Remaining Useful Life Prediction Results Research Articles

Life Extension of Wind Turbine Drivetrains by means of SCADA Data: Case Study of Generator Bearings in an Onshore Wind Farm

Accurate predictions for remaining useful life (RUL) of wind turbine drivetrains is crucial in reducing downtime, optimising maintenance strategies, extending operational life and improving costs. The purpose of this study is to present a method which utilises SCADA data for RUL estimation. It aims to answer whether, by only having the basic SCADA data, normally collected from all wind turbines, valuable RUL prediction results can be obtained. The work intends to develop a reliable, accurate, user-friendly and informative tool to enable observation of any growing trends in the proposed metrics, such as temperature difference, cumulative sum of temperature difference and the moving average of the cumulative sum. These metrics are defined based on the differences between the actual temperatures of the components, that might have undergone some damage and the model predicted temperatures of those components if they would have remained healthy, throughout years of operation. A machine learning model is used, along with selected SCADA input parameters, to predict the healthy state temperature of components. The proposed method is implemented on SCADA data collected from an actual wind farm over seven years. The results of this study show that while the SCADA data analysis can contribute to fault detection, the estimation of RUL purely based on SCADA appears to be uncertain. Such tools however, can be used as a monitoring method, during operation, to record abnormality trend of the components over the years and can be used as important inputs for the life extension evaluation of wind turbine drivetrains.

Remaining useful life prediction with uncertainty quantification based on multi-distribution fusion structure

The issue of uncertainty in Remaining Useful Life (RUL) prediction based on the Deep Learning (DL) framework has gained increased attention in recent years. The probabilistic distribution is usually the effective way to capture the associated uncertainty in RUL prediction. However, most existing research relies on a single distribution that assumes a singular underlying pattern in the data. This type of approach restricts the ability to describe dynamic industrial processes and lacks robustness in adapting to the complexity of time-varying degradation. This paper proposes an approach to improve uncertainty quantification in RUL prediction based on the Multi-Distribution Fusion (MDF) structure. Initially, multiple possible RUL prediction results are produced. Subsequently, the MDF is used to integrate the former results with different weights and outputs the final RUL prediction distribution. The proposed method excels in uncertainty capturing in complex scenarios and provides a deeper understanding of the underlying dynamics of the monitoring data. The application of MDF resulted in more enhanced and robust uncertainty concerning RUL predictions. To validate the effectiveness of the proposed method, two neural networks, the Long Short-Term Memory (LSTM) network and Convolutional Neural Network (CNN), are individually combined with MDF and applied to the C-MAPSS datasets.

Adaptive staged remaining useful life prediction of roller in a hot strip mill based on multi-scale LSTM with multi-head attention

Accurate prediction of a roller's remaining useful life (RUL) is significant for a hot strip mill to avoid major safety accidents and substantial economic losses. Since rollers’ degradation processes represent multi-stage characteristics, traditional single-stage models cannot acquire accurate RUL prediction results. Therefore, this paper proposes an adaptive staged RUL prediction method based on multi-scale long short-term memory network with multi-head self-attention (LSTM-MHA). The roller's production data and operation data are fused to construct interpretable health indicators (HI), which can represent the roller's remaining rotatable angle. Then the degradation process is divided into multiple stages and analyzed. The proposed multi-scale LSTM-MHA can adaptively update the model weights based on changes in the roller degradation stages by the multi-scale memory structure. The MHA embedding mechanism filters important temporal information for LSTM units. The trained model can predict the roller's RUL by inputting the current roller's HI. The proposed method is verified on an industrial hot strip mill roller dataset from a well-known steel company. The validation results show that the prediction accuracy of the proposed method is higher than 98.98%. Compared with existing deep learn-based methods, the proposed multi-scale LSTM-MHA method has significant advantages in roller's RUL prediction.

A treelike framework combining fault diagnosis and RUL prediction

Most existing deep learning methods consider the remaining useful life (RUL) prediction problem under a single failure mode and cannot solve the RUL prediction problem with multiple failure modes coexisting caused by component coupling in actual engineering systems. Thus, considering these issues, this paper proposes a novel tree network framework to address fault classification and RUL prediction in parallel, and the RUL prediction results are fused output, which are suitable for bearing RUL prediction with multiple faults. First, this paper develops a fault recognizer combining a frequency domain classifier and deep convolutional neural network to improve model selection accuracy. Secondly, this paper proposes a feature fusion algorithm based on the Gini coefficient, and the fused indicators are input into the RUL prediction sub-network for model training. Finally, the RUL sub-network prediction results are dynamically weighted and fused with the fault classification results to obtain the RUL based on SoftMax. The bearing dataset XJTU-SY is introduced to verify the efficiency of the proposed method, and computational results show that the developed framework can effectively predict RUL compared with other traditional methods, especially for RUL prediction under multiple failure modes.

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.

Transfer learning based remaining useful life prediction of lithium-ion battery considering capacity regeneration phenomenon

Lithium-ion battery (LIB) has been widely used in various energy storage systems, and the accurate remaining useful life (RUL) prediction for LIB is critical to ensure the normal operation of system. However, the capacity regeneration (CR) phenomenon caused by the non-working state of LIB will seriously affect the capacity degradation trajectory of LIB, thus leading to an inaccurate RUL prediction result. To solve this problem, this paper proposes a non-parametric CR detection algorithm to detect the CR phenomenon online and quantitatively correct the prediction errors caused by different CR phenomena. In addition, a data reconstruction algorithm and a parameter update scheme are proposed to solve the problem of unit-to-unit difference among LIBs. Finally, the long and short term memory neural network-based transfer learning model is applied to predict the RUL of target LIB. The effectiveness of the proposed model is verified by the real LIB dataset of NASA. Compared with some existing RUL prediction models considering CR phenomenon, the proposed model exhibits higher accuracy and better universality.

Physics-informed multi-state temporal frequency network for RUL prediction of rolling bearings

Accurate prediction of remaining useful life (RUL) has been a key issue in the field of Prognostic and Health Management (PHM), which aims at predictive maintenance to improve equipment reliability and safety. Aiming at the problem that the existing RUL prediction methods are weak in perceiving long-term features and poor in capturing periodic dependence, which leads to inaccurate and seriously lagging RUL prediction results of rolling bearings, this article proposes a physics-informed bearing RUL prediction approach, namely multi-state temporal frequency network (MSTFN). Firstly, a physics-informed dynamic adaptive inverse discrete Fourier transform (IDFT) frequency domain block is constructed, which maps the spectral information from the known domain to the spectral interval of the unknown domain for extracting the periodic features of the sequence. Secondly, a residual self-attention multi-state gated control unit (RSA-MSGCU) is proposed, which incorporates a novel multi-state hierarchical division mechanism in memory cells to enhance the medium- and long-term feature perception capability. Based on RSA-MSGCU, a trend prediction time domain block is built to extract the trend features of the sequence. Finally, the periodic and trend features are fused to achieve compatibility between the IDFT physical model and the data-driven model and then make the final RUL prediction.

Remaining useful life prediction method of lithium-ion batteries is based on variational modal decomposition and deep learning integrated approach

A common method based on variational modal decomposition (VMD) and an integrated depth model is proposed to address the problem that it is difficult to precisely anticipate the remaining useful life (RUL) of lithium-ion batteries (LIBs). Initially, VMD is employed to decompose the LIBs capacity data in multiple scales to obtain the signal's global degradation tendency and local random fluctuation components. Then, the global degradation trend and each fluctuation component are modeled using an echo state network (ESN) and a Bayesian optimized long short-term memory (LSTM) network, respectively. The final LIBs RUL prediction results are obtained by integrating the prediction outcomes. On three public LIBs datasets with distinct degradation characteristics, the performance of the proposed model is tested, and alternative prediction algorithms are compared. The results of the experiments show that the proposed model's maximum average absolute percentage error does not exceed 0.43%. The average relative error not more than 0.5%, indicating great accuracy and stability in prediction.

Research on Calendar Aging for Lithium-Ion Batteries Used in Uninterruptible Power Supply System Based on Particle Filtering

The aging process of lithium-ion batteries is an extremely complex process, and the prediction of the calendar life of the lithium-ion battery is important to further guide battery maintenance, extend the battery life and reduce the risk of battery use. In the uninterruptible power supply (UPS) system, the battery is in a floating state for a long time, so the aging of the battery is approximated by calendar aging, and its decay rate is slow and difficult to estimate accurately. This paper proposes a particle filtering-based algorithm for battery state-of-health (SOH) and remaining useful life (RUL) predictions. First, the calendar aging modeling for the batteries used in the UPS system for the Shanghai rail transportation energy storage power station is presented. Then, the particle filtering algorithm is employed for the SOH estimation and RUL prediction for the single-cell battery calendar aging model. Finally, the single-cell SOH and RUL estimation algorithm is expanded to the pack and group scales estimation. The experimental results indicate that the proposed method can achieve accurate SOH estimation and RUL prediction results.

Predicting future capacity of lithium-ion batteries using transfer learning method

Lithium-ion (Li-ion) batteries have numerous applications, such as electric vehicles, power tools, medical devices, drones, satellites, and utility-scale storage. Remaining useful life (RUL) prediction is a challenging task due to the highly nonlinear degradation behavior and prediction of batteries dealing with unknown future data. Online RUL prediction for Li-ion batteries plays an important role in proper battery health management. To improve the prediction accuracy of RUL, we propose a novel hybrid method based on transfer learning to predict the future capacity of Li-ion batteries. The proposed method consists of integrating an ensemble empirical mode decomposition algorithm, a support vector regression model, and a bidirectional long short-term memory with attention mechanism model to predict the state of health (SOH), where SOH is the battery cycle life and discharge capacity measured in number of cycles. In addition, we also consider keen-onset effects. The performance of RUL predictions using different starting RUL prediction points is compared with experiments. The proposed method shows that the larger the cycle number of the RUL prediction starting point, the more accurate the RUL prediction results. The relative error values for the three different charging policy target batteries using the proposed method are 6.96 %, 0.6 %, and 6.25 %, respectively. The proposed method is effective for predicting the future capacity of Li-ion batteries and can be applied to predictive maintenance.

A Hybrid Model to Assess the Remaining Useful Life of Proton Exchange Membrane Fuel Cells

Durability and remaining useful life (RUL) prediction techniques are ones of the key issues for proton exchange membrane fuel cell (FC) commercialization. Herein, the performance degradation of an FC is analyzed based on the whole lifetime experimental data (up to 6500 h). The voltage model with different patterns is developed based on the voltage data, which can be easily measured. The mechanism model is developed based on the evolution of degradation indices reflecting the degradation state. However, the former is sensitive to the local and periodic changes in the voltage curve, leading to a large prediction error, and the latter requires aging data from complex and high-cost characterization, limiting the practical applications. Therefore, a hybrid prediction model combining the voltage and mechanism model is proposed where the respective weight of each model is dynamically determined based on their local prediction errors. The results reveal that the maximum errors in RUL prediction are 9.72%, 3.90% and 2.01% for the voltage, mechanism and hybrid model, respectively, and the RUL prediction results of the hybrid model are close to actual RUL when those of the voltage model are far from the accuracy zone, indicating that the hybrid model provides credible RUL predictions with the highest accuracy.

Remaining Useful Life Prediction for a Catenary, Utilizing Bayesian Optimization of Stacking

This article addresses the problem that the remaining useful life (RUL) prediction accuracy for a high-speed rail catenary is not accurate enough, leading to costly and time-consuming periodic planned and reactive maintenance costs. A new method for predicting the RUL of a catenary is proposed based on the Bayesian optimization stacking ensemble learning method. Taking the uplink and downlink catenary data of a high-speed railway line as an example, the preprocessed historical maintenance and maintenance data are input into the integrated prediction model of Bayesian hyperparameter optimization for training, and the root mean square error (RMSE) of the final optimized RUL prediction result is 0.068, with an R-square (R2) of 0.957, and a mean absolute error (MAE) of 0.053. The calculation example results show that the improved stacking ensemble algorithm improves the RMSE by 28.42%, 30.61% and 32.67% when compared with the extreme gradient boosting (XGBoost), support vector machine (SVM) and random forests (RF) algorithms, respectively. The improved accuracy prediction lays the foundation for targeted equipment maintenance and system maintenance performed before the catenary system fails, thus potentially saving both planned and reactive maintenance costs and time.

State of health and remaining useful life prediction of lithium-ion batteries based on a disturbance-free incremental capacity and differential voltage analysis method

State of health (SOH) and remaining useful life (RUL) prediction are crucial for battery management systems (BMS). However, accurate SOH and RUL prediction still need to be improved due to the complicated battery aging mechanism. This work combines incremental capacity analysis (ICA) and differential voltage analysis (DVA) based on the second-order RC model with an improved Bidirectional Gated Recurrent Unit (BiGRU) to develop SOH and RUL prediction framework. Firstly, the voltage is reconstructed through the second-order RC model to obtain the incremental capacity (IC) and differential voltage (DV) curves to avoid the influence of measurement noise and the complex parameter adjustment process in the filtering method on the IC and DV curves. Then, a new set of battery aging features are extracted from the reshaped IC and DV curves to improve SOH and RUL prediction accuracy and robustness. Next, the BiGRU method with attention mechanism (BiGRU-AM) is used to build the prediction models for battery aging features, SOH, and RUL. To reduce the impact of the capacity regeneration phenomenon, the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) method is used to decompose the SOH prediction results, and the decomposed residual is used as the input to improve the prediction accuracy of RUL. The uncertainty of RUL prediction results is analyzed by Monte Carlo (MC) simulation. Finally, the proposed method is verified by experimental battery data from Center for Advanced Life Cycle Engineering (CALCE) and Sandia National Laboratory. Experimental results show that the voltage reconstruction results based on the second-order RC model are applied to ICA and DVA analysis, effectively avoiding the influence of noise. The RMSE of voltage reconstruction is within 0.0006, and the Pearson correlation coefficient between the four aging features extracted from the reconstructed IC/DV curve and SOH is above 0.9. Moreover, this method has good robustness to the cell inconsistency, temperature uncertainty, and a satisfied generalization ability to different battery chemistries, which the maximum RUL predicted AE of CALCE and Sandia battery is within 10 and 5, respectively.

Prognostics of Remaining Useful Life for Lithium-Ion Batteries Based on Hybrid Approach of Linear Pattern Extraction and Nonlinear Relationship Mining

The accurate prediction of the remaining useful life (RUL) of lithium-ion batteries (LIBs) is a key, challenging research direction. In this study, a battery degradation model is built based on the LIB dataset of NASA. A data decomposition prediction method, which extracts the linear trend from the capacity degradation data and then predicts the residuals of time series with nonlinear relations, is proposed. Then, an autoregressive integrated moving average-long short-term memory (ARIMA-LSTM) combined model for predicting the RUL of LIBs is established. The linear trend of capacity degradation is predicted by the ARIMA model and the LSTM model is established to predict the nonlinear residuals separated from the capacity degradation. The summation of both obtains the final battery RUL prediction result. Compared with the LSTM model and prediction models of similar types in existing literature in recent two years, the experimental results show that the RUL prediction error of the proposed model in this article is no more than one cycle, which is smaller than that of all the models involved in the comparison. Thus, this method has higher prediction accuracy and model generalization ability in the prediction of the remaining useful life of LIBs.

Attention-based Dual-channel Deep Neural Network For Aero-engine RUL Prediction Under Time-varying Operating Conditions

For complex systems such as aerospace, remaining useful life (RUL) prediction is a general technique that provides information for decision-making in predictive maintenance. In the industrial field, RUL prediction under time-varying operating conditions is a challenging task. In this paper, an attention-based dual-channel deep neural network is proposed to fuse the time-varying operating conditions, with both prediction channels using long short-term memory (LSTM) neural networks. First, the features are extracted by a one-dimensional convolutional neural network (CNN). The operating conditions and sensor data are put into the dual-channel LSTM neural networks separately for prediction. The obtained results are combined with the attention mechanism to assign weights and finally put into the fully connected network for linear mapping to get the final RUL prediction results. This study is based on the N-CMAPSS dataset published by NASA. Compared with traditional methods, this method demonstrates its superiority and effectiveness.

Prediction of Remaining Useful Life of Lithium Batteries Based on WOA-VMD and LSTM

The remaining useful life (RUL) of a lithium-ion battery is directly related to the safety and reliability of the electric system powered by a lithium-ion battery. Accurate prediction of RUL can ensure timely replacement and maintenance of the batteries of the power supply system, and avoid potential safety hazards in the lithium-ion battery power supply system. In order to solve the problem that the prediction accuracy of the RUL of lithium-ion batteries is reduced due to the local capacity recovery phenomenon in the process of the capacity degradation of lithium-ion batteries, a prediction model based on the combination of the whale optimization algorithm (WOA)-variational mode decomposition (VMD) and short-term memory neural network (LSTM) was proposed. First, WOA was used to optimize the VMD parameters, so that the WOA-VMD could fully decompose the capacity signal of the lithium-ion battery and separate the dual component with global attenuation trend and a series of fluctuating components representing the capacity recovery from the capacity signal of the lithium-ion battery. Then, LSTML was used to predict the dual component and fluctuation components, so that LSTM could avoid the interference of the capacity recovery to the prediction. Finally, the RUL prediction results were obtained by stacking and reconstructing the component prediction results. The experimental results show that WOA-VMD-LSTM can effectively improve the prediction accuracy of the RUL of lithium-ion batteries. The average cycle error was one cycle, the average RMSE was less than 0.69%, and the average MAPE was less than 0.43%.

Uncertainty-Controlled Remaining Useful Life Prediction of Bearings with a New Data-Augmentation Strategy

The remaining useful life (RUL) of bearings based on deep learning methods has been increasingly used. However, there are still two obstacles in deep learning RUL prediction: (1) the training process of the deep learning model requires enough data, but run-to-failure data are limited in the actual industry; (2) the mutual dependence between RUL predictions at different time instants are commonly ignored in existing RUL prediction methods. To overcome these problems, a RUL prediction method combining the data augmentation strategy and Wiener–LSTM network is proposed. First, the Sobol sampling strategy is implemented to augment run-to-failure data based on the degradation model. Then, the Wiener–LSTM model is developed for the RUL prediction of bearings. Different from the existing LSTM-based bearing RUL methods, the Wiener–LSTM model utilizes the Wiener process to represent the mutual dependence between the predicted RUL results at different time instants and embeds the Wiener process into the LSTM to control the uncertainty of the result. A joint optimization strategy is applied in the construction of the loss function. The efficacy and superiority of the proposed method are verified on a rolling bearing dataset obtained from the PRONOSTIA platform. Compared with the conventional bearing RUL prediction methods, the proposed method can effectively augment the bearing run-to-failure data and, thus, improve the prediction results. Meanwhile, fluctuations of the bearing RUL prediction result are significantly suppressed by the proposed method, and the prediction errors of the proposed method are much lower than other comparative methods.

Condition-based maintenance method for multicomponent system considering maintenance delay based on remaining useful life prediction: Subsea tree system as a case

Maintenance resources and personnel may not arrive in time after formulating a maintenance strategy due to the particular working environment and poor accessibility for multicomponent systems. A condition-based maintenance (CBM) method considering maintenance preparation delay based on remaining useful life (RUL) prediction is proposed. Maintenance preparations under normal and emergency conditions are considered. Inspection time is dynamically set based on the maintenance preparation time and the RUL prediction result. At the inspection time point, RUL for each component in the multicomponent system is predicted based on offline inspection data according to the degradation trend, which is simulated by combining Gamma and Poisson processes. The next inspection time can be determined subsequently. A threshold based on RUL prediction results is set as the maintenance standard to evaluate the maintenance effectiveness. Inspection interval, safety threshold, and maintenance threshold are optimized by minimizing the maintenance cost. A case study of the subsea tree system for offshore oil and gas development is conducted to demonstrate the application of the proposed CBM methodology.

Prediction of Remaining Useful Life of the Lithium-Ion Battery Based on Improved Particle Filtering

Remaining useful life (RUL) prediction of lithium-ion batteries plays an important role in battery failure prediction and health management (PHM). By accurately predicting the RUL of the battery, the battery can be replaced accordingly, thereby effectively avoiding the occurrence of an accident and ensuring the normal operation of the entire system. In the prediction of the remaining service life of lithium-ion batteries, it is difficult to ensure accuracy due to the problem of particle degradation and the influence of singular values in the particle filter algorithm. In view of these problems, this article introduces the unscented Kalman algorithm to improve the particle filter algorithm from the perspective of re-weighting the particles, so as to improve the accuracy of the prediction results of the remaining service life of lithium-ion batteries. The improved particle filter is simulated and verified using the battery sample data in the Arbin experimental test platform. Comparing the simulation results with the traditional particle filter method, when the number of reference samples is the same, the PDF width of the prediction results of the improved particle filter algorithm is slightly smaller than that of the particle filter algorithm, indicating that the fluctuation of the prediction result is more accurate. It is proved that the improved particle filter method proposed in this article can provide more accurate battery RUL prediction results and can effectively improve the accuracy and robustness of the remaining service life prediction of lithium-ion batteries.

RETRACTED: Remaining useful life prediction of wind turbine generator based on 1D-CNN and Bi-LSTM

• The RUL prediction method for samples with unknown the performance degradation . • Extracted deep features to describe the performance degradation of the WTG . • Combined the repeat segmentation and the sliding window processing to generate samples. The remaining useful life (RUL) prediction has an important guiding role in the preventive maintenance of wind turbine generators (WTGs). In this article, a real-time dynamic perception model of the RUL prediction is proposed for WTGs, which contains the multi-state parameters processing, the performance degradation analysis, the performance degradation prediction and the RUL prediction. First, the degradation process is fitted to the random distribution on the basis of the principal component analysis (PCA) of the multi-state parameters. Secondly, the multivariate time series samples and the corresponding performance degradation amount are all brought into the 1-dimensional convolution neural network (1D-CNN) for the regression analysis. Then, the bidirectional long short memory (Bi-LSTM) neural network is set to predict the amount of performance degradation in time series to obtain the future trend of performance degradation. Finally, the result of the RUL prediction is obtained by contrasting the setting threshold of the performance degradation. The results show that the proposed model has lower regression analysis error and degradation prediction error than the single deep learning and traditional models, and can obtain more accurate and reliable RUL prediction results.