Prognostics And Health Management Research Articles

Machine learning predictions of lithium-ion battery state-of-health for eVTOL applications

Electric vertical take-off and landing (eVTOL) aircrafts is one solution for the urban and peri-urban mobility. To monitor the aircraft systems and functions and to mitigate safety concerns, prognostic and health management (PHM) uses state-of-the-art machine learning approaches. Lithium-ion battery, as the main power source of eVTOL, needs to be monitored. To anticipate the battery ageing, the most important parameter to predict is its state-of-health (SoH). In this article, we present a machine learning methodology to predict and forecast the batteries SoH (regression). This study was based on a public dataset consisting of 22 cells that were tested over several hundreds of charge/discharge cycles, with power demands simulating typical eVTOL missions. After feature engineering and preprocessing using a rolling window, 5 machine learning models were adjusted to the train dataset using cross-validation and grid search (linear regression, support vector machines, k-nearest neighbors (kNN), random forest and gradient boosted trees). kNN was found to give the best validation and test scores for the lowest training time (1 μs/point). Finally, after finer hyperparameters tuning (number of observations in the rolling window and k neighbors), kNN was found accurate to forecast SoH up to 200 cycles (≈500 h) ahead (test-R2 ≈ 0.98).

Development of a time series imaging approach for fault classification of marine systems

As part of any Prognostics and Health Management (PHM) system for the shipping industry, the determination of the current health of marine systems is fundamental. As such, diagnostic analytics is performed; a process that is typically constituted by fault detection, fault isolation, and fault identification. Although some efforts have been made to distinguish the faults and malfunctions (fault detection) that can occur in marine systems, the implementation of fault identification to provide a description of any considered fault type and its nature is still an unexplored area due to the lack of fault data. To overcome this, a methodology for the identification of anomalies in marine systems is presented in this paper. The proposed approach aims to analyse the implementation of time series imaging through the application of the first-order Markov chain in tandem with an analysis of both ResNet50V2 and Convolutional Neural Networks (CNNs) as part of the image classification task. To highlight the performance of this methodology, anomalies have been simulated considering the power parameter of a diesel generator. Results demonstrated the potential of time series imaging and image classification approaches, as the Markov-CNN achieved an accuracy of 95% when performing the fault classification task.

Data-Driven Voltage Prognostic for Solid Oxide Fuel Cell System Based on Deep Learning

A solid oxide fuel cell (SOFC) is an innovative power generation system that is green, efficient, and promising for a wide range of applications. The prediction and evaluation of the operation state of a solid oxide fuel cell system is of great significance for the stable and long-term operation of the power generation system. Prognostics and Health Management (PHM) technology is widely used to perform preventive and predictive maintenance on equipment. Unlike prediction based on the SOFC mechanistic model, the combination of PHM and deep learning has shown wide application prospects. Therefore, this study first obtains an experimental dataset through short-term degradation experiments of a 1 kW SOFC system, and then proposes an encoder-decoder RNN-based SOFC state prediction model. Based on the experimental dataset, the model can accurately predict the voltage variation of the SOFC system. The prediction results of the four different prediction models developed are compared and analyzed, namely, long short-term memory (LSTM), gated recurrent unit (GRU), encoder–decoder LSTM, and encoder–decoder GRU. The results show that for the SOFC test set, the mean square error of encoder–decoder LSTM and encoder–decoder GRU are 0.015121 and 0.014966, respectively, whereas the corresponding error results of LSTM and GRU are 0.017050 and 0.017456, respectively. The encoder–decoder RNN model displays high prediction precision, which proves that it can improve the accuracy of prediction, which is expected to be combined with control strategies and further help the implementation of PHM in fuel cells.

A hybrid method for prognostics of lithium-ion batteries capacity considering regeneration phenomena

Prognostics and health management (PHM) is crucial to the reliability and safety of lithium-ion batteries. In this respect, the capacity regeneration phenomenon that occurs during the process of battery degradation brings a challenge to the accuracy of capacity prediction. In this paper, a hybrid method is proposed for the accurate prediction of lithium-ion batteries capacity considering regeneration. Firstly, the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is applied to decompose the raw capacity signal into the global degradation trend components and the local fluctuation components. Then, each component is separately fed to the adaptive neuro-fuzzy inference system (ANFIS) for prediction. Finally, the individual outputs of the ANFIS models are recomposed to obtain the ultimate prediction results. The proposed method is validated by application to NASA lithium-ion battery experimental data. The results obtained show that the proposed method can obtain satisfactory prediction accuracy, wherein the negative impact of capacity regeneration on the prediction accuracy is reduced.

Remaining useful life prediction of mechanical system based on performance evaluation and geometric fractional Lévy stable motion with adaptive nonlinear drift

Remaining useful life (RUL) prediction is of great significance for prognostic and health management (PHM). To accurately predict the RUL of mechanical system under complex conditions, an RUL prediction framework is proposed based on performance evaluation and geometric fractional Lévy stable motion (GFLSM) with adaptive nonlinear drift. The early fault identification of degradation process is realized by setting a threshold for the constructed monotonic health indicator (HI). The dynamic updating method of failure threshold depending on confidence interval is proposed to determine the time of zero RUL. The heavy-tailed distribution degradation model based on GFLSM is constructed to overcome the limitation of Gaussian distribution. The multiple degradation stages are mapped to a relatively unified mode through GFLSM. The long-range dependence and self-similarity of degradation process are described through the relationship between Hurst exponent and stability exponent. The adaptive updating method of nonlinear drift coefficient is put forward to satisfy different degradation trajectories, and other parameters of GFLSM are estimated by the characteristic function method. The predicted RUL and corresponding probability density function (PDF) are obtained by Monte Carlo. The proposed RUL prediction framework is verified by the degradation simulation signal and two different practical industrial experiments. The experimental results demonstrate that the proposed framework is more effective and superior to other state-of-the-art techniques in RUL prediction of mechanical system.

Prognostics and health management for piezoresistive pressure sensor based on improved gated recurrent unit networks

Application fields for piezoresistive pressure sensors have become increasingly extensive in recent years. So a high reliability of the sensor is also required. However, considering that some sensors operate in hostile environments,the need to ensure continuous operation accuracy, prognostics, and health management (PHM) for piezoresistive pressure sensors should not be ignored. To solve this problem, a fault diagnosis and prognostic method that combines a support vector machine (SVM) and deep gated recurrent unit (DGRU) network optimized by hunter–prey optimization (HPO) is proposed in this paper. First, three sensor fault types are defined. Second, SVM is adopted to realize the fault diagnosis. Third, two layers of DGRU are employed to predict the health index, which is defined to represent the health state of the sensor. Meanwhile, the optimal parameters of the DGRU are optimized by HPO algorithms. Finally, the remaining useful life can be estimated by the predicted health index and failure threshold. The method proposed in this paper is proved to be effective and accurate. The fault diagnosis accuracy is 100% in the three fault types defined by this paper. The minimum mean absolute error is 6 × 10−5. It proves the proposed method of PHM in this paper is viable in a real application.

Privacy-preserving and sensor-fused framework for prognostic & health management in leased manufacturing system

Nowadays, increased investments and fierce competition have made the leased manufacturing paradigm being widely adopted. Advances in multi-sensor technologies and data-driven methodologies have also supported the original equipment manufacturer (OEM) to focus on outsourcing prognostic and health management (PHM) services for the leased manufacturing system. However, under this product-service paradigm, the degraded-related signal analysis for massive sensors of leased machines is always limited by privacy concerns, fusion lack, and varying correlation. To promote the security performance and analysis capability of PHM from the OEM to the lessee, this paper proposes a privacy-preserving and sensor-fused framework to achieve time-to-failure (TTF) updating and predictive maintenance (PdM) arrangement for the leased manufacturing system. We first aggregate degraded-related historical datasets through encrypted functional features to preserve the privacy signals of two parties. Next, informative sensors of each leased machine are identified by the penalized functional (log)-location-scale (LLS) regression, while keeping the signals private. Then, the real-time observed signals from the identified sensors on the lessee side are locally fused and cryptographically uploaded to the OEM for the adaptive functional LLS regression. The TTF distributions can thus be securely updated in real time and integrated to output PdM intervals for each leased machine. We further validate our privacy-preserving and sensor-fused framework on a real leased crankshaft manufacturing system, which shows our secure mechanism and fused prognostic can significantly ensure privacy protection, computational performance, fusion effectiveness, and PdM cost reduction.

Predicting the Remaining Useful Life of Landing Gear with Prognostics and Health Management (PHM)

Landing gear is an essential part of an aircraft. However, the components of landing gear are susceptible to degradation over the life of their operation, which can result in the shimmy effect occurring during take-off and landing. In order to reduce unplanned flight disruptions and increase the availability of aircraft, the predictive maintenance (PdM) technique is investigated in this study. This paper presents a case study on the implementation of a health assessment and prediction workflow for remaining useful life (RUL) based on the prognostics and health management (PHM) framework of currently in-service aircraft, which could significantly benefit fleet operators and aircraft maintenance. Machine learning is utilized to develop a health indicator (HI) for landing gear using a data-driven approach, whereas a time-series analysis (TSA) is used to predict its degradation. The degradation models are evaluated using large volumes of real sensor data from in-service aircraft. Finally, the challenges of implementing a built-in PHM system for next-generation aircraft are outlined.

Failure prognosis of the components with unlike degradation trends: A data-driven approach

Precise remaining useful life (RUL) estimation of components is critical for the prognostic and health management (PHM) of the systems to improve reliability and reduce downtime and maintenance costs. One component may show multiple degradation patterns throughout its life cycle. The degradation trends’ occurrence and recurrence are highly unpredictable. This article suggests an RUL prediction model based on artificial neural network (ANN), for components that show different patterns of degradation while operating under similar working conditions. For the ANN learning, some key time-domain features based on the high correlation of the features with the target output, that is, Life ratio (LR) of the components, are extracted from the history of degradation profiles. Prediction intervals are also estimated to account for the various uncertainties in the degradation profile data. In an application involving accelerated aging of capacitors, when the results of the ANN model are compared to the results of conventional machine learning models for example, Linear Regression, Decision Tree, Support Vector Regression, and Bayesian Neural Network (BNN), it is found that the ANN model gives lowest Mean Square Error (MSE) with limited data, thereby demonstrating the effectiveness of the proposed methodology.

On Predictive Maintenance in Industry 4.0: Overview, Models, and Challenges

In the era of the fourth industrial revolution, several concepts have arisen in parallel with this new revolution, such as predictive maintenance, which today plays a key role in sustainable manufacturing and production systems by introducing a digital version of machine maintenance. The data extracted from production processes have increased exponentially due to the proliferation of sensing technologies. Even if Maintenance 4.0 faces organizational, financial, or even data source and machine repair challenges, it remains a strong point for the companies that use it. Indeed, it allows for minimizing machine downtime and associated costs, maximizing the life cycle of the machine, and improving the quality and cadence of production. This approach is generally characterized by a very precise workflow, starting with project understanding and data collection and ending with the decision-making phase. This paper presents an exhaustive literature review of methods and applied tools for intelligent predictive maintenance models in Industry 4.0 by identifying and categorizing the life cycle of maintenance projects and the challenges encountered, and presents the models associated with this type of maintenance: condition-based maintenance (CBM), prognostics and health management (PHM), and remaining useful life (RUL). Finally, a novel applied industrial workflow of predictive maintenance is presented including the decision support phase wherein a recommendation for a predictive maintenance platform is presented. This platform ensures the management and fluid data communication between equipment throughout their life cycle in the context of smart maintenance.

A deep learning framework for sensor-equipped machine health indicator construction and remaining useful life prediction

Prognostic and health management (PHM) effectively reduces the economic loss of sensor-equipped machine downtime caused by under-maintenance and the waste of resources resulted from over-maintenance. The remaining useful life (RUL) prediction is the most critical step in PHM. However, accurate RUL prediction in a multiple sensors data environment faces challenges and difficulties. In this paper, we firstly consider the change point where the sensor-equipped machine drifts from a health state (initial stage) to the degradation stage, and combine the deep learning model with the change point to construct the health indicator (HI). Then, a long short-term memory model combined with an attention mechanism (LSTM_Att) is used to iteratively predict the future HI. Additionally, the predicted RUL distribution of the studied sensor-equipped machine is estimated using the similarity method based on the HI data of historical multiple sensor-equipped machines. Then, the confidence interval of RUL is obtained. Finally, the proposed method is verified on the publicly available turbofan engine degradation data set. The experimental results show that the proposed method outperforms the state-of-art benchmark methods.

Application of prognostic and health management in avionics system

Currently, most aircraft avionics systems are maintained based on reported failures or periodic system replacement. However, the evolution of prognostic and health management (PHM) concepts from mechanical to electronic systems and further to avionics system maintenance has been driven by changes in weapon platform procurement and support requirements. At the same time, with the increasing complexity of avionics design, integrated modular avionics (IMA) came into being. The appearance of IMA design concept marks the gradual transition of avionics system from distributed joint architecture to integrated architecture, which also provides the foundation for PHM technology to be applied to avionics system. This paper reviews the application and research status of predictive and health management system technology in avionics system.

Multi-criteria sensor placement determination in prognostics and health management using combined fault diagnosis, fault detection, and risk indexes

In framework of a prognostics and health management (PHM), a stochastic method is proposed for optimal sensor locations determination using three independent criteria indexes, reflecting the efficiency of sensor network: (i) the uncertainty of sensor information, the ability of fault diagnosis of a sensor network, (ii) the reliability of sensors, reflecting the fault detectability in sensor networks, (iii) risk of sensor failure including the consequence of sensor failure scenarios. The dynamic system failure model is developed and analyzed to consider the variation of environmental factors and their related failure threshold characterization along with statistical variance estimation as the information value that each possible sensor placement scenario provides through sensor information. A dynamic failure model is developed to incorporate the interaction of sensors and their corresponding components. It is demonstrated here that considering each criterion independently will not comprehensively determine the optimal placement scenario. An augmented index is formulated through Shannon Entropy theory, ranking all sensor placement scenarios based on proposed combined index. Optimization of sensor placement is demonstrated on a typical steam turbine. As shown in detail, the ranking based on each criterion provides different inconsistent rank for the location of the sensors. The combined index is found practically proper and recommended for rank and selection process.

Predictive Maintenance with PCA Approach for Multi Automated Railroad Crossing System (ARCS) in The Framework of Prognostic and Health Management (PHM) Planning

The existence of level crossings between railroads and road vehicles that do not have gates in areas far from crowds, such conditions require gates that are made automatically to avoid accidents. The Automated Railroad Crossing System (ARCS) is an automatically activated railroad crossing gate where train arrival information is obtained through sensors. For one level crossing, there are several electronic devices installed in the automatic railroad crossing system. It is planned that the automatic railroad crossing system will be installed at several level crossings. The problem is how to estimate the time to perform automatic railroad crossing maintenance at several different locations, For this reason, it is necessary to know the estimated remaining useful life (Remaining Useful Life) of the subsystems. The purpose of this research is to find the estimated remaining useful time (RUL) of the subsystem in the automatic railroad crossing system in order to estimate the time to perform maintenance. The process that is carried out to obtain the remaining useful time is through the Prognostic Health Management System development plan, while the analysis of the estimated remaining useful time is carried out using Principle Component Analysis (PCA), the results of this simulation show promising results to determine the estimated value of the remaining useful time. If it can be known the estimated remaining time of the benefit, it is hoped that the maintenance plan for each automatic railroad crossing system can be carried out more efficiently.

Failure analysis of repairable systems with k-out-of-m configuration considering with common cause failure and maintenance correlation based on goal oriented method

System reliability directly affects the application of Prognostic and Health Management (PHM) technology. This paper proposed a new Goal Oriented (GO) methodology for failure probability analysis of repairable systems with the Common Cause Failure (CCF) and maintenance correlation (MC). It should be noted that the considered repariable systems in this study are k-out-of-m configuration. Firstly, an improved GO algorithm solving the CCF is elaborated. Second, a failure analysis framework for repairable systems taken into account of CCF and MC is developed based on Markov theory. Finally, the new GO methodology is applied in analysising the dynamic unavailability of Hydraulic Oil Supply System (HOSS) of heavy vehicles. Comparisons with two traditional methods indicated that the proposed methodology provides higher efficiency. Furthermore, this work can widen the application of GO methodology.

Knowledge Informed Machine Learning using a Weibull-based Loss Function

Machine learning can be enhanced through the integration of external knowledge. This method, called knowledge informed machine learning, is also applicable within the field of Prognostics and Health Management (PHM). In this paper, the various methods of knowledge informed machine learning, from a PHM context, are reviewed with the goal of helping the reader understand the domain. In addition, a knowledge informed machine learning technique is demonstrated, using the common IMS and PRONOSTIA bearing data sets, for remaining useful life (RUL) prediction. Specifically, knowledge is garnered from the field of reliability engineering which is represented through the Weibull distribution. The knowledge is then integrated into a neural network through a novel Weibull-based loss function. A thorough statistical analysis of the Weibull-based loss function is conducted, demonstrating the effectiveness of the method on the PRONOSTIA data set. However, the Weibull-based loss function is less effective on the IMS data set. The results, shortcomings, and benefits of the approach are discussed in length. Finally, all the code is publicly available for the benefit of other researchers.

A Study on Deep Learning-Based Fault Diagnosis and Classification for Marine Engine System Auxiliary Equipment

Maritime autonomous surface ships (MASS) are proposed as a future technology of the maritime industry. One of the key technologies for the development of MASS is condition-based maintenance (CBM) based on prognostics and health management (PHM). The CBM technology can be used for early detection of abnormalities based on the database and for a prediction of the fault occurring in the future. However, this technology has a problem that requires a high-quality database that reproduces the operation state of the actual ships and quantitatively and systematically indicates the characteristics for the various fault state of the device. To solve this problem, this paper presents a study on the development method of the fault database based on the reliability. Firstly, the reliability analysis of the target device was performed to select five types of the core fault modes. After that, a fault simulation scenario that defined the fault simulation test methodology was drawn. A land-based testbed was built for the fault simulation test. The fault simulation database was developed with a total of 109 sets through the fault simulation test. Additionally, a fault classification algorithm based on deep learning is proposed. The classification performance was evaluated with a confusion matrix. The developed database will be expected to serve as the basis for the development CBM technology of MASS in the future.

Deep imbalanced regression using cost-sensitive learning and deep feature transfer for bearing remaining useful life estimation

Deep learning (DL) techniques have revolutionized the landscape of prognostic and health management (PHM), with the capability to learn discriminative representations from “big data”. However, realistic industry data often show imbalanced distributions, which greatly weakens the method relying on manually balanced datasets. To fill the research gap of bearing remaining useful life (RUL) estimation with imbalanced data, a novel framework is proposed to learn imbalanced regression using cost-sensitive learning and deep feature transfer (CSL-DFT), which introduces the idea of discretization and makes full use of techniques of imbalanced learning. Our CSL-DFT includes these main points: discretization & label distribution smoothing, deep feature transfer via CORrelation ALignment (CORAL), and cost-sensitive learning via class-balanced re-weighting. Considering this is the first application of deep imbalanced regression (DIR) in RUL prediction, a variety of imbalanced bearing training sets are designed based on experimental data, and verified the effectiveness of CSL-DFT. Comparison with other methods further shows its superior performance and rationality of design.

Prognostics and Health Management for Maintenance-Dependent Processes

Industrial components and systems undergo degradation in process operations. Prognostics and health management (PHM) is a process to assess and predict health conditions of components and can be applied to condition-based monitoring and maintenance. PHM is commonly utilized to analyze the health condition of a single lifecycle until failure. When maintenance occurs, degradation can be removed, and the PHM model can be restarted with new parameters related to the expected postmaintenance conditions. Maintenance actions, mostly imperfect repairs, may not entirely reset the condition to as good as new, and further degradation may occur at a higher rate. Maintenance repairs should be considered in prognostic models to predict component health more accurately. Furthermore, processes typically have more than one component that degrade and influence process measurements. The dependence of process measurements to multiple fault modes and related degradation can make individual component health monitoring complex. Commonly, faults and their related effects on process parameters must be isolated. In these cases, the diagnostics and prognostics framework should handle unsynchronized failure and maintenance reinitialization of different components for multiple fault processes. This research paper presents the Maintenance-Dependent Monitoring and Prognostics Model (MDMPM) to detect anomalies, decouple faults for different components, and predict future health conditions to calculate remaining useful life (RUL). The model is demonstrated with semisimulated nuclear power plant (NPP) data, with simultaneous condenser pump degradation and condenser tube fouling. The MDMPM shows a reliable prediction of RULs of NPP maintenance-dependent processes with interacting component degradation modes.

The Aramis Data Challenge to prognostics and health management methods for application in evolving environments

A recurrent difficulty for the effective application of Prognostics and Health Management (PHM) methods is related to the “evolving environments” in which industrial components typically operate. Several factors render the operational environments evolving, including deterioration of components, effects of maintenance activities and changes in working conditions. The issue of evolving environments is even more complicated for multi-component systems, where the degradation of one component can affect the degradation processes of other components, thus modifying their lifetime distributions and the statistical properties of the monitored signals. In an effort to convey research toward practical PHM solutions capable of dealing with evolving environments, the “Aramis challenge on degradation state assessment in evolving environments,” has been launched for the ESREL2020-PSAM15 conference. This work describes the Aramis Data Challenge and associated public dataset, illustrates the methods proposed for its solution and the related results obtained. For the evaluation of the goodness of the fault detection methods, an original metric is introduced, which is a variant of the timeliness metric that has been used in the PHM08 data challenge.