Traditional Data-driven Methods Research Articles

Physics-informed neural networks for V-notch stress intensity factor calculation

This paper proposes a physics-informed neural networks (PINNs) based approach for elastic structures with a V-notch, by which the displacement field, stress field as well as the V-notch stress intensity factor (NSIF) can be obtained through artificial neural networks. A PINN model is established for V-notch structures, integrating physical information into a deep neural network to ensure adherence to physical laws while fitting observational data. Subsequently, an adaptive local sampling strategy for V-notch structures is adopted, generating locally dense Gaussian points sampling around regions of stress concentration. Based on this, a sequential PINNs approach for V-notch structures is then established to calculate the NSIF for V-notch structures with arbitrary notch angles. Finally, the effectiveness of the proposed method is validated through three numerical examples. The results demonstrate the method can accurately predict the NSIFs for V-notch structures across a spectrum of opening angles. Compared to the traditional data-driven method, the proposed method is able to more effectively compute the NSIF of V-notch structures due to the integration of physical information and observational data.

Fatigue life prediction of cold expansion hole using physics-enhanced data-driven method

Cold expansion (CE) serves as a practical surface enhancement to improve the fatigue life of hole structures by improving surface integrity in both macro-scale and micro-scale. Due to the inaccessibility and high cost of experimental measurements, the physical relation between surface integrity and fatigue life are always implicit, serving as the major challenge for accurate life prediction. To address this issue, a novel method is proposed by introducing physical information to traditional data-driven method, where surface integrity enriched by multi-scale simulation is mapped to fatigue life via machine learning (ML) mechanism. As integrated to four typical ML algorithms, the proposed physics-enhanced data-driven method exhibit outstanding capability for accuracy improvement, decreasing the scatter band by amplitude between 27.3 % and 71.4 %. The proposed method offers a promising option for fatigue life prediction on surface treated structures with limited physical information.

Joint Physics and Data Driven Geometrical Factors of Array Laterolog in Layered Formations

Numerical geometrical factors (GFs) are developed for array laterolog measurements in one-dimensional (1D) cylindrically and planarly layered formations. The kernel of GFs assumes that laterolog responses can be expressed in terms of linear superposition of formation resistivities at different units. Two techniques, namely physics-driven modeling and data-driven approximation, have been employed to ensure both computational accuracy and efficiency of GFs. Physics-driven modeling utilizes a three-dimensional finite element algorithm to generate high-precision and fully labeled GFs data and library. Subsequently, these predetermined data are trained offline using classical neural network algorithm to establish an implicit relationship between formation parameters and GFs. The joint physics and data-driven GFs presents three main advantages. Firstly, it enables the direct representation of the spatial sensitivity of layered formations both laterally and horizontally in vertical and horizontal wells. Secondly, GFs can be used to rapidly calculate the array laterolog responses. The computation speed can be enhanced by thousands of times in comparison with traditional physic-driven modeling, allowing for real-time data processing. Another noteworthy aspect is that GFs forward modeling can handle formations with arbitrary layers, thereby resolving the fixed-layer modeling issue inherent in traditional data-driven methods. Thirdly, the GFs can be employed to approximate the derivative of logging response with respect to formation resistivity, significantly reducing the complexity of Jacobian matrix calculation for deterministic inversion. These numerical GFs not only serve as an excellent alternative simulator for array laterolog, but also will be helpful to accelerate the modeling and inversion of other logging measurement in layered structures.

A novel DRL-guided sparse voxel decoding model for reconstructing perceived images from brain activity

BackgroundDue to the sparse encoding character of the human visual cortex and the scarcity of paired training samples for {images, fMRIs}, voxel selection is an effective means of reconstructing perceived images from fMRI. However, the existing data-driven voxel selection methods have not achieved satisfactory results. New methodHere, a novel deep reinforcement learning-guided sparse voxel (DRL-SV) decoding model is proposed to reconstruct perceived images from fMRI. We innovatively describe voxel selection as a Markov decision process (MDP), training agents to select voxels that are highly involved in specific visual encoding. ResultsExperimental results on two public datasets verify the effectiveness of the proposed DRL-SV, which can accurately select voxels highly involved in neural encoding, thereby improving the quality of visual image reconstruction. Comparison with existing methodsWe qualitatively and quantitatively compared our results with the state-of-the-art (SOTA) methods, getting better reconstruction results. We compared the proposed DRL-SV with traditional data-driven baseline methods, obtaining sparser voxel selection results, but better reconstruction performance. ConclusionsDRL-SV can accurately select voxels involved in visual encoding on few-shot, compared to data-driven voxel selection methods. The proposed decoding model provides a new avenue to improving the image reconstruction quality of the primary visual cortex.

Beyond seen faults: Zero-shot diagnosis of power circuit breakers using symptom description transfer

Power circuit breakers (CBs) are vital for the control and protection of power systems, yet diagnosing their faults accurately remains a challenge due to the diversity of fault types and the complexity of their structures. Traditional data-driven methods, although effective, require extensive labeled data for each fault class, limiting their applicability in real-world scenarios where many faults are unseen. This paper addresses these limitations by introducing symptom description transfer-based zero-shot fault diagnosis (SDT-ZSFD), a method that leverages zero-shot learning for fault diagnosis. Our approach constructs a fault symptom description (FSD) framework, which embeds a fault symptom layer between the feature layer and the label layer to facilitate knowledge transfer from seen to unseen fault classes. The method utilizes current and acceleration signals collected during CB operation to extract features. By applying sparse principal component analysis to these signals, we derive high-quality features that are mapped to the FSD framework, enabling effective zero-shot learning. Our method achieves a satisfactory recognition rate by accurately diagnosing unseen faults based on these symptoms. This approach not only overcomes the data scarcity problem but also holds potential for practical applications in power system maintenance. The SDT-ZSFD method offers a reliable solution for CB fault diagnosis and provides a foundation for future improvements in symptom-based zero-shot diagnostic mechanisms and algorithmic robustness.

A label-free battery state of health estimation method based on adversarial multi-domain adaptation network and relaxation voltage

The state of health (SOH) estimation of lithium-ion batteries is crucial for the operational reliability and safety of electric vehicles. However, traditional data-driven methods face problems of label shortage and domain discrepancy caused by diverse battery types and operating conditions. This paper proposes a label-free SOH estimation method based on adversarial multi-domain adaptation technique and relaxation voltage (RV). Firstly, the raw RV and integral voltage are proposed to construct the two-dimensional input sequence to ensure high adaptability across various target domains. Secondly, a two-dimensional convolutional neural network integrated with fully connected layers is developed to establish the feature extractor and SOH estimator, which paves the way for effective knowledge transfer. Furthermore, an adversarial multi-domain adaptation method with a domain discriminator is developed to optimize the extraction of domain-invariant features and enable accurate SOH estimation without SOH labels. Datasets of 50 batteries with different temperatures and materials are used for the validation. The proposed method shows outstanding performance, achieving the average RMSE and MAE of 0.0274 and 0.0240 across various working temperatures, and 0.0177 and 0.0148 across different battery types. It suggests that the proposed method can achieve robust adaptive battery SOH estimation across multiple target domains without using SOH labels.

A method for predicting methane production from anaerobic digestion of kitchen waste under small sample conditions.

Anaerobic digestion (AD) has the great potential to treat organic waste and achieve remarkable results effectively. However, it is very tough to establish an accurate mechanistic model for this process. Data-driven modeling technology has opened a new door to solving this problem. While when the sample set is small, traditional data-driven modeling methods are often powerless. In this paper, an effective method is proposed for data-driven high-precision modeling in small sample scenarios. A time series generative adversarial network (TimeGAN) is first utilized to augment the original high-quality small-sample data collected during the AD methane production. A novel hybrid kernel extreme learning machine (HKELM) is then designed to form a better structure of the data-driven model, whose regularization coefficient C0 is optimized by the sparrow search algorithm (SSA). Finally, this semi-finished model (SSA-HKELM) is trained by the augmented data to form the final mathematical model (TimeGAN-SSA-HKELM) for the AD methane generation process. Comparative experiments of the methane daily production prediction error haveverified the effectiveness of the method, which can be extended to other similar small sample data-driven modeling scenarios.

Time series modeling and forecasting of epidemic spreading processes using deep transfer learning

Traditional data-driven methods for modeling and predicting epidemic spreading typically operate in an independent and identically distributed setting. However, epidemic spreading on complex networks exhibits significant heterogeneity across different phases, regions, and viruses, indicating that epidemic time series may not be independent and identically distributed due to temporal and spatial variations. In this article, a novel deep transfer learning method integrating convolutional neural networks (CNNs) and bi-directional long short-term memory (BiLSTM) networks is proposed to model and forecast epidemics with heterogeneous data. The proposed method combines a CNN-based layer for local feature extraction, a BiLSTM-based layer for temporal analysis, and a fully connected layer for prediction, and employs transfer learning to enhance the generalization ability of the CNN-BiLSTM model. To improve prediction performance, hyperparameter tuning is conducted using particle swarm optimization during model training. Finally, we adopt the proposed approach to characterize the spatio-temporal spreading dynamics of COVID-19 and infer the pathological heterogeneity among epidemics of severe acute respiratory syndrome (SARS), influenza A (H1N1), and COVID-19. The comprehensive results demonstrate the effectiveness of the proposed approach in exploring the spatiotemporal variations in the spread of epidemics and characterizing the epidemiological features of different viruses. Moreover, the proposed method can significantly reduce modeling and predicting errors in epidemic spread to some extent.

Model-data hybrid driven approach for remaining useful life prediction of cutting tool based on improved inverse Gaussian process

In the cutting process, reasonable evaluation of tool remaining useful life (RUL) is essential to ensure machining stability and safety. However, traditional data-driven methods rely on datasets with complete degradation data and prior knowledge, which are usually difficult to obtain in real scenarios. For these problems, this paper proposed a model-data hybrid driven approach for RUL prediction of cutting tool based on improved inverse Gaussian process. The tool wear degradation is considered as a random process, the cutting physical model and the data-driven stochastic process model are fused, so that the prior knowledge from the physical model and the statistical law of measured data are fully utilized to improve the accuracy of tool RUL prediction. The cutting physical model is established, and the prior knowledge of tool wear degradation is obtained based on physical simulation method. A data-driven random effects inverse Gaussian process model is constructed for tool RUL modeling and prediction. According to Bayesian method, the unknown parameters are estimated and updated integrating physical simulated and measured wear data, so as to obtain the probability density function for specific tool RUL, and quantify the prediction uncertainty. Finally, comparative experiments are conducted to validate the performance of the proposed method. The results suggest that the hybrid approach has the best performance in different tool RUL prediction.

Remaining Useful Life Prediction Based on Deep Learning: A Survey.

Remaining useful life (RUL) is a metric of health state for essential equipment. It plays a significant role in health management. However, RUL is often random and unknown. One type of physics-based method builds a mathematical model for RUL using prior principles, but this is a tough task in real-world applications. Another type of method estimates RUL from available information through condition and health monitoring; this is known as the data-driven method. Traditional data-driven methods require significant human effort in designing health features to represent performance degradation, yet the prediction accuracy is limited. With breakthroughs in various application scenarios in recent years, deep learning techniques provide new insights into this problem. Over the past few years, deep-learning-based RUL prediction has attracted increasing attention from the academic community. Therefore, it is necessary to conduct a survey on deep-learning-based RUL prediction. To ensure a comprehensive survey, the literature is reviewed from three dimensions. Firstly, a unified framework is proposed for deep-learning-based RUL prediction and the models and approaches in the literature are reviewed under this framework. Secondly, detailed estimation processes are compared from the perspective of different deep learning models. Thirdly, the literature is examined from the perspective of specific problems, such as scenarios where the collected data consist of limited labeled data. Finally, the main challenges and future directions are summarized.

Identification of Milling Cutter Wear State under Variable Working Conditions Based on Optimized SDP

Traditional data-driven tool wear state recognition methods rely on complete data under targeted working conditions. However, in actual cutting operations, working conditions vary, and data for many conditions lack labels, with data distribution characteristics differing between conditions. To address these issues, this article proposes a method for recognizing the wear state of milling cutters under varying working conditions based on an optimized symmetrized dot pattern (SDP). This method utilizes complete data from source working conditions for representation learning, transferring a generalized milling cutter wear state recognition model to varying working condition scenarios. By leveraging computer image processing features, the vibration signals produced by milling are converted into desymmetrization dot pattern images. Clustering analysis is used to extract template images of different wear states, and differential evolution algorithms are employed to adaptively optimize parameters using the maximization of Euclidean distance as an indicator. Transfer learning with a residual network incorporating an attention mechanism is used to recognize the wear state of milling cutters under varying working conditions. The experimental results indicate that the method proposed in this paper reduces the impact of working condition changes on the mapping relationship of milling cutter wear states. In the wear state identification experiment under varying conditions, the accuracy reached 97.39%, demonstrating good recognition precision and generalization ability.

A deep learning method for solving thermoelastic coupling problem

Abstract The study of thermoelasticity problems holds significant importance in the field of engineering. When analyzing non-Fourier thermoelastic problems, it was found that as the thermal relaxation time increases, the finite element solution will face convergence difficulties. Therefore, it is necessary to use alternative methods to solve. This paper proposes a physics-informed neural network (PINN) based on the DeepXDE deep learning library to analyze thermoelastic problems, including classical thermoelastic problems, thermoelastic coupling problems, and generalized thermoelastic problems. The loss function is constructed based on equations, initial conditions, and boundary conditions. Unlike traditional data-driven methods, this approach does not rely on known solutions. By comparing with analytical and finite element solutions, the applicability and accuracy of the deep learning method have been validated, providing new insights for the study of thermoelastic problems.

The Remaining Life Prediction of Rails Based on Convolutional Bi-Directional Long and Short-Term Memory Neural Network with Residual Self-Attention Mechanism

In the railway industry, the rail is the basic load-bearing structure of railway tracks. The prediction of the remaining useful life (RUL) for rails is important to avoid unexpected system failures and reduce the cost of maintaining the system. However, the existing detection of rail flaws is difficult, the rail deterioration mechanisms are diverse, and the traditional data-driven methods have insufficient feature extraction. This causes low prediction accuracy. With objectives set in relation to the problems outlined above, a rail RUL prediction approach based on a convolutional bidirectional long- and short-term memory neural network with a residual self-attention (CNNBiLSTM-RSA) mechanism is designed. Firstly, the pre-processed vibration data are taken as the input for the convolutional bi-directional long- and short-term memory neural network (CNNBiLSTM) to extract the forward and backward dependencies and features of the rail data. Secondly, the RSA mechanism is introduced in order to obtain the contributions of the features at different moments during the degradation process of the rail. Finally, an end-to-end RUL prediction implementation based on the convolutional bi-directional long- and short-term memory neural network with the residual self-attention mechanism is established. The experiments were carried out using the full life-cycle data of rails collected at the railway site. The results show that the method achieves a higher accuracy in the RUL prediction of rails.

Physics-informed State-space Neural Networks for transport phenomena

This work introduces Physics-informed State-space neural network Models (PSMs), a novel solution to achieving real-time optimization, flexibility, and fault tolerance in autonomous systems, particularly in transport-dominated systems such as chemical, biomedical, and power plants. Traditional data-driven methods fall short due to a lack of physical constraints like mass conservation; PSMs address this issue by training deep neural networks with sensor data and physics-informing using components’ Partial Differential Equations (PDEs), resulting in a physics-constrained, end-to-end differentiable forward dynamics model. Through two in silico experiments – a heated channel and a cooling system loop – we demonstrate that PSMs offer a more accurate approach than a purely data-driven model. In the former experiment, PSMs demonstrated significantly lower average root-mean-square errors across test datasets compared to a purely data-driven neural network, with reductions of 44 %, 48 %, and 94 % in predicting pressure, velocity, and temperature, respectively.Beyond accuracy, PSMs demonstrate a compelling multitask capability, making them highly versatile. In this work, we showcase two: supervisory control of a nonlinear system through a sequentially updated state-space representation and the proposal of a diagnostic algorithm using residuals from each of the PDEs. The former demonstrates PSMs’ ability to handle constant and time-dependent constraints, while the latter illustrates their value in system diagnostics and fault detection.

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.

Physics-informed graphical neural network for power system state estimation

State estimation is highly critical for accurately observing the dynamic behavior of the power grids and minimizing risks from cyber threats. However, existing state estimation methods encounter challenges in accurately capturing power system dynamics, primarily because of limitations in encoding the grid topology and sparse measurements. This paper proposes a physics-informed graphical learning state estimation method to address these limitations by leveraging both domain physical knowledge and a graph neural network (GNN). We employ a GNN architecture that can handle the graph-structured data of power systems more effectively than traditional data-driven methods. The physics-based knowledge is constructed from the branch current formulation, making the approach adaptable to both transmission and distribution systems. The validation results of three IEEE test systems show that the proposed method can achieve lower mean square error more than 20% than the conventional methods.

EXPLORING SOME SPATIALLY CONSTRAINED DELINEATION METHODS IN SEGMENTING THE MALAYSIAN COMMERCIAL PROPERTY MARKET

This study delves into the property submarket in Kuala Lumpur and Selangor, Malaysia. The submarket is anticipated to be simple, uniform, and dense, making it highly influenced by neighbouring properties. However, traditional data-driven methods that overlook spatial contiguity disregard this density condition. To tackle this problem, the study investigates spatially constrained data-driven methods utilizing Principal Component Analysis (PCA) and cluster analysis. The findings reveal that spatially constrained methods outperform traditional methods by minimizing errors and enhancing model fit. Specifically, the two-step cluster method and k-means cluster method reduce errors by 6.96% and 7.22%, respectively, but at the cost of model fit by 11.23% and 13.94%. Conversely, the spatial k-means and spatial agglomerative hierarchical cluster methods reduce errors by 8.68% and 8.17%, respectively, while improving model fit by 7.1% and 6.35%. Hence, the study concludes that spatially constrained data-driven methods are more effective in differentiating commercial property submarkets than traditional methods.

A hierarchical enhanced data-driven battery pack capacity estimation framework for real-world operating conditions with fewer labeled data

Battery pack capacity estimation under real-world operating conditions is important for battery performance optimization and health management, contributing to the reliability and longevity of battery-powered systems. However, complex operating conditions, coupling cell-to-cell inconsistency, and limited labeled data pose great challenges to accurate and robust battery pack capacity estimation. To address these issues, this paper proposes a hierarchical data-driven framework aimed at enhancing the training of machine learning models with fewer labeled data. Unlike traditional data-driven methods that lack interpretability, the hierarchical data-driven framework unveils the “mechanism” of the black box inside the data-driven framework by splitting the final estimation target into cell-level and pack-level intermediate targets. A generalized feature matrix is devised without requiring all cell voltages, significantly reducing the computational cost and memory resources. The generated intermediate target labels and the corresponding features are hierarchically employed to enhance the training of two machine learning models, effectively alleviating the difficulty of learning the relationship from all features due to fewer labeled data and addressing the dilemma of requiring extensive labeled data for accurate estimation. Using only 10% of degradation data, the proposed framework outperforms the state-of-the-art battery pack capacity estimation methods, achieving mean absolute percentage errors of 0.608%, 0.601%, and 1.128% for three battery packs whose degradation load profiles represent real-world operating conditions. Its high accuracy, adaptability, and robustness indicate the potential in different application scenarios, which is promising for reducing laborious and expensive aging experiments at the pack level and facilitating the development of battery technology.

Planetary gearbox fault diagnosis based on FDKNN-DGAT with few labeled data

Although data-driven methods have been widely used in planetary gearbox fault diagnosis, the difficulty and high cost of manual labeling leads to little labeled training data, which limits the classification performance of traditional data-driven methods. Therefore, the semi-supervised fault diagnosis method with few labeled samples becomes one of the main research directions. Graph attention network (GAT) is distinguished from traditional classification network by using graph structure for fault node information aggregation and feature extraction, which is an effective semi-supervised learning algorithm. This paper uses fast Fourier transform to process the original vibration signal of gearbox and use it as graph nodes, and propose a KNN graph construction method using pooling for fuzzy distance calculation. In addition, this paper improves the distribution of attention weights by introducing dynamic graph attention networks to correct the problem that classical static GATs cannot clearly distinguish the weights of different categories of nodes. Experiments show that the method proposed in this paper can better extract fault features in complex gearbox vibration signals with an accuracy of more than 99% with very few labeled samples, and has better diagnostic performance compared with other graph neural network architectures and traditional classification networks.

State of Charge Estimation of Lithium-Ion Battery Based on Back Propagation Neural Network and AdaBoost Algorithm

The accurate estimation of the state of charge (SOC) of lithium-ion batteries is critical in battery energy storage systems. This paper introduces a novel approach, the AdaBoost–BPNN model, to overcome the limitations of traditional data-driven estimation methods, such as a low estimation accuracy and poor generalization ability. The proposed model employs a back propagation neural network (BPNN) for the preliminary estimation. Subsequently, an AdaBoost–BPNN model is developed as a strong learner using the AdaBoost integration algorithm. Each BPNN sub-model serves as a weak learner within the AdaBoost framework. The final output of the strong learner is obtained by combining the individual outputs from the weak learners using weighting factors. This adaptive adjustment of weighting factors enhances the accuracy of SOC estimation. The proposed SOC estimation algorithm is evaluated and validated through experimental analysis. Throughout the paper, theoretical analysis is conducted, and the proposed AdaBoost–BPNN model is validated and verified using experimental results. The results demonstrate that the AdaBoost–BPNN model outperforms traditional methods in accurately estimating SOC under various conditions, including constant current-constant voltage (CCCV) charging, dynamical stress testing (DST), US06, a federal urban driving schedule (FUDS), and pulse discharge conditions.