSE) ranging from 0.05% to 2.2% and 0.8851 to 1.6326, respectively. The robustness of the model is also tested by injecting noise in the training data during training.

Feature Extraction for Data-Driven Remaining Useful Life Prediction of Rolling Bearings

A variety of data-driven methods have been proposed to predict remaining useful life (RUL) of key component for rolling bearings. The accuracy of data-driven RUL prediction model largely depends on the extraction method of performance degradation features. The individual heterogeneity and working condition difference of rolling bearings lead to the different performance degradation curves of rolling bearings, which result in the mismatch between the established RUL prediction model by the training rolling bearings and the test rolling bearings. If a feature is found, which can reflect the consistency of the performance degradation curve of each rolling bearings, and give the indicator to determine the node and predictable interval of the declining period, the accuracy of the RUL prediction model will be improved. To solve this problem, a new feature extraction method based on the data-driven method, namely, Fitting Curve Derivative Method of Maximum Power Spectrum Density (FDMPD), is proposed to extract the performance degradation features of the same or similar rolling bearings from the historical state monitoring data in this article. The FDMPD can make the performance degradation feature curves of life cycle, which takes on consistency trend for different rolling bearings, and the starting point of the rolling bearings to enter the degenerating period is defined and the working stage of rolling bearings is divided. Based on this, the kernel extreme learning machine (KELM) and weight application to failure times (WAFT) are combined with FDMPD to establish a new RUL prediction model of rolling bearings, which can effectively realize the RUL prediction of rolling bearings. The whole life cycle data of rolling bearings are used to verify the validity of the RUL prediction model. The experimental results show that the established RUL prediction model can accurately predict the RUL of rolling bearings. It has the advantages of rapidity, stability, and applicability.

Health indicator construction by quadratic function-based deep convolutional auto-encoder and its application into bearing RUL prediction

As one of the most important components of machinery, once the bearing has a failure, serious catastrophe may happen. Hence, for avoiding the catastrophe, it is valuable to predict the remaining useful life (RUL) of bearing. Health indicators (HIs) construction plays a greatly important role in the data-driven RUL prediction. Unfortunately, most of the existing HIs construction methods need prior knowledge and few of them construct HIs from raw vibration signals. For dealing with the above issues, a novel quadratic function-based deep convolutional auto-encoder is developed in this work. The raw bearing vibration signals are first preprocessed by low-pass filtering. Then the cleaned vibration signals are input into the quadratic function-based DCAE neural networks for constructing HIs of bearings. Compared with AE, DNN, KPCA, ISOMAP, PCA and VAE, it is revealed that the proposed methodology can construct a better HI from the raw bearing vibration signal in terms of comprehensive performance. Several comparative experiments have been implemented, and the results indicate that the HI constructed by quadratic function-based DCAE neural network has stronger predictive power than the traditional data-driven HIs.

Data-Driven Method for Predicting Remaining Useful Life of Bearing Based on Bayesian Theory.

Bearings are some of the most critical industrial parts and are widely used in various types of mechanical equipment. Bearing health status can have a significant impact on the overall equipment performance, and bearing failures often cause serious economic losses and even casualties. Thus, estimating the remaining useful life (RUL) of bearings in real time is of utmost importance. This paper proposes a data-driven RUL prediction method for bearings based on Bayesian theory. First, time-domain features are extracted from the bearing vibration signal and data are fused to build a health indicator (HI) and a state model of bearing degradation. Then, according to Bayesian theory, a Bayesian model of state parameters and bearing life is established. The parameters of the Bayesian model are updated and bearing RUL is predicted by the Metropolis–Hastings algorithm. The method was validated by the XJTU-SY bearing open datasets and the prediction results are compared with the existing methods. Accuracy of the proposed method was demonstrated.

Multi-Sensor Data-Driven Remaining Useful Life Prediction of Semi-Observable Systems

Semi-observable systems are referred to as a kind of widely used industrial equipment whose physical degradation state is only observable via shutdown inspection. To monitor the degradation process of semi-observable systems online, different types of sensors are generally employed to collect monitoring signals. Lots of studies have been conducted to fuse multi-sensor signals to predict remaining useful life (RUL). Majority of them, however, ignored the partially available state observations which can be viewed as ground truth measurements of physical degradation. To deal with this problem, this article proposes a multi-sensor data-driven RUL prediction method for semi-observable systems, which leverages degradation information from online multi-sensor signals as well as offline state observations. This method is developed based on a generalizable state-space model combined with particle filtering framework. In the framework, a state transition function is used to describe the degradation process of system states. A multidimensional measurement function is constructed to describe the mapping between states and multi-sensor signals. To enhance the performance of prediction, an algorithm named prioritized sensor group selection is also proposed to select the optimal sensor group for RUL prediction. The effectiveness of the proposed method is demonstrated using an experiment of cutting tool wear.

Contrastive Adversarial Domain Adaptation for Machine Remaining Useful Life Prediction

Enabling precise forecasting of the remaining useful life (RUL) for machines can reduce maintenance cost, increase availability, and prevent catastrophic consequences. Data-driven RUL prediction methods have already achieved acclaimed performance. However, they usually assume that the training and testing data are collected from the same condition (same distribution or domain), which is generally not valid in real industry. Conventional approaches to address domain shift problems attempt to derive domain-invariant features, but fail to consider target-specific information, leading to limited performance. To tackle this issue, in this article, we propose a contrastive adversarial domain adaptation (CADA) method for cross-domain RUL prediction. The proposed CADA approach is built upon an adversarial domain adaptation architecture with a contrastive loss, such that it is able to take target-specific information into consideration when learning domain-invariant features. To validate the superiority of the proposed approach, comprehensive experiments have been conducted to predict the RULs of aeroengines across 12 cross-domain scenarios. The experimental results show that the proposed method significantly outperforms state-of-the-arts with over 21% and 38% improvements in terms of two different evaluation metrics.

Sparse auto-encoder with regularization method for health indicator construction and remaining useful life prediction of rolling bearing

Remaining useful life (RUL) prediction, allowing for mechanical predictive maintenance, reduces unplanned and expensive maintenance greatly. One of the great challenges of data-driven RUL prediction is to extract the features that describe the actual degradation process. This paper presents a health indicator (HI) construction method based on a sparse auto-encoder with regularization (SAEwR) model for rolling bearings. This paper includes two modules, HI construction and RUL prediction. In the stage of the HI construction, the original features are compressed and extracted by the SAEwR model. The extracted features are sorted according to the trendability, and the features with large trendability are selected to construct the HI by using minimum quantization error. In the module of RUL prediction, the maximum likelihood estimation method is used to estimate the parameters of the prediction model, and a particle filter-based RUL prediction with degradation model is proposed. The proposed method is benchmarked with variational auto-encoder, auto-encoder methods and principal component analysis. The data from PRONOSTIA and ABLT-1A platform support the value of our approach.

Remaining useful lifetime prediction for equipment based on nonlinear implicit degradation modeling

Nonlinearity and implicitness are common degradation features of the stochastic degradation equipment for prognostics. These features have an uncertain effect on the remaining useful life (RUL) prediction of the equipment. The current data-driven RUL prediction method has not systematically studied the nonlinear hidden degradation modeling and the RUL distribution function. This paper uses the nonlinear Wiener process to build a dual nonlinear implicit degradation model. Based on the historical measured data of similar equipment, the maximum likelihood estimation algorithm is used to estimate the fixed coefficients and the prior distribution of a random coefficient. Using the on-site measured data of the target equipment, the posterior distribution of a random coefficient and actual degradation state are step-by-step updated based on Bayesian inference and the extended Kalman filtering algorithm. The analytical form of the RUL distribution function is derived based on the first hitting time distribution. Combined with the two case studies, the proposed method is verified to have certain advantages over the existing methods in the accuracy of prediction.

Data-Driven Remaining Useful Life Prediction Considering Sensor Anomaly Detection and Data Recovery

Prognostics and health management (PHM) is being adopted more and more in the modern engineering systems. As one of the most important technologies in the PHM domain, remaining useful life (RUL) prediction has attracted much attention from the researchers in the scholar and industrial field. Although many methods have been proposed to improve the prediction result, the problem of sensor anomaly detection and data recovery has not been considered together. To achieve this object, the data-driven RUL prediction framework considering sensor anomaly detection and data recovery is proposed, which is expected to improve the performance of RUL prediction caused by sensor anomaly. The selected sensor data are first detected to decide whether they are anomalous. If the data of this selected sensor are normal, they are continuously adopted as the input of the RUL prediction algorithm. But, if the data are anomalous, they will be recovered by the related algorithm. The recovered data will be utilized as the input of the RUL prediction algorithm. In the proposed framework, mutual information, Kernel principal component analysis (KPCA), least square-support vector machine (LS-SVM), and Gaussian process regression (GPR) are utilized. Both simulation data and practical data are used to evaluate the performance of the proposed method. Compared with abandoning the anomalous sensor data, the recovered data can indeed help to enhance the RUL prediction result.