Remaining Useful Life Prediction Research Articles

A novel unsupervised adaptive density-based clustering filter for remaining useful life prediction of bearings

Abstract Constructing the health indicator (HI) and predicting the remaining useful life (RUL) are essential steps in bearing health management. Some prediction methods depend on prior information about HIs, especially when these indicators are generated by deep learning models. However, acquiring such prior information can be challenging in practical applications. This paper introduces a novel unsupervised adaptive density-based clustering filter (UADCF) for RUL prediction of bearings, which operates without the need for prior knowledge. Firstly, a post-hoc interpretation HI model (PIHIM) is proposed to characterize the deep learning constructed HIs from the perspective of what the deep learning has done. Then, leveraging the classical density-based clustering algorithm, we introduce the UADCF for unsupervised estimation of model parameters, which can dynamically adjust density parameters based on the current conditions. Finally, we develop a prediction framework combining PIHIM and UADCF, enabling unsupervised RUL prediction of bearings. The experimental studies validate the effectiveness of the proposed method.

A Novel Hybrid Approach for Remaining Useful Life (RUL) and Short-Term Capacity Prediction of Batteries

Lithium-ion battery management requires an accurate determination of the batteries’ remaining usable lives. A novel hybrid method used for the RUL and short-term capacity prediction of batteries is presented in this manuscript. The proposed technique is combined with a White Shark Optimizer (WSO) with the Multi-kernel Support Vector Machine (MSVM); therefore, it is called the WSO-MSVM technique. The major goal of this WSO-MSVM method is to attain effective future capacities and RUL forecast for the Lithium-Ion Battery (LIB) with reliable management of uncertainty. The hybrid technique solves the difficulty of forecasting the RUL of LIB. This proposal trains the WSO-MSVM model using the LIB data. The trained model is utilized to forecast the capacity and RUL of LIBs. In this proposed system, capacity, current, and voltage are considered in the discharge operation. The proposed technique validates good reliability and prediction accuracy with this suitable mechanism. Compared to the existing techniques like artificial neural network (ANN), ant lion optimizer-optimized artificial neural network, and adaptive network-based fuzzy inference system, the proposed technique shows fewer errors under charge and discharge conditions. In addition, the uncertainty intervals of the proposed WSO-MSVM technique for the predicted RUL may include the actual remaining life of the battery. The proposed technique’s charging error is lower the other exciting methods.

Prediction of Remaining Useful Life of Aero-engines Based on CNN-LSTM-Attention

Accurately predicting the remaining useful life (RUL) of aircraft engines is crucial for maintaining financial stability and aviation safety. To further enhance the prediction accuracy of aircraft engine RUL, a deep learning-based RUL prediction method is proposed. This method possesses the potential to strengthen the recognition of data features, thereby improving the prediction accuracy of the model. First, the input features are normalized and the CMAPSS (Commercial Modular Aero-Propulsion System Simulation) dataset is utilized to calculate the RUL for aircraft engines. After extracting attributes from the input data using a convolutional neural network (CNN), the extracted data are input into a long short-term memory (LSTM) network model, with the addition of attention mechanisms to predict the RUL of aircraft engines. Finally, the proposed aircraft engine model is evaluated and compared through ablation studies and comparative model experiments. The results indicate that the CNN-LSTM-Attention model exhibits superior prediction performance for datasets FD001, FD002, FD003, and FD004, with RMSEs of 15.977, 14.452, 13.907, and 16.637, respectively. Compared with CNN, LSTM, and CNN-LSTM models, the CNN-LSTM model demonstrates better prediction performance across datasets. In comparison with other models, this model achieves the highest prediction accuracy on the CMAPSS dataset, showcasing strong reliability and accuracy.

Method for Remaining Useful Life Prediction of Turbofan Engines Combining Adam Optimization-Based Self-Attention Mechanism with Temporal Convolutional Networks

Conducting the remaining useful life (RUL) prediction for an aircraft engines is of significant importance in enhancing aircraft operation safety and formulating reasonable maintenance plans. Addressing the issue of low prediction model accuracy due to traditional neural networks’ inability to fully extract key features, this paper proposes an engine RUL prediction model based on the adaptive moment estimation (Adam) optimized self-attention mechanism–temporal convolutional network (SAM-TCN) neural network. Firstly, the raw data monitored by sensors are normalized, and RUL labels are set. A sliding window is utilized for overlapping sampling of the data, capturing more temporal features while eliminating data dimensionality. Secondly, the SAM-TCN neural network prediction model is constructed. The temporal convolutional network (TCN) neural network is used to capture the temporal dependency between data, solving the mapping relationship of engine degradation characteristics. A self-attention mechanism (SAM) is employed to adaptively assign different weight contributions to different input features. In the experiments, the root mean square error (RMSE) values on four datasets are 11.50, 16.45, 11.62, and 15.47 respectively. These values indicate further reduction in errors compared to methods reported in other literature. Finally, the SAM-TCN prediction model is optimized using the Adam optimizer to improve the training effectiveness and convergence speed of the model. Experimental results demonstrate that the proposed method can effectively learn feature data, with prediction accuracy superior to other models.

Method for remaining useful life prediction of rolling bearings based on deep reinforcement learning.

In contemporary industrial systems, the prediction of remaining useful life (RUL) is recognized as a valuable maintenance strategy for health management due to its ability to monitor equipment operational status in real time and ensure the safety of industrial production. Current studies have largely concentrated on deep learning (DL) techniques, leading to a shortage of RUL prediction methods that utilize deep reinforcement learning (DRL). To further enhance application and research, this paper introduces a novel approach to RUL prediction based on DRL, specifically using a combination of Convolutional Neural Network-Bidirectional Long Short-Term Memory Network (CNN-BiLSTM) and the Deep Deterministic Policy Gradient (DDPG) algorithm. The proposed method reframes the conventional task of estimating RUL as a Markov decision process (MDP), effectively integrating the feature extraction capabilities of DL with the decision-making abilities of DRL. Initially, a hybrid CNN-BiLSTM is employed to establish an agent that can extract degradation features from raw signals. Subsequently, the DDPG algorithm within DRL is leveraged to develop the RUL prediction mechanism, completing the MDP by defining appropriate action spaces and reward functions. The agent, through repeated trials and optimization, learns to map the current operational state of the rolling bearing to its remaining service life. Validation analysis was performed on the intelligent maintenance systems (IMS) bearing dataset. The findings suggest that the DRL-based approach outperforms the current methodologies, demonstrating a superior performance in root mean square error (MSE) and MSE metrics. The predicted outcomes align more closely with the actual lifespan values.

Advanced data-driven techniques in AI for predicting lithium-ion battery remaining useful life: A comprehensive review

As artificial intelligence (AI) technology evolves, data-driven approaches are gaining attention in predicting lithium-ion battery's remaining useful life (RUL). Indeed, accurate RUL prediction is challenging, primarily because of the complex nature of the work and dynamic shifts in model parameters. To address these challenges, this article comprehensively explores five significant publicly accessible lithium-ion battery datasets, encompassing diverse usage conditions and battery types, offering researchers a rich repository of experimental data. In particular, we not only provide detailed information and access addresses for each dataset, but also present, for the first time, four innovative methods for battery aging health factor extraction. These methods, based on advanced AI techniques, are able to effectively identify and quantify key indicators of battery performance degradation, thereby enhancing the precision and dependability of RUL prediction. Additionally, the article identifies major challenges faced by current predictive techniques, including data quality, model generalization capabilities, and computational cost, highlighting the need for research focused on dataset diversity, multiple algorithm fusion, and hybrid physical-data-driven models to enhance prediction accuracy. We believe that this review will help researchers gain a comprehensive understanding of RUL estimation methods and promote the development of AI in battery.

An aircraft engine remaining useful life prediction method based on predictive vector angle minimization and feature fusion gate improved transformer model

Remaining useful life (RUL) prediction is crucial for achieving intelligent and predictive maintenance of aircraft engines. In practical applications, advance prediction values smaller than the true values can prevent serious deferred maintenance accidents. Using asymmetric loss functions directly leads to noticeable accuracy degradation, and existing methods often fail to satisfy accuracy and advance prediction requirements. To address this problem, this paper proposes a novel RUL prediction method based on the Prediction Vector Angle (PVA) minimization and Feature Fusion Gate (FFG) improved Transformer network. Specifically, the FFG is proposed to enhance Transformer prediction accuracy by dynamically fusing global and local features. The concept of PVA is first introduced based on the tilting properties of the RUL descent process. The target of the prediction model is cleverly transformed from error minimization to PVA minimization through the cosine similarity loss function. Various experiments on the CMAPSS dataset demonstrate the effectiveness of the proposed method in achieving high accuracy and advanced prediction. Compared to the state-of-the-art method, RMSE is reduced by at least 2.94 % and Score by 7.00 %. Finally, the PVA minimization mechanism significantly improves long short-term memory and convolutional neural network performance. The proposed method is noteworthy for its superiority and applicability.

Life prediction method of battery energy storage system in frequency modulation application

Abstract To tackle the challenge of lifespan reduction in lithium batteries during frequency modulation, this study introduces a novel Remaining Useful Life (RUL) prediction methodology. The proposed approach integrates Variational Mode Decomposition (VMD) with Gated Recurrent Unit (GRU) networks, thereby efficiently synthesizing data from both operational parameters and capacity metrics. Firstly, VMD is utilized to decompose the initial capacity data of lithium batteries into separate components characterized by high and low frequencies. Subsequently, for the high-frequency elements, GRU is employed for rolling iterative prediction, while for the low-frequency elements, features are extracted from the operational data and inputted into GRU for prediction. Finally, the component prediction results are restored to obtain capacity prediction values. Validation conducted using the NASA dataset shows that the root mean square error of capacity prediction is minimized to 0.0122, with a corresponding minimum average absolute error of 0.0096, and RUL prediction errors are generally within 2 cycles.

RUL prediction of lithium-ion battery based on BiLSTM

Abstract In this paper, in order to avoid the unavoidable degradation of the Remaining Useful Life(RUL) of lithium batteries in course of use, which may lead to the inability of lithium batteries to work normally or even cause safety accidents, finding a way to effectively predict the remaining life of lithium batteries in advance is significant. Therefore, the generator and discriminator models are constructed through a multilayer bidirectional length and Bidirectional Long Short-Term Memory(BiLSTM) model, which expands fewer data samples, making the sample data volume greatly improved, and avoiding the overfitting of the subsequent deep learning models under the fewer sample data. In BiLSTM, the hidden layer preserves the internal state of bidirectional data, and the key information is preserved and discarded through the internal Long Short-Term Memory(LSTM) module, so as not to cause any damage to the battery. information is preserved and discarded to avoid the problem of gradient disappearance that occurs when the data is dependent for a long time, and the GAN-BiLSTM model is suggested to deal with the issue of low precision and accuracy of deep learning methods with small samples.

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.

An embedding layer-based quantum long short-term memory model with transfer learning for proton exchange membrane fuel stack remaining useful life prediction

The battery management system (BMS) plays a critical role in electric vehicles (EVs), with the prediction of remaining useful life (RUL) being of utmost importance. This functionality enables real-time monitoring and enhances driver assistance through intelligent driving mechanisms. To improve the precision of stack voltage degradation and RUL prognostication for proton exchange membrane fuel cell (PEMFC) aging, especially with limited real-time training data, we propose a novel approach using a layer-enhanced quantum long short-term memory model with transfer learning. The proposed model is trained offline using historical stack voltage degradation data from a source stack, which is then fine-tuned and transferred to a target stack. Comparative evaluations demonstrate the model's robustness, with consistently low root mean square errors in stack voltage predictions, even with increased training data. The proposed model significantly improves RUL prediction accuracy, achieving a maximum relative accuracy enhancement of 1.33 % when initiating predictions at 712 h. These results affirm the effectiveness of our approach in enhancing the accuracy of PEMFC aging prognosis, thereby offering valuable insights for practical integration within the BMS framework for EVs.

Utilizing multiple inputs autoregressive models for bearing remaining useful life prediction

Accurate prediction of the remaining useful life (RUL) of rolling bearings is crucial in industrial production, yet existing models often struggle with limited generalization capabilities due to their inability to fully process all vibration signal patterns. We introduce a novel multi-input autoregressive model to address this challenge in RUL prediction for bearings. Our approach uniquely integrates vibration signals with previously predicted RUL values, employing feature fusion to output current window RUL values. Through autoregressive iterations, the model attains a global receptive field, effectively overcoming the limitations in generalization. Furthermore, we innovatively incorporate a segmentation method and multiple training iterations to mitigate error accumulation in autoregressive models. Empirical evaluation on the PMH2012 dataset demonstrates that our model, compared to other backbone networks using similar autoregressive approaches, achieves significantly lower root mean square error (RMSE) and Score. Notably, it outperforms traditional autoregressive models that use label values as inputs and non-autoregressive networks, showing superior generalization abilities with a marked lead in RMSE and Score metrics.

Combined improved tuna swarm optimization with graph convolutional neural network for remaining useful life of engine

AbstractAccurate prediction of the engine's remaining useful life (RUL) is essential to ensure the safe operation of the aircraft because. However, traditional deep‐learning based methods for RUL prediction has been limited by interpretability and adjustment for hyperparameters in practical applications due to the intricate potential relations during the degradation process. To address these dilemmas, an improved multi‐strategy tuna swarm optimization‐assisted graph convolutional neural network (IMTSO‐GCN) is developed to achieve RUL prediction in this work. Specifically, mutual information is used to describe potential relationships among measured parameters so that they could be utilized in building edges for these parameters. Besides, considering that not all relational nodes will positively affect the RUL prediction and the inherent hyperparameters of the GCN are high‐dimensional. Inspired by “No Free Lunch (NFL)”, IMTSO is proposed to reduce the optimization cost of hyperparameters, in which cycle chaotic mapping is employed to achieve initialization of the population for improving the uniformity of the initial population distribution. Besides, a novel adaptive approach is proposed to enhance the exploration and exploitation of tuna swarm optimization (TSO). The CMAPSS dataset was used to validate the effectiveness and advancedness of IMTSO‐GCN, and the experimental results show that the R2 of the IMTSO‐GCN is greater than 0.99, RMSE is less than 3, the Score error is within 1.

A novel spatio-temporal characteristic extraction network for bearing remaining useful life prediction

Abstract Remaining useful life (RUL) is an important index indicating the health status of equipment, which has attracted extensive attention. Nevertheless, many existing RUL prediction methods encounter difficulties in effectively capturing comprehensive degradation features hidden in the data. Moreover, within real-world industrial scenarios, noisy signals are inevitably collected in the raw signals, thereby posing a big challenge to the precision of RUL predictions. To address the aforementioned problems, a robust RUL estimation approach based on degradation intrinsic mode functions (IMFs) selection and spatio-temporal feature regression is developed in this paper. The former addresses the issue of deep learning models struggling to extract degradation features of rolling bearings due to interference factors in vibration signals, while the latter resolves the problem of incomplete degradation features extracted by traditional RUL models under complex operating conditions. Firstly, complete ensemble empirical mode decomposition with adaptive noise is adopted to process the raw signals, separating components with degradation features, ineffective components, and noise. Subsequently, an IMFs selection method employing fast dynamic time warping and cosine coefficients is designed to obtain the valuable degradation features. Finally, a spatio-temporal feature extraction network is presented to comprehensively and effectively capture both spatial and temporal features within the chosen degradation IMFs, achieving the prediction of RUL with high accuracy and strong robustness. The experimental part containing two case studies has validated the effectiveness and superiority of the proposed method.

Predicting the Future Capacity and Remaining Useful Life of Lithium-Ion Batteries Based on Deep Transfer Learning

Lithium-ion batteries are widely utilized in numerous applications, making it essential to precisely predict their degradation trajectory and remaining useful life (RUL). To improve the stability and applicability of RUL prediction for lithium-ion batteries, this paper uses a new method to predict RUL by combining CNN-LSTM-Attention with transfer learning. The presented model merges the strengths of both convolutional and sequential architectures, and it enhances the model’s capability to grasp comprehensive information by utilizing the attention mechanism, thereby boosting overall performance. The CEEMDAN algorithm is used for NASA batteries with obvious capacity regeneration phenomena to alleviate the difficulties caused by capacity regeneration on model prediction. During the model transfer phase, the CNN and LSTM layers of the pre-trained model from the source domain are kept unchanged during retraining, while the attention and fully connected layers are fine-tuned for NASA batteries and self-tested NCM batteries. The final results indicate that this method achieves superior accuracy relative to other methods while addressing the issue of limited labeled data in the target domain through transfer learning, thereby enhancing the model’s transferability and generalization capabilities.

A deep learning approach for supercapacitor remaining useful life prediction using pre-classifying strategy

Accurate prediction of the remaining useful life (RUL) of supercapacitors (SCs) is essential for the application of energy storage systems. In the existing end-to-end RUL prediction methods, there exist large errors in life prediction in SCs, especially in the early cycle prediction. This paper proposes an augmentation of the end-to-end RUL prediction method to improve the precision named pre-classifying. First, the pre-classifying is used to divide SCs’ cycling data into different classes to ensure consistency in each class. Then the end-to-end models are built for each class to predict the RULs. In this work, a novel hybrid model that combines convolutional neural networks and long short-term memory networks is proposed to capture spatiotemporal features from the cycling data. The experiment results demonstrate that the proposed hybrid model outperforms various deep neural networks in terms of prediction accuracy. When using the same hybrid model as the backbone, we report competitive results of the pre-classifying method on the public dataset with the root mean square error (RMSE) of 147.67 cycles, a 69.5% improvement compared with the RMSE of 484.18 cycles of the classical end-to-end RUL prediction model. Furthermore, experiment results indicate that the proposed pre-classification method significantly enhances the prediction accuracy of the standard end-to-end RUL prediction method with different deep neural network architectures. These results demonstrate that the pre-classifying method can be a general and effective methodology for RUL prediction of complex SC systems.

Hybrid firefly algorithm–neural network for battery remaining useful life estimation

Abstract Accurately estimating the remaining useful life (RUL) of batteries is crucial for optimizing maintenance, preventing failures, and enhancing reliability, thereby saving costs and resources. This study introduces a hybrid approach for estimating the RUL of a battery based on the firefly algorithm–neural network (FA–NN) model, in which the FA is employed as an optimizer to fine-tune the network weights and hidden layer biases in the NN. The performance of the FA–NN is comprehensively compared against two hybrid models, namely the harmony search algorithm (HSA)–NN and cultural algorithm (CA)–NN, as well as a single model, namely the autoregressive integrated moving average (ARIMA). The comparative analysis is based mean absolute error (MAE) and root mean squared error (RMSE). Findings reveal that the FA–NN outperforms the HSA–NN, CA–NN, and ARIMA in both employed metrics, demonstrating superior predictive capabilities for estimating the RUL of a battery. Specifically, the FA–NN achieved a MAE of 2.5371 and a RMSE of 2.9488 compared with the HSA–NN with a MAE of 22.0583 and RMSE of 34.5154, the CA–NN with a MAE of 9.1189 and RMSE of 22.4646, and the ARIMA with a MAE of 494.6275 and RMSE of 584.3098. Additionally, the FA–NN exhibits significantly smaller maximum errors at 34.3737 compared with the HSA–NN at 490.3125, the CA–NN at 827.0163, and the ARIMA at 1.16e + 03, further emphasizing its robust performance in minimizing prediction inaccuracies. This study offers important insights into battery health management, showing that the proposed method is a promising solution for precise RUL predictions.

A multi-source data fusion driven power field effect transistor health state assessment and remaining useful life prediction method

Abstract The metal oxide semiconductor field effect transistor (MOSFET) is subjected to various stresses due to the external and internal operating environments, leading to ageing and failure of the device. With multiple degradation mechanisms, a single piece of information can no longer fully reflect the health state of MOSFETs, so how to use multi-source data to characterise the state of the device and predict the remaining useful life (RUL) is an issue worth exploring. To address this problem, a method for constructing health indicators (HI) as well as predicting RUL using multi-source data is proposed. In this method, firstly, the features are computed by selecting the appropriate ageing signal from the ageing mechanism. Secondly, the extracted features are filtered using Pearson’s algorithm to find the features that are strongly correlated with longevity. Then, the filtered features are merged by dimensionality reduction using the kernel principal component analysis algorithm and the HI is constructed from the reduced result. Finally, an unsupervised clustering algorithm is used to classify the states based on the data distribution in HI, and the filtered features are used as input to the grey wolf optimisation bidirectional long short-term memory neural network to predict the RUL of the device. In this paper, the proposed method is validated using data from the MOSFET Accelerated Aging Experiment at the NASA Ames Centre of Excellence for Prediction. The results show that the method is able to achieve good results in health state assessment and RUL prediction of MOSFETs. The proposed method is an effective and comprehensive strategy that is more helpful for the operation and maintenance of electronics.

Data-driven predictive maintenance framework considering the multi-source information fusion and uncertainty in remaining useful life prediction

The existing equipment maintenance methods mainly separate the two related phases of prediction and predictive maintenance (PdM) by looking at remaining useful life (RUL) prediction without considering maintenance or optimizing maintenance schedules based on the given prediction information. To address this issue, we propose a framework based on a multi-stream attention fusion network with quantile regression model and deep reinforcement learning (DRL). In this novel framework, the multi-stream attention fusion block is used to comprehensively capture the operating status of industrial equipment and eliminate the extracted duplicate information. Quantile regression is employed to obtain the probabilistic RUL prediction results with uncertainty expression. We further formulate the PdM problem for DRL, where maintenance actions are triggered based on the estimates of the RUL distribution. To illustrate the superiority of our framework, we compare it with some state-of-the-art models using a public data and our private data. The experimental results indicate that our method exhibits high accuracy and stability in bearing RUL prediction and predictive replacement.

Prediction of residual life of rotating components based on adaptive dynamic weighting and gated double attention unit

Gears and bearings play vital roles as essential transmission components in mechanical drivetrains. Accurately predicting the remaining useful life (RUL) of these components is paramount to ensure optimal performance and prevent unexpected failures. To enhance the precision of RUL prediction, a novel method has been developed which involves constructing health indicators (HI) and implementing an adaptive dynamic weighting (ADW) on a gated dual attention unit (GDAU). The process commences by extracting multi-dimensional time-frequency domain features from vibration signals, which are then refined using an improved kernel principal component analysis (Adaptive Kernel Principal Component Analysis – AKPCA) to extract key components. Subsequently, the constructed HI is fine-tuned through an optimization process utilizing the exponentially weighted moving average method. Finally, the ADW strategy dynamically adjusts the input weights of the HI, and the GDAU model is employed to predict the RUL of gears and bearings. Experiment and comparison results have validated the effectiveness and advantages of the proposed method.