Remaining Useful Life Estimation Research Articles

Machine Learning Approach for LPRE Bearings Remaining Useful Life Estimation Based on Hidden Markov Models and Fatigue Modelling

Ball bearings are one of the most critical components of rotating machines. They ensure shaft support and friction reduction, thus their malfunctioning directly affects the machine’s performance. As a consequence, it is necessary to monitor the health conditions of such a component to avoid major degradations which could permanently damage the entire machine. In this context, HMS (Health Monitoring Systems) and PHM (Prognosis and Health Monitoring) methodologies propose a wide range of algorithms for bearing diagnosis and prognosis. The present article proposes an end-to-end PHM approach for ball bearing RUL (Remaining Useful Life) estimation. The proposed methodology is composed of three main steps: HI (Health Indicator) construction, bearing diagnosis and RUL estimation. The HI is obtained by processing non-stationary vibration data with the MODWPT (Maximum Overlap Discrete Wavelet Packet Transform). After that, a degradation profile is defined and coupled with crack initiation and crack propagation fatigue models. Lastly, a MB-HMM (Hidden Markov Model) is trained to capture the bearing degradation dynamics. This latter model is used to estimate the current degradation state as well as the RUL. The obtained results show good RUL prediction capabilities. In particular, the fatigue models allowed a reduction of the ML (Machine Learning) model size, improving the algorithms training phase.

A novel transformer-based DL model enhanced by position-sensitive attention and gated hierarchical LSTM for aero-engine RUL prediction

Accurate prediction of remaining useful life (RUL) for aircraft engines is essential for proactive maintenance and safety assurance. However, existing methods such as physics-based models, classical recurrent neural networks, and convolutional neural networks face limitations in capturing long-term dependencies and modeling complex degradation patterns. In this study, we propose a novel deep-learning model based on the Transformer architecture to address these limitations. Specifically, to address the issue of insensitivity to local context in the attention mechanism employed by the Transformer encoder, we introduce a position-sensitive self-attention (PSA) unit to enhance the model's ability to incorporate local context by attending to the positional relationships of the input data at each time step. Additionally, a gated hierarchical long short-term memory network (GHLSTM) is designed to perform regression prediction at different time scales on the latent features, thereby improving the accuracy of RUL estimation for mechanical equipment. Experiments on the C-MAPSS dataset demonstrate that the proposed model outperforms existing methods in RUL prediction, showcasing its effectiveness in modeling complex degradation patterns and long-term dependencies.

Using transformer and a reweighting technique to develop a remaining useful life estimation method for turbofan engines

Estimating the Remaining Useful Life (RUL) of industrial machinery is an important task in Prognostics and Health Management (PHM). Accurate RUL prediction based on current health status and real-time sensor measurements can help establish efficient maintenance strategies in advance, thus improving the productivity and efficiency of machinery operations. However, considering the nature of industrial machinery, a data imbalance problem, which exists in current RUL estimation datasets like the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) dataset, hinders data-driven approaches. To address the detrimental data imbalance problem and improve estimation performance, this research proposes an Adaptive RUL-wise Reweighting (ARR) technique. ARR uses kernel smoothing to maintain the inherent continuity of RUL and a reweighting method to rebalance the effects of samples with different ground-truth RULs. In addition, this research proposes a novel RUL estimation method that uses a transformer architecture, which has not yet been widely applied for PHM tasks. Using a transformer-based estimation model with high expressive power capable of handling multivariate time-series data has proven effective in RUL estimation. Comprehensive experiments using the C-MAPSS dataset based on cross validation are conducted to find optimal hyperparameter configurations and to validate the proposed method’s effectiveness in RUL estimation. For subsets FD001 and FD003 of the C-MAPSS dataset, the proposed method provides state-of-the-art estimation performance with the lowest Root Mean Squared Error (RMSE) values of 11.36 and 11.28 and score values of 192.22 and 133.41, respectively. The proposed method also shows highly competitive performance on subsets FD002 and FD004 of the C-MAPSS dataset.

An Attention-Based Deep Learning Approach for Lithium-ion Battery Lifespan Prediction: Analysis and Experimental Validation

The potential for lithium-ion batteries to become unstable can lead to operational malfunctions within the system and result in safety incidents. Therefore, accurately forecasting the remaining useful life (RUL) is beneficial in mitigating the likelihood of battery failure and prolonging its operational lifespan. Hence, precise estimation of RUL can help prevent numerous safety incidents and minimize resource wastage, presenting a significant and complex issue. This paper introduces a Deep Learning (DL) model that utilizes Long Short-Term Memory (LSTM) and attention mechanism to improve the accuracy of predicting the RUL of lithium-ion batteries. Initially, the battery capacity regeneration phenomenon is captured by applying four LSTM layers, followed by implementing an attention mechanism to align input and output sequences based on the content or semantics of the input sequence. Finally, the final prediction outcomes are generated via a Fully Connected (FC) layer. The efficacy of the proposed model is assessed through the utilization of the NASA dataset, and its performance is contrasted with various deep learning models to highlight its efficacy. Results from the experiments demonstrate that the suggested At-LSTM presents a robust option for forecasting the RUL of lithium-ion batteries, as it delivers superior results compared to all other models examined.

An adaptive model with dual-dimensional attention for remaining useful life prediction of aero-engine

As an indispensable part in prognostics and health management (PHM) of mechanical systems, the remaining useful life (RUL) estimation has been studied widely and intensively. Recently, with the boom in deep learning, the traditional data-driven prognostic methods like the temporal convolutional neural networks (TCNs), have made remarkable progress in the RUL estimation. However, the conventional TCNs methods do not have a well-established mechanism to parallel weigh the dual-dimensional features of multi-sensory data. Moreover, the conventional TCNs have a fixed network structure which is not flexible to learn deep temporal representations. Here, we propose an adaptive model with dual-dimensional attention for aero-engine RUL prognostics. Specifically, the dual-dimensional multi-head attention (DMHA) module is constructed to assign weights to features in the spatial and temporal dimensions, and then preforms the feature fusion. The mathematical derivations of the DMHA module parameter iterations are performed, which enhances the interpretability of the model. Next, the adaptive TCN (ATCN) is designed to further learn the time correlations of the fused features. Compared with the traditional TCN, the ATCN can adjust the network structure adaptively and extract the deep temporal representations better. Last, the prediction ability of the DMHA-ATCN is verified by conducting experiments on the C-MAPSS dataset.

Explainable RUL estimation of turbofan engines based on prognostic indicators and heterogeneous ensemble machine learning predictors

Data-driven prognostics of systems exploit sensor measurements to predict the degradation evolution and anticipate failures, corresponding to the estimation of the remaining useful life (RUL). This task uses feature engineering to build prognostic indicators (HI) and machine learning (ML) to estimate the RUL. However, high variability in data coming from similar systems operating under different conditions negatively affects the RUL performance. Hence, this paper presents a new methodology that combines feature and ML engineering methods to provide an explainable RUL prediction. The key contributions lie in constructing efficient prognostic indicators that isolate distinct profile trajectories, enabling adaptive RUL extraction for each system. An ensemble of heterogeneous ML predictors is also trained using these indicators and RUL trajectories, effectively addressing variability issues and enhancing RUL performance. The proposed methodology is rigorously validated using NASA-provided turbofan engine data (C-MAPSS), demonstrating its performance, compared to state-of-the-art results, with improved score and accuracy of prediction.

Hybrid Neural Networks for Enhanced Predictions of Remaining Useful Life in Lithium-Ion Batteries

With the proliferation of electric vehicles (EVs) and the consequential increase in EV battery circulation, the need for accurate assessments of battery health and remaining useful life (RUL) is paramount, driven by environmentally friendly and sustainable goals. This study addresses this pressing concern by employing data-driven methods, specifically harnessing deep learning techniques to enhance RUL estimation for lithium-ion batteries (LIB). Leveraging the Toyota Research Institute Dataset, consisting of 124 lithium-ion batteries cycled to failure and encompassing key metrics such as capacity, temperature, resistance, and discharge time, our analysis substantially improves RUL prediction accuracy. Notably, the convolutional long short-term memory deep neural network (CLDNN) model and the transformer LSTM (temporal transformer) model have emerged as standout remaining useful life (RUL) predictors. The CLDNN model, in particular, achieved a remarkable mean absolute error (MAE) of 84.012 and a mean absolute percentage error (MAPE) of 25.676. Similarly, the temporal transformer model exhibited a notable performance, with an MAE of 85.134 and a MAPE of 28.7932. These impressive results were achieved by applying Bayesian hyperparameter optimization, further enhancing the accuracy of predictive methods. These models were bench-marked against existing approaches, demonstrating superior results with an improvement in MAPE ranging from 4.01% to 7.12%.

A New Physics-Informed Condition Indicator for Cost-Effective Direct Current Solenoid Valves Using Significant Points of the Excitation Current

Abstract This paper proposes a new condition indicator (CI) for cost-effective industrial direct current (DC)-powered solenoid valves (SVs). First, a literature review of the failure modes for SVs is reported. The failure regarding the impeded movement of the valve plunger is considered. The proposed CI for this failure mode is computed from only some significant points of the SVs excitation current and their time instants. The proposed CI is evaluated by a large endurance campaign on a set of SVs. Then, a Remaining Useful Life (RUL) estimation algorithm is proposed and evaluated on the experimental data. Experimental results show the goodness of the proposed CI for RUL estimation of cost-effective industrial DC SVs.

A novel WaveNet-GRU deep learning model for PEM fuel cells degradation prediction based on transfer learning

Precise prediction of Remaining Useful Life (RUL) within the transportation industry is essential for cost reduction and enhanced energy efficiency, focusing on extending the operational lifespan of proton exchange membrane fuel cells (PEMFCs). In pursuit of this objective, this study employs data-driven prediction methodologies centered on deep neural networks and transfer learning. The fundamental premise is that these approaches hinge on the compatibility of functional conditions across diverse datasets. Multiple strategies, amalgamating transfer learning, and deep neural networks, are introduced to forecast the PEMFC stack's behavior and its associated RUL. Network hyperparameters are optimized through Bayesian optimization, targeting root-mean-square error (RMSE) minimization in voltage predictions. The efficacy of these prediction techniques is evaluated through essential performance metrics, including the mean absolute percentage error (MAPE), RMSE, and coefficient of determination (R2), applied to both voltage predictions and RUL estimations. For the first time, a WaveNet-GRU model has been developed. Comparative assessment of models trained on 50% of the dataset underscores its supremacy. This model attains R2, RMSE, and MAPE scores of 99.1, 2.16E-4, and 0.166E-1, respectively, in predicting stack voltage. Also, RUL has increased by 21% compared to the best contemporary research. The WaveNet-GRU model demonstrates exceptional transfer learning capabilities when applied to stacks influenced by current ripples. In this context, it achieves optimal R2, RMSE, and MAPE values of 99.69, 1.37E-4, and 0.31E-1, respectively.

A novel gear RUL prediction method by diffusion model generation health index and attention guided multi-hierarchy LSTM

Gears, as indispensable components of machinery, demand accurate prediction of their Remaining Useful Life (RUL). To enhance the utilization of ordered information within time series data and elevate RUL prediction precision, this study introduces the attention-guided multi-hierarchy LSTM (AGMLSTM). This innovative approach leverages attention mechanisms to capture the intricate interplay between high and low hierarchical features of the input data, marking the first application of such a technique in gear RUL prediction. Additionally, a refined health indicator (HI) is introduced, constructed through a diffusion model, to precisely reflect the gears' health condition. The proposed RUL prediction method unfolds as follows: firstly, HIs are computed from gear vibration data. Subsequently, leveraging the known HIs, AGMLSTM predicts future HIs, and the RUL of the gear is determined upon surpassing the failure threshold. Quantitative analysis of experimental results conclusively demonstrates the superiority of the proposed RUL prediction method over existing approaches for gear RUL estimation.

Remaining Useful Life Estimation of Hollow Worn Railway Vehicle Wheels via On-Board Random Vibration-Based Wheel Tread Depth Estimation.

The problem of remaining useful life estimation (RULE) of hollow worn railway vehicle wheels in terms of remaining mileage via wheel tread depth estimation using on-board vibration signals from a single accelerometer on the bogie frame is presently investigated. This is achieved based on the introduction of a statistical time series method that employs: (i) advanced data-driven stochastic Functionally Pooled models for the modeling of the vehicle dynamics under different wheel tread depths in a range of interest until a critical limit, as well as tread depth estimation through a proper optimization procedure, and (ii) a wheel tread depth evolution function with respect to the vehicle running mileage that interconnects the estimated hollow wear with the remaining useful mileage. The method's RULE performance is investigated via hundreds of Simpack-based Monte Carlo simulations with an Attiko Metro S.A. vehicle and many hollow worn wheels scenarios which are not used for the method's training. The obtained results indicate the accurate estimation of the wheels tread depth with a mean absolute error of ∼0.07 mm that leads to a corresponding small error of ∼3% with respect to the wheels remaining useful mileage. In addition, the comparison with a recently introduced Multiple Model (MM)-based multi-health state classification method for RULE, demonstrates the better performance of the postulated method that achieves 81.17% True Positive Rate (TPR) which is significantly higher than the 45.44% of the MM method.

Anomaly-informed remaining useful life estimation (AIRULE) of bearing machinery using deep learning framework

A rolling bearing is a crucial element within rotating machinery, and its smooth operation profoundly influences the overall well-being of the equipment. Consequently, analyzing its operational condition is crucial to prevent production losses or, in extreme cases, potential fatalities due to catastrophic failures. Accurate estimates of the Remaining Useful Life (RUL) of rolling bearings ensure manufacturing safety while also leading to cost savings.•This paper proposes an intelligent deep learning-based framework for remaining useful life estimation of bearings on the basis of informed detection of anomalies.•The paper demonstrates the setup of an experimental bearing test rig and the collection of bearing condition monitoring data such as vibration data.•Advanced hybrid models of Encoder-Decoder LSTM demonstrate high forecasting accuracy in RUL estimation.

A hybrid deep learning approach for remaining useful life prediction of lithium-ion batteries based on discharging fragments

Accurate remaining useful life (RUL) estimation is crucial for the normal and safe operations of lithium-ion batteries (LIBs). Traditionally, every cycle’s maximum discharging capacity should be measured and then serve as a model input to predict iteratively the degradation trajectory. Unfortunately, full discharge stages are not always present in practice. Herein, this study presents a hybrid approach consisting of signal decomposition and deep learning to overcome the above limitations. Firstly, for the collected discharging fragments, the convolutional neural networks model predicts every cycle’s maximum discharging capacity which combines to form a predicted capacity degradation curve before the start point of RUL prediction. Then, via empirical mode decomposition, this curve’s global degradation trend is extracted and serves as the subsequent model input. Finally, the entire degradation trajectory and RUL value could be inferred based on the well-trained gated recurrent unit-fully connected model. The superior prediction performance of the proposed method is verified on two open battery datasets. All the estimation errors can be maintained within 7.0% based on the discharging fragment of the ∼20% capacity ratio ranges from 40% to 60% of the degradation data. This result illustrates the promising accuracy and robustness of the developed LIBs RUL estimation method, especially for not full discharge process in practice.

GAN-Based Multi-Task Learning Approach for Prognostics and Health Management of IIoT

Health assessment (HA) & remaining useful life (RUL) estimation, the two essential pillars of the prognostics and health management (PHM) paradigm, help improve industrial equipment reliability while reducing maintenance costs. However, reported works treat HA and RUL estimation as disjoint problems though there exist unexploited similarities among these related issues. Additionally, in practical industrial working scenarios, equipment(s) stay in a normal condition for most of its lifespan, leading to a disproportionate training dataset, hampering the prediction accuracy. To overcome the above problems, we propose a data-driven multi-task learning framework aided by a novel least squares recurrent auxiliary classifier generative adversarial network (LS-R-ACGAN). LS-R-ACGAN employs recurrent neural networks (RNNs) in its generator & discriminator networks for multi-variate fault data generation while overcoming the vanishing gradient problem of ACGANs. Post-data-augmentation, a balanced training dataset, trains a multi-task learning model based on a deep gated RNN (DGRU) for joint HA and RUL estimation. Our simulations use the C-MAPSS dataset for testing the proposed approach’s accuracy. The final results showcase improvements by at least <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\sim$</tex-math> </inline-formula> 3.54% and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\sim$</tex-math> </inline-formula> 2.38% on the RMSE and Score metric over existing state-of-the-art works suggesting its competitiveness and competence for real-world implementations. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This study attacks multiple presumptions concerning the maintenance of complex systems including, ample availability of fault samples alongside normal samples and solving only one of the problems, i.e., health assessment (HA) or remaining useful life (RUL) estimations rather than both. Popular artificial intelligence (AI) based solutions rely heavily upon the quantity of training data for desired performance outcomes. However, in real-life industrial scenarios, fault instances are rare, inherently resulting in an imbalanced dataset. Additionally, it is both labor-intensive and costly to repeat multiple experiments for gathering required data as the equipment(s) pass through short-lived fault states. Thus conventional AI-based solutions may end up producing biased predictions. This work proposes to solve the imbalanced training dataset problem along with joint HA and RUL estimations, in one solution pipeline, currently unexplored in majority of the existing works in the literature. LS-R-ACGAN generates ample diverse fault samples for reducing the degree of imbalance between normal and fault classes. Subsequently, after data augmentation, the balanced training dataset helps train a multi-task learning model for joint HA and RUL estimations. The proposed framework has been implemented and tested on a benchmark dataset proving its superiority over multiple existing methods in the literature.

A Hybrid Predictive Maintenance Solution for Fault Classification and Remaining Useful Life Estimation of Bearings Using Low-Cost Sensor Hardware

In recent years, predictive maintenance tasks, especially for bearings, have become increasingly important. Solutions for these use cases concentrate on the classification of faults and the estimation of the Remaining Useful Life (RUL). As of today, these solutions suffer from a lack of training samples. In addition, these solutions often require high-frequency accelerometers, incurring significant costs. To overcome these challenges, this research proposes a combined classification and RUL estimation solution based on a Convolutional Neural Network (CNN) and a Long Short-Term Memory (LSTM) network. This solution relies on a hybrid feature extraction approach, making it especially appropriate for low-cost accelerometers with low sampling frequencies. In addition, it uses transfer learning to be suitable for applications with only a few training samples.

An unsupervised subdomain adversarial network for remaining useful life estimation under various conditions

AbstractDomain adaptation is a transfer learning method that is widely applied for remaining useful life (RUL) prediction under different operating conditions. However, current RUL prediction methods based on domain adaptation only consider minimizing degradation feature differences, while ignoring the correlation between features and labels, which may result in significant errors in the mapping relationship between the two learned by the model, and thus reduce the accuracy of prediction. To address this issue, this paper proposes an unsupervised subdomain adversarial network (USDAN) for RUL estimation under various conditions. On the basis of traditional domain adversarial adaptation networks, the network uses temporal convolutional networks to construct the feature extractor to learn the long‐term dependencies between degradation data, and a multilinear conditioning scheme is designed to combine the features and subdomain labels as input features to the domain discriminator. Finally, with backpropagation and gradient reversal, the feature extractor is prompted to generate cross‐domain invariant features that are correlated with the labels, improving the prediction accuracy and enhancing the transferability of the network. In the experimental section, the public turbofan engine degradation dataset is used to validate the effectiveness of the network, and the comparison with other approaches demonstrates its advantages.

MHT: A multiscale hourglass-transformer for remaining useful life prediction of aircraft engine

Remaining useful life (RUL) prediction of aircraft engines is significant in the health monitoring, operation, and maintenance of aircraft. Capturing more comprehensive device degradation trends at different time scales and extracting long-term dependencies effectively among elements in long time series are two challenges in the field of aircraft engine RUL estimation. To address the aforementioned challenges, this paper proposes a novel multiscale Hourglass-Transformer (MHT) aircraft engine RUL prognostics. Specifically, an hourglass-shaped multiscale feature extractor (HME) is designed based on one-dimensional convolutional neural network, which can scale the time sequence into multi-time scales for feature fusion. Then, a transformer network is employed to further extract features from the fused feature map and output the RUL. To enhance inter-scale data attention, a pyramid self-attention mechanism is employed in both the encoder and decoder. Finally, the superiority and effectiveness of this approach are verified on the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) dataset. Furthermore, the robustness and generalization capability of this method are further validated on New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS) dataset.

Joint evaluation and prediction of SOH and RUL for lithium batteries based on a GBLS booster multi-task model

From the perspective of safe electric vehicle operation, accurately assessing the state of health (SOH) and remaining useful life (RUL) of lithium batteries holds paramount importance. This paper introduces a novel multi-task learning data-driven model named GBLS Booster, focusing on the joint estimation and prediction of SOH and RUL. GBLS Booster integrates the strengths of GBLS, offering reduced computation, swift computing speed, and harnesses the powerful feature extraction capabilities of the CNN-Transformers algorithm-based Booster. Additionally, the Tree-structured Parzen Estimator (TPE) algorithm is applied to optimize the model.In this study, 10 healthy indicators (HIs) are devised to capture variations in battery SOH. These HIs are derived from readily available sensor data, encompassing current, voltage, and temperature information. The random forest method (RF) is employed to further refine features and minimize data dimensions. Concerning the RUL prediction, the capacity data is often plagued by significant noise. To address this challenge, the complete empirical mode decomposition (CEEMDAN) method is employed for noise reduction decomposition, followed by the utilization of the Pearson correlation coefficient to eliminate noisy data points.The proposed method is rigorously evaluated using the NASA dataset and CLACE dataset for modeling simulation and verification. Comparative analysis with other algorithms is conducted. The results demonstrate the superior performance of the proposed model, showcasing exceptional accuracy (with a minimum Mean Absolute Percentage Error (MAPE) of 0.3348 % for SOH and a minimum Relative Error (RE) of 0.01 % for RUL), robustness, and generalization capabilities.

Anomaly detection and multi-step estimation based remaining useful life prediction for rolling element bearings

The prediction of the Remaining Useful Life (RUL) of rotating machinery targets the avoidance of sudden machine failures and the optimization of the maintenance schedules as well as the management of repairs and replacements. A number of statistical model-based prognostics methodologies have been proposed in recent literature but a number of open challenges still exist, blocking their integration in the industry: (i) The extraction of a high-quality Health Indicator (HI) is critical for bearings’ prognostics but till now the trend of many classic HIs is often disturbed; deviating from a monotonical increase or decrease. (ii) The detection of the Start Prediction Time (SPT) on the HI is usually required in order to improve the accuracy of the RUL estimation but it can be incorrect especially when distinct random outliers occur. (iii) The widely applied RUL estimators, e.g. the Kalman filter (KF) and the Particle filter (PF), belong to the category of single-step estimation techniques that assume that the current step estimation depends only on the state of the previous step. However, the occurrence of bearings’ damage is usually not a sudden change but an accumulated process. Therefore a novel prognostic strategy for Rolling Element Bearings (REB) is proposed in this paper, which integrates a robust anomaly detection technique, the Support Vector Data Description (SVDD), with a multi-step estimation method, the Moving Horizon Estimation (MHE). Various advanced entropy and sparsity HIs are extracted by signals filtered at different frequency bands. By comparing all the extracted HIs based on a proposed criterion, a premium prognostic HI is chosen from a specific frequency band. Moreover, the SVDD, deployed and combined with several proposed constraints, can robustly detect the occurrence of a bearing fault and decide the proper SPT. Furthermore, the MHE is combined with two popular exponential statistical models and a second-order polynomial model for bearings’ prognostics. Experimental results from fifteen degraded bearings demonstrate that the MHE outperforms the classic Kalman filter and Particle filter. Moreover, the polynomial model achieves a better RUL estimation compared to the two classic exponential models in combination with the MHE.

Similarity learning hidden semi-Markov model for adaptive prognostics of composite structures

Data-driven methodologies have found increasing usage in the last decade for remaining useful life (RUL) prognostics of composite materials utilizing structural health monitoring (SHM) data. Of particular interest is the reliable RUL prediction in cases where the end-of-life is not in between the extreme values within the testing dataset. For example, when unexpected phenomena that severely compromise the structural integrity occur during the service life. Such cases are often referred as outliers and the RUL prognosis based on a data-driven model that learns from past data is often erroneous. This study addresses this challenge by proposing a new stochastic model; the Similarity Learning Hidden Semi Markov Model (SLHSMM), an extension of the Non-Homogenous Hidden Semi Markov Model (NHHSMM). Through the utilization of a nonparametric discrete distribution, which characterizes the similarity between the testing structure and the training structures, a dynamic re-estimation process is employed. This process assigns higher importance to the training structures that display greater similarity to the testing one. As a result, the estimated parameters effectively capture the specific characteristics of the testing structure. The training and testing SHM data sets consist of strain measurements collected from a case study where carbon–epoxy single-stringered panels, are subjected to constant, variable, and random amplitude fatigue loading until failure. RUL estimations from the SLHSMM, the NHHSMM, and the Gaussian Process Regression (GPR) are compared. The SLHSMM clearly outperforms its classical counterpart and GPR providing more accurate outlier and inlier prognostics, demonstrating its capability to adapt to unexpected phenomena and integrate unforeseen data into a prognostic platform.