Performance Degradation Process Research Articles

Machine cross-domain remaining useful life prediction via contrastive adversarial variational recurrent method

The performance degradation process of the machine is non-stationary which varies with the operating environment and is reflected as a temporal shift phenomenon. Models or methods that assume the test and training sets have the same distribution will no longer be suitable for solving problems where domain shifts exist. This brings new challenges for accurately evaluating the machine’s remaining useful life (RUL). This paper studies a contrastive adversarial variational recurrent method for machine RUL prediction under different service conditions. In this new approach, the variational recurrent networks are developed to extract distribution features and latent spaces of raw data, meanwhile, the adversarial strategies are applied to enable the learning of domain-invariant features to make the model achieve cross-domain task processing. To supervise the learning process of the model to learn more mutual information between the extracted features and the raw input data, a contrastive loss is also introduced in the proposed method. Sufficient experiments were conducted to verify the feasibility and superiority of the suggested approach, including 12 sets of cross-scenario tests on the turbofan engine dataset C-MAPSS from NASA. Experimental findings confirm that the proposed method performs satisfactorily and competitively even compared to current state-of-the-art works.

Exponential Dispersion accelerated degradation modelling and reliability assessment considering initial value and processes heterogeneity

In the production and operation, inherent variability and uncertainty necessitate addressing unit-to-unit heterogeneity in initial performance values and degradation processes. This article presents a bi-stochastic exponential dispersion process (BS-ED) designed to account for heterogeneity in both initial performance values and degradation processes. First, based on the ED process, the time and acceleration covariates are introduced to form a nonlinear accelerated ED process, and a random effect coefficient associated with the accelerated stress is incorporated to consider the heterogeneity of the process. Meanwhile, through the modelling of degradation time-shift, a degradation model considering the stochastic initial value of the product performance is developed. To effectively conduct the statistic inference of the BS-ED process, an improved stochastic EM algorithm is proposed, and the information matrix and Ito calculus are combined to estimate the confidence intervals. Finally, the stability of the method is verified by simulation and analyzed by two real cases.

Performance degradation assessment of rolling bearings based on the comprehensive characteristic index and improved SVDD

Rolling bearing is one of the most critical parts of mechanical equipment, so the performance degradation assessment of rolling bearing is vital to ensure the normal operation of the whole mechanical equipment. Aiming at the problems that a single degradation characteristic can only contain limited performance degradation information of rolling bearings, a large number of redundant characteristics exist in the high-dimension characteristic set resulting in the inability to effectively mine the characteristic information of rolling bearings, and that traditional degradation assessment models are not suitable for the shortage of fault data during the actual operation of rolling bearings, a performance degradation assessment method of rolling bearings based on the comprehensive characteristic index and improved support vector data description (SVDD) is proposed in this paper. Firstly, to solve the parameter selection problem of variational mode decomposition (VMD), a parameter-adaptive VMD method based on salp swarm algorithm based on mixed strategy (MSSSA) is proposed. Secondly, to extract the performance degradation information of rolling bearings more comprehensively and fully, the comprehensive characteristic index is proposed. Then, a kernel locality preserving projection orthogonal kernel principal component analysis (KLPPOKPCA) method is proposed to reduce the dimensionality of the extracted multi-domain characteristic set of the rolling bearing. Finally, a support vector data description with negative samples (NSSVDD) is proposed and optimized by MSSSA to solve the problem that traditional degradation assessment models are not suitable for the shortage of fault data during the actual operation of rolling bearings and improve the detection performance of abnormal data. The experimental results show that the proposed method can accurately divide the performance degradation process of the rolling bearing. Moreover, the comparison with other methods further highlights the superiority of the proposed method in determining the point in time of early fault of the rolling bearing.

Remaining useful life prediction method for Degradation–Shock dependence: Case of a subsea hydraulic control system

Predicting the remaining useful life (RUL) is crucial for ensuring reliable equipment operation. This study addresses the challenges of RUL prediction in modeling the effects of external shocks and internal degradation, with emphasis on the dependency relationship between the two factors. A novel hybrid method for RUL prediction is proposed. The proposed method combines degradation models and field data to realize RUL assessment under the influence of system degradation and external shocks. A dependency factor is introduced to effectively incorporate the influence of degradation stage and shock intensity on system failure. A comprehensive uncertainty analysis is performed, in consideration of random degradation, unpredictable shock events, and the dynamic evolution of degradation stages and performance degradation processes. Numerical simulations are conducted on a hydraulic control system to demonstrate the effectiveness and accuracy of the proposed method. The robustness and reliability of the proposed approach in predicting the RUL of the hydraulic control system are demonstrated, with particular emphasis on the method's capability to handle the complexities of dependency modeling and uncertainty analysis. Results show that the proposed method can realize dependency modeling of the hydraulic control system, accurately predict its RUL, and analyze the vulnerability of its subsystems.

Cascading failure and resilience optimization of unmanned vehicle distribution networks in IoT

With the development of science and technology, unmanned vehicles have become one of the main ways of distribution, and the appearance of unmanned vehicles undoubtedly brings great convenience to people. However, few researchers consider the cascading failure times for different situations in unmanned vehicle distribution networks. Especially, few researchers have integrated them into maintenance strategies to real-time path changes to increase the reliability of accomplishing distribution tasks and improve resilience after congestion. One of the key issues is how to improve the resilience of distribution network and what maintenance strategies should be developed. In this paper, cascading failures are analyzed and resilience is optimized in unmanned vehicle distribution network. First, cascading failures for different situations are analyzed, and two importance-based diversion methods are proposed to describe the three processes of congestion formation, persistence, and evacuation. Second, failure and performance degradation processes are investigated. Third, the Floyd time-varying resilience optimization algorithm is proposed, and the calculation method of resilience under different strategies is given. At last, simulation results show that the proposed strategy 3 improves the original resilience by 101 % compared with the traditional waiting path strategy.

Efficient light-driven peroxymonosulfate activation by CoSx/TiOx nanorod for rapid degradation of levofloxacin: Performance and continuous degradation process

The preparation of cobalt-based compounds with the same morphology and the comparison of their performance in activating peroxymonosulfate (PMS) to remove pollutants is still rarely reported. In this paper, CoMx/TiOx nanorods were prepared by a simple glycol-mediated approach. Among CoMx/TiOx (M = S, N, P, O), CoSx/TiOx + PMS system displayed the greatest degradation rate for levofloxacin (Levo) irradiated by simulated solar light, completing 100 % degradation of Levo during 5 mins with k (rate constant) = 0.797 min−1, which could be improved by roughly 2.8 times, 12.2 times and 18.1 times compared to CoNx/TiOx (0.276 min−1), CoPx/TiOx (0.065 min−1), CoTiO3 (0.044 min−1) respectively. The remarkable catalytic performance of CoSx/ TiOx might be due to its highest specific surface area, optimal electron transfer capacity, and greatest capacity to activate PMS, according to DFT calculations and experiment results. Under several operating factors (pH, inorganic ions and NOMs, water bodies, and pollutants), the CoSx/TiOx catalyst still maintained excellent degradation performance of Levo, which fully demonstrated its exceptional anti-interference and practical water treatment ability. Singlet oxygen (1O2) was crucial to the degradation of Levo in the CoSx/TiOx + PMS + Light system. The Levo degradation route was inferred and the decrease in toxicity of the intermediate was seen. A self-assembled flow-bed unit could effective purification of 1140 mL Levo with 80 mg catalyst. This study offers a straightforward approach for creating an efficient and stable cobalt-based composite catalysts for applications in advanced oxidation processes (AOPs).

Early Prediction of Student’s Performance

Abstract: Performance degradation assessment (PDA) is of great significance to ensure safety and availability of mechanical equipment. As an important issue of PDA, the robustness of the trained model directly affects the assessment efficiency and restricts its application in practice. This paper proposes a robust modeling approach based on Student's t-hidden Markov model (Student's t-HMM) and nuisance attribute projection (NAP). NAP can remove nuisance attributes caused by individual differences from the feature space. Student's t-HMM utilizes the finite Student's t-mixture models (SMMs) to describe the observation emission densities associated with each hidden state, which can be more tolerant towards outliers than conventional HMMs. Based on these two techniques, the proposed method is supposed to be more robust and can assess the performance degradation process of new objects based on data of tested objects.

Striking a Balance between Carbon Mediated Cathode Degradation and SEI Engineering Towards Stable High-Performance Solid-State Batteries

Solid-state batteries (SSBs) can render a paradigm shift in battery safety and energy density. The promise hinges on integrating high-capacity alkali metal electrodes with state-of-the-art inorganic solid electrolytes and high-energy Li-ion cathodes. However, the development has been restricted, mainly owing to the limited understanding and partial overcoming of the interfacial electro-chemo-mechanical issues at the electrode-electrolyte interfaces, leading to inferior power capability, energy efficiency, and long-term stability of the cell.1 Sulfide-type solid electrolytes (SEs) like Li6PS5Cl, Li10GeP2S12, and Li3PS4 are promising for commercial SSB development, but they seem to suffer from interfacial stability issues with the high-energy NMC (LiNixCoyMnzO2) cathodes, particularly when a conductive carbon additive is employed, which is required for improved capacity and high current density operations. Earlier reports suggest that the carbon enables a faster electronic percolation pathway, leading to a quicker NMC-SE interfacial degradation over cycling. Parallelly, the reactive functional groups on the carbon surface may aggravate the decomposition pathway by reacting with the thiophosphate SEs.2–4 , 5 To shut down this conductive additive-led decomposition and to stabilize NMC-sulfide SE-based SSBs, often a protective coating of LiNbO3, LiZrO3, Li3PO4, etc., is applied on the NMC particles in addition to avoiding the use of carbon in the cathode composite. Relatively lesser-known approaches looked into the effect of modifying the conductive additive surface itself with a non-conductive polymer or insulator-type material to tame the degradation mechanism led by the electron percolation pathway.1,6–9 In this context, by thorough experimentation, we have established an in-depth understanding of carbon’s role in the NMC performance degradation process. This led to the reckoning that one of the insulating degradation products can be in situ generated with functionalized carbon as the conducting additive, rendering stable solid electrolyte interphase (SEI) during electrochemical cycling. We found that the controlled degradation kinetics with an optimally functionalized carbon material help achieve the required conductivity (low impedance) at the cathode for improved energy and power densities and a stable cathode SEI for durable SSB cycling. In conclusion, we present a novel and facile idea of engineering the conductive carbon additive for achieving high-capacity, long-life NMC-based SSBs without the complication of expensive coating strategies.Reference: Ding, Z., Li, J., Li, J. & An, C. Review-Interfaces: Key Issue to Be Solved for All-Solid-State Lithium Battery Technologies. (2020)Koerver, R. et al. Redox-active cathode interphases in solid-state batteries. J. Mater. Chem. A 5, 22750–22760 (2017).Walther, F. et al. Influence of Carbon Additives on the Decomposition Pathways in Cathodes of Lithium Thiophosphate-Based All-Solid-State Batteries. Chem. Mater. 32, 6123–6136 (2020).Walther, F. et al. Visualization of the Interfacial Decomposition of Composite Cathodes in Argyrodite-Based All-Solid-State Batteries Using Time-of-Flight Secondary-Ion Mass Spectrometry. Chem. Mater. (2019)Park, S. W. et al. Graphitic Hollow Nanocarbon as a Promising Conducting Agent for Solid-State Lithium Batteries. Small 15, (2019).Takada, K. et al. Interfacial modification for high-power solid-state lithium batteries. Solid State Ionics 179, 1333–1337 (2008).Banerjee, A., Wang, X., Fang, C., Wu, E. A. & Meng, Y. S. Interfaces and Interphases in All-Solid-State Batteries with Inorganic Solid Electrolytes. Chem. Rev. 120, 6878–6933 (2020).Zhang, Y. et al. Direct Visualization of the Interfacial Degradation of Cathode Coatings in Solid State Batteries: A Combined Experimental and Computational Study. Adv. Energy Mater. 10, 1903778 (2020).Randau, S. et al. On the Additive Microstructure in Composite Cathodes and Alumina-Coated Carbon Microwires for Improved All-Solid-State Batteries. Chem. Mater. 33, 1380–1393 (2021). Figure 1

Personalized Transfer Learning Framework for Remaining Useful Life Prediction Using Adaptive Deconstruction and Dynamic Weight Informer

The precise remaining useful life (RUL) prediction of turbofan engines benefits maintenance decisions. The training data quantity and quality are crucial for effective prediction modeling and accuracy improvement. However, the performance degradation process of the same type of turbofan engine usually exhibits different trajectories because of engines’ differences in degradation degrees, degradation rates, and initial health states. In addition, the initial part of the trajectory is a stationary health stage, which contains very little information on degradation and is not helpful for modeling. Considering the differential degradation characteristics and the requirement for accurate prediction modeling of the same type of turbofan engines with individual differences, we specifically propose a personalized transfer learning framework for RUL prediction by answering three key questions: when, what, and how to transfer in prediction modeling. The framework tries to maximumly utilize multi-source-domain data (samples of the same type of engines that run to failure) to improve the training data quantity and quality. Firstly, a transfer time identification method based on a dual-baseline performance assessment and the Wasserstein distance is designed to eliminate the worthless part of a trajectory for transfer and prediction modeling. Then, the transferability of each sample in the multi-source domain is measured by an approach, named the time-lag ensemble distance measurement, and then the source domain is ranked and adaptively deconstructed into two parts according to transferability. Ultimately, a new training loss function considering the transferability of the weighted multi-source-domain data and a two-stage transfer learning scheme is introduced into an informer-based RUL prediction model, which has a great advantage for long-time-series prediction. The simulation data of 100 of the same type of turbofan engine with individual differences and five comparison experiments validate the effectiveness and accuracy of the proposed method.

Bearing performance degradation assessment using adversarial fusion convolutional autoencoder based on multi-source information

Bearing operation states will directly determine the performance of the equipment; thus, monitoring operation status and degradation indicators is the key to ensuring continuous and healthy operation of the equipment. However, most of the research uses single-source information data, which makes it difficult to model when dealing with multi-source information, complex data distribution, and noise. In this paper, a bearing performance degradation assessment method based on multi-source information is proposed to comprehensively utilize the data signals of different structures, spaces, types, and sources. First, the adversarial fusion convolutional autoencoder is constructed for obtaining the degradation index of the bearing, while the adversarial learning strategy is applied to achieve the effect of enhancing the robustness and sensitivity of the degradation indicators extracted by the network. Then the degradation index is input into the support vector data description to determine the fault anomalies of the degradation index adaptively and the fuzzy c-means algorithms to obtain the final rolling bearing performance degradation evaluation results. Through the verification results of two experiment datasets, it is found that the proposed model can achieve accurate evaluation and quantitative analysis of the performance degradation process of bearings. As a result, the entire network ensures the reconstruction accuracy of normal samples while simultaneously stretching the reconstruction error of abnormal samples to achieve accurate monitoring of degradation onset.

Improving electrochemical performance of LiNi0.5Mn1.5O4 positive electrodes via regulated cathode electrolyte interphase

LiNi0.5Mn1.5O4 has been thoroughly investigated as one of the most promising high-voltage positive materials for Li-ion batteries, owing to its low cost, high energy density and good stability. Nonetheless, the limited reversible capacity and poor cycle efficiency limit its use to the high-end digital domain. Herein, the SmF3 cathode electrolyte interphase modified strategy has been first explored for LiNi0.5Mn1.5O4, which demonstrated outstanding electrochemical performance. The surface regulated sample demonstrates a preferred initial discharge-specific capacity of 137.85 mAh g−1 at a 1 C rate and sustains a discharge capacity of 110.21 mAh g−1 after 200 cycles. More importantly, SmF3 can successfully reduce the occurrence of adverse reactions while producing a thin and homogenous CEI layer. The basic understanding of cathode electrolyte interphase alteration on surface morphology, crystal structure, and chemical components of the modified samples are revealed through means of various testing methods. The surface modification with SmF3 can effectively mitigate the formation of generated by-products and enhance the transport of Li+. This technique of modifying the cathode electrolyte interphase with SmF3 serves as a starting point for further research into the performance degradation process and opens up a new avenue for improving the electrochemical performance of the spinel positive materials.

Evolution of photoelectric conversion and device stability of PM6:N2200 all-polymer solar cells

In this work, to study the performance degradation process in PM6:N2200 all-polymer solar cells, including the optical absorption, morphology, photo-physical process of the active layer and current–voltage characteristics of the device were analyzed by employing multiple characterization techniques. The results showed that the morphology (phase sizes and phase separation scales) of PM6:N2200 would change with time. Unstable morphology of active layer determined the changeable photoelectric conversion process in operating solar cells. Besides, we found that morphological change of PM6:N2200 had a little influence on the exciton behaviors (diffusion and dissociation) within 40 days. However, it showed an undesired impact on photogenerated charge behaviors (transport and recombination). Unstable morphology of active layer and high charge recombination probability were essential reasons for the performance degradation of polymer solar cells in operating time.

Bearing performance degradation assessment and remaining useful life prediction based on data-driven and physical model

Intelligent health maintenance of bearings usually consists of two stages: constructing effective health assessment indicators and accurate remaining useful life prediction models. However, many prediction models are available only when many constraints are met, and the health assessment indicators may not be able to accurately track the performance degradation process of bearings. This study proposes a bearing performance degradation assessment and remaining life prediction method. First, the health evaluation index family (referred to as the generalized high-order moment coefficient) was constructed based on the generalized power mean and high-order origin moments for health assessment. Subsequently, an improved Paris–Erdogan model is proposed, which uses the optimal health evaluation index as input to predict the remaining life of the bearing after the initial failure. The experimental results show that the proposed method has a higher performance degradation tracking accuracy and a smaller prediction error than the combination of traditional statistical indicators and prediction models.

Solid oxide electrolysis cells – Interplay between operating conditions, fuel electrode overpotential and degradation

With the integration of increased share of energy from renewable, but fluctuating, energy sources there will be an increased demand for large-scale flexible conversion of electrical energy into easily storable chemical fuels. Electrolysis will be a key component in the future of energy conversion and storage solutions. Solid oxide cells (SOC) can provide the most efficient electrolysis technology. Furthermore, solid oxide electrolysis cell (SOEC) technology has improved tremendously in terms of initial performance, long-term durability, and up-scaling of cell and stack sizes; and with respect to commissioning and operation of SOEC plants. Nevertheless, it is evident that further improvement of electrochemical performance and especially long-term durability is crucial for the coming up-scaling and integrated SOEC systems as part of our energy grid. This calls for research within 1) thorough mapping of degradation mechanisms, 2) determination of thresholds for the onset of irreversible degradation processes, 3) safe operating conditions and means to mitigate known degradation processes, etc. However, degradation is not a simple singular property and there is often a puzzling interplay between parameters such as: the chosen electro-catalysts, the manufacturing of cell components, obtained microstructures, the resulting initial electrochemical performance, operating conditions, and the nature and extent of different degradation processes. This work will present11This work has been presented as invited keynote speak at symposium E01.07 as part of the 23rd International Conference on Solid State Ionics, MRS Meeting, Boston (US), July 2022. an overview of the puzzling interplay between initial electrochemical performance and long-term durability. It will then provide examples of multiple complementary characterization techniques applied for the analyses of these features and processes (e.g., SEM, TEM, Raman spectroscopy and electrochemical impedance spectroscopy). Furthermore, strategies to mitigate and/or limit specific degradation processes and insight into data required to establish degradation wise safe operating conditions are discussed. The present work will touch upon multiple performance characteristics and degradation processes: 1) triple-phase-boundaries (TPB) and critical micro/nano-structural parameters of the porous nickel/yttria-stabilized-zirconia (Ni/YSZ) electrode; 2) the perception of electrode overpotential and why this parameter is more important than the set current density or operating cell/stack voltage, 3) the effect of impurities, 4) the interplay between impurities and carbon deposition and 5) the degradation caused by Ni migration and actions to mitigate this degradation process.

DLVR-NWP: A Novel Data-Driven Bearing Degradation Model for RUL Estimation

Performance degradation process modeling and remaining useful life (RUL) estimation have gained much attention for condition-based maintenance of important engineering equipment such as the bearings of gantry cranes. The existing work mainly uses the original or elaborately selected features that may not necessarily correspond to the current degradation process. In view of this, this paper proposes a novel data-driven bearing degradation modeling method, called dynamic latent variable reconstruction nonlinear Wiener process (DLVR-NWP). The proposed DLVR-NWP method is composed of a feature generation, a dynamic latent variable (DLV)-based nonlinear degradation detection, a dynamic latent variable reconstruction (DLVR)-based degradation-relevant dynamic feature extraction, and a failure magnitude-based nonlinear Wiener process (NWP) model for RUL estimation. The reduced-dimensional degradation-relevant dynamic feature is first extracted from multiple highly-correlated time-domain and frequency-domain features using a dynamic latent variable reconstruction method that finds out the magnitude feature along the historical similar degradation processes. By incorporating the DLV-based nonlinear degradation detection mechanism, degradation-relevant dynamic features during the degradation stage instead of the overall operation duration are integrated into the NWP model for RUL estimation. Results from case studies on real bearings indicate that the proposed method using degradation-relevant dynamic feature significantly improves the RUL estimation accuracy compared to the traditional methods.

A performance degradation model of nitrile butadiene rubber (NBR) O-rings for hydrogen permeation effects under high-pressure.

The integration of molecular chain changes on a microscopic scale to achieve macroscopic performance is crucial in degradation processes concerning O-ring seals. Nonetheless, a comprehensive and compelling mathematical model that can describe molecular chains' material properties and macroscopic material properties simultaneously for O-rings under high-pressure conditions is yet to be established. In this paper, we propose a degradation model based on viscoelasticity and molecular chain statistics for hydrogen permeation. The proposed model aims to establish the relationship between the material molecular chains and macroscopic material properties, with a primary focus on accurately recognizing the performance degradation process of rubber sealing rings. We verify the model's reliability through uniaxial tensile strength experiments and high-pressure hydrogen immersion experiments, respectively. Predictions of the model exhibit favorable conformity with the experimental data concerning the above phenomena. Furthermore, we derive the number of molecular chains and maximum strain of the degradation process. Based on the similarity of the degradation process's descent, it is plausible to speculate that NBR properties' degradation can be characterized by the average number of molecular chains.

Smart meter life prediction method based on health index and random forest

Smart electricity meter is the key instrument to realize the accurate measurement of electricity consumption. However, with the increase of the service life of electricity meter, its measurement accuracy will deteriorate. Aiming at the problems that a single parameter cannot accurately characterize the performance degradation process of the electric meter, and the traditional intelligent learning model is difficult to accurately fit the degradation model of the smart electric meter, a new method is proposed to fuse the data to construct the electric meter health index, and combine the multi-model similarity matching and the integrated model to calculate the residual energy of the electric meter. Methods of life prediction. Hierarchical clustering and silhouette coefficients are used to filter parameters and fuse them into a meter health index. Multi-model similarity matching is used for regression model prediction to optimize the prediction results of the model. Select a meter measurement data set to verify the effectiveness of the proposed method. The results show that the method has the advantages of fast prediction speed, and the fusion of health index and multi-model similarity matching greatly improves the prediction accuracy of the remaining life of the electric meter.

Digital twin based condition monitoring approach for rolling bearings

Digital twin is an important technology for grasping states of mechanical systems in real time. However, there are few studies on how to establish life-cycle digital twin models of bearings. In order to accurately estimate the condition of bearings, a digital twin model of bearing life cycle (BLDT) is proposed to achieve equivalent information on the virtual entity (VE) model and physical entity (PE) model. First, a dynamic model of rolling bearings and defect evolution model are established to simulate the dynamic response of the bearing performance degradation process. Then, the physical characteristics and degradation information of the PE model are exchanged with the VE model to evaluate the time-varying defect size and the equivalent comprehensive stiffness. The evolution law of the life-cycle is obtained through a neural network. Finally, the network parameters are introduced into the VE model to obtain dynamic response results of the life-cycle bearing dynamic model of other datasets under the same working conditions. By comparing the obtained digital twin results with experiment signals in the time and frequency domains, the accuracy and effectiveness of the BLDT model are verified.

Evaluation of Rolling Bearing Performance Degradation Based on Comprehensive Index Reduction and SVDD

The evaluation of rolling bearing performance degradation has important implications for the prediction and health management (PHM) of rotating equipment. A method for evaluation of rolling bearing performance degradation based on comprehensive index reduction and support vector data description (SVDD) is proposed in this study. Firstly, the improved variational mode decomposition (VMD) method was used to decompose vibration signals, and the defect frequency amplitude ratio index which is sensitive to early faults is extracted. Secondly, a comprehensive feature index set of rolling bearings is constructed by combining traditional time-domain and time–frequency-domain indexes, and the main features are extracted by the dimensionality reduction algorithm of locally linear embedding (LLE). Finally, the SVDD evaluation model was utilized to characterize and evaluate the rolling bearing lifetime degradation process using the distance from the test sample to the trained hypersphere center. Results showed that the proposed comprehensive degradation index can accurately detect the occurrence of early weak fault stage of rolling bearings and objectively reveal the performance degradation process of rolling bearings.

Neural-ILT 2.0: Migrating ILT to Domain-Specific and Multitask-Enabled Neural Network

Optical proximity correction (OPC) in modern design closures has become extremely expensive and challenging. Conventional model-based OPC encounters performance degradation and large process variation, while aggressive approach, such as inverse lithography technology (ILT), suffers from large computational overhead for both mask optimization and mask writing processes. In this article, we developed Neural-ILT, an end-to-end learning-based OPC framework, which literally conducts mask prediction and ILT correction for a given layout in a single neural network, with the objectives of: 1) mask printability enhancement; 2) mask complexity optimization; and 3) flow acceleration. A domain-specific model pretraining recipe, which introduces the domain knowledge of lithography system, is proposed to help Neural-ILT achieving faster and better convergence. Quantitative results show that compared to the state-of-the-art (SOTA) learning-based OPC solutions and conventional OPC flows, Neural-ILT can achieve <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$15\times $ </tex-math></inline-formula> to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$30\times $ </tex-math></inline-formula> turnaround time (TAT) speedup and the best mask printability with relatively lower mask complexity. Based on the developed infrastructure, we further investigated the feasibility of handling multiple mask optimization tasks for different datasets within a common Neural-ILT platform. We believe this work could bridge well-developed deep learning toolkits to GPU-based high-performance lithographic computations to achieve groundbreaking performance boosting on various computational lithography-related tasks.