Pub Date : 2023-12-18DOI: 10.36001/ijphm.2023.v14i2.3485
Hang Xiao, J. Coble, J. Hines
Accurately predicting the remaining useful life (RUL) of a system is a crucial factor in prognostics and health management (PHM). This paper introduces an auxiliary particle filter (APF) model, which has the advantages of dynamically updating the model parameters and being optimized in computational speed for prognosis applications in real engineering problems. The development of particle filter (PF) in the recent decade focused on increasing the PF model’s complexity to solve more difficult problems. However, the added complexity negatively impacts the computational speed. The number of particles is commonly reduced to compensate for this increased computational burden, but this significantly reduces the accuracy of PF’s posterior distribution. The developed APF model can estimate unknown states and model parameters at the same time with a large number of particles. This algorithm was demonstrated with a dataset from an electric motor accelerated aging experiment. The results show that this model can quickly and accurately predict the RUL and is robust to measurement noise.
{"title":"Auxiliary Particle Filter for Prognostics and Health Management","authors":"Hang Xiao, J. Coble, J. Hines","doi":"10.36001/ijphm.2023.v14i2.3485","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i2.3485","url":null,"abstract":"Accurately predicting the remaining useful life (RUL) of a system is a crucial factor in prognostics and health management (PHM). This paper introduces an auxiliary particle filter (APF) model, which has the advantages of dynamically updating the model parameters and being optimized in computational speed for prognosis applications in real engineering problems. The development of particle filter (PF) in the recent decade focused on increasing the PF model’s complexity to solve more difficult problems. However, the added complexity negatively impacts the computational speed. The number of particles is commonly reduced to compensate for this increased computational burden, but this significantly reduces the accuracy of PF’s posterior distribution. The developed APF model can estimate unknown states and model parameters at the same time with a large number of particles. This algorithm was demonstrated with a dataset from an electric motor accelerated aging experiment. The results show that this model can quickly and accurately predict the RUL and is robust to measurement noise.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":2.1,"publicationDate":"2023-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138995341","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-11-30DOI: 10.36001/ijphm.2023.v14i2.3545
P. Sharma K., V. T.
A high-end critical electronic system is expected to have hundreds of electronic subsystems, which rely on the Power Management Unit (PMU) to be energized. Having an efficient PMU is crucial and it requires reliable and well-structured voltage buck converters to translate the supplied voltage levels. The buck converters employed in PMU are expected to be fault tolerant and supply uninterrupted power while serving critical subsystems. Active redundant parallel buck converters employed in PMU to achieve fault tolerance increases overhead in terms of area, cost and power dissipation. In this paper, a DC-DC converter is designed for the PMU by combining two legs of buck converters with an effective output of 3.3 V. A simple yet effective technique is proposed to design a fault-tolerant buck DC-DC converter by bypassing a faulty converter leg. The proposed system utilizes an online signal processing-based method for prognostic fault detection. Ripple content in the voltage of the output Aluminum Electrolytic Capacitor (AEC) is monitored and used as a primary health indicator for the primary buck converter leg. Increase in the output ripple due to degradation is used for the prognosis of primary converter failure. The secondary buck converter leg is activated only upon the confirmed prognosis of a faulty primary converter leg to avoid false triggering. The timely prognosis of primary converter failure and activation of secondary converter facilitates uninterrupted power supply. An experimental setup is built and tested in the laboratory. Experimental results indicate a smooth transition from the primary converter leg to the secondary demonstrating an uninterrupted power supply along with the simplicity and effectiveness of the proposed solution
{"title":"Fault- Tolerant DC-DC Converter with Zero Interruption Time Using Capacitor Health Prognosis","authors":"P. Sharma K., V. T.","doi":"10.36001/ijphm.2023.v14i2.3545","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i2.3545","url":null,"abstract":"A high-end critical electronic system is expected to have hundreds of electronic subsystems, which rely on the Power Management Unit (PMU) to be energized. Having an efficient PMU is crucial and it requires reliable and well-structured voltage buck converters to translate the supplied voltage levels. The buck converters employed in PMU are expected to be fault tolerant and supply uninterrupted power while serving critical subsystems. Active redundant parallel buck converters employed in PMU to achieve fault tolerance increases overhead in terms of area, cost and power dissipation. In this paper, a DC-DC converter is designed for the PMU by combining two legs of buck converters with an effective output of 3.3 V. A simple yet effective technique is proposed to design a fault-tolerant buck DC-DC converter by bypassing a faulty converter leg. The proposed system utilizes an online signal processing-based method for prognostic fault detection. Ripple content in the voltage of the output Aluminum Electrolytic Capacitor (AEC) is monitored and used as a primary health indicator for the primary buck converter leg. Increase in the output ripple due to degradation is used for the prognosis of primary converter failure. The secondary buck converter leg is activated only upon the confirmed prognosis of a faulty primary converter leg to avoid false triggering. The timely prognosis of primary converter failure and activation of secondary converter facilitates uninterrupted power supply. An experimental setup is built and tested in the laboratory. Experimental results indicate a smooth transition from the primary converter leg to the secondary demonstrating an uninterrupted power supply along with the simplicity and effectiveness of the proposed solution","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":2.1,"publicationDate":"2023-11-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139198997","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-11-10DOI: 10.36001/ijphm.2023.v14i2.3528
Junhyun Byun, Suhong Min, Jihoon Kang
With the rising complexity of manufacturing processes, resulting from rapid industrial development, the utilization of remaining useful lifecycle (RUL) prediction, based on failure physics and traditional reliability, has remained limited. Although data-driven approaches of RUL prediction were developed using machine learning algorithms, uncertainty-induced challenges have emerged, such as sensor noise and modeling error. To address these uncertainty-induced problems, this study proposes a stochastic ensemble-modeling concept for improving the RUL prediction result. The proposed ensemble model combines artificial degradation patterns and fitness weights, which incorporate formulas reflecting failure patterns and various reliability function data with the observed degradation factor. Furthermore, a recursive Bayesian updating technique, reflecting the difference between expected and observed remaining life sequentially, was leveraged to reduce the prediction uncertainty. Moreover, we comparatively studied the predictive performance of the proposed model (recursive Bayesian ensemble model) against an existing baseline method (exponentially weighted linear regression model). Through simulation and case datasets, this experiment demonstrated the robustness and utility of the proposed algorithm.
{"title":"RUL Prognostics","authors":"Junhyun Byun, Suhong Min, Jihoon Kang","doi":"10.36001/ijphm.2023.v14i2.3528","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i2.3528","url":null,"abstract":"With the rising complexity of manufacturing processes, resulting from rapid industrial development, the utilization of remaining useful lifecycle (RUL) prediction, based on failure physics and traditional reliability, has remained limited. Although data-driven approaches of RUL prediction were developed using machine learning algorithms, uncertainty-induced challenges have emerged, such as sensor noise and modeling error. To address these uncertainty-induced problems, this study proposes a stochastic ensemble-modeling concept for improving the RUL prediction result. The proposed ensemble model combines artificial degradation patterns and fitness weights, which incorporate formulas reflecting failure patterns and various reliability function data with the observed degradation factor. Furthermore, a recursive Bayesian updating technique, reflecting the difference between expected and observed remaining life sequentially, was leveraged to reduce the prediction uncertainty. Moreover, we comparatively studied the predictive performance of the proposed model (recursive Bayesian ensemble model) against an existing baseline method (exponentially weighted linear regression model). Through simulation and case datasets, this experiment demonstrated the robustness and utility of the proposed algorithm.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135092601","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-10-17DOI: 10.36001/ijphm.2023.v14i2.3583
Wei Li, Guoyan Li, Sagar Kamarthi
The increasing complexity of a vehicle's digital architecture has created new opportunities to revolutionize the maintenance paradigm. The Artificial Intelligence (AI) assisted maintenance system is a promising solution to enhance efficiency and reduce costs. This review paper studies the research trends in AI-assisted vehicle maintenance via keyword co-occurrence network (KCN) analysis. The KCN methodology is applied to systematically analyze the keywords extracted from 3153 peer-reviewed papers published between 2011 and 2022. The network metrics and trend analysis uncovered important knowledge components and structure of the research field covering AI applications for vehicle maintenance. The emerging and declining research trends in AI models and vehicle maintenance application scenarios were identified through trend visualizations. In summary, this review paper provides a comprehensive high-level overview of AI-assisted vehicle maintenance. It serves as a valuable resource for researchers and practitioners in the automotive industry. This paper also highlights potential research opportunities, limitations, and challenges related to AI-assisted vehicle maintenance.
{"title":"The Study of Trends in AI Applications for Vehicle Maintenance Through Keyword Co-occurrence Network Analysis","authors":"Wei Li, Guoyan Li, Sagar Kamarthi","doi":"10.36001/ijphm.2023.v14i2.3583","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i2.3583","url":null,"abstract":"The increasing complexity of a vehicle's digital architecture has created new opportunities to revolutionize the maintenance paradigm. The Artificial Intelligence (AI) assisted maintenance system is a promising solution to enhance efficiency and reduce costs. This review paper studies the research trends in AI-assisted vehicle maintenance via keyword co-occurrence network (KCN) analysis. The KCN methodology is applied to systematically analyze the keywords extracted from 3153 peer-reviewed papers published between 2011 and 2022. The network metrics and trend analysis uncovered important knowledge components and structure of the research field covering AI applications for vehicle maintenance. The emerging and declining research trends in AI models and vehicle maintenance application scenarios were identified through trend visualizations. In summary, this review paper provides a comprehensive high-level overview of AI-assisted vehicle maintenance. It serves as a valuable resource for researchers and practitioners in the automotive industry. This paper also highlights potential research opportunities, limitations, and challenges related to AI-assisted vehicle maintenance.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136034602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-24DOI: 10.36001/ijphm.2023.v14i2.3497
M. Soualhi, A. Soualhi, K. Nguyen, K. Medjaher, C. Guy, Razik Hubert
Rotating machines are widely used in several fields such as railways, renewable energies, robotics, etc. This diversity of application implies a large variety of faults of critical components susceptible to fail. For this purpose, prognostics and health management (PHM) is deployed to effectively monitor these components through the detection, diagnostics as well as prognostics of faults. In the literature, there exist numerous methods to ensure the above monitoring activities. However, few of them consider different failure types using heterogeneous data and various operating conditions. Also, there are no dominant methods that can be generalized for monitoring. For this reason, the genericity of these methods and their applicability in several systems is a crucial issue. To help researchers to achieve the above challenges, this paper presents a detailed description of data sources from experimental test benches. These data-sets correspond to different case studies that monitor the health states of multiple critical components in various operating conditions using numerous sensors.
{"title":"Open Heterogeneous Data for Condition Monitoring of Multi Faults in Rotating Machines Used in Different Operating Conditions","authors":"M. Soualhi, A. Soualhi, K. Nguyen, K. Medjaher, C. Guy, Razik Hubert","doi":"10.36001/ijphm.2023.v14i2.3497","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i2.3497","url":null,"abstract":"Rotating machines are widely used in several fields such as railways, renewable energies, robotics, etc. This diversity of application implies a large variety of faults of critical components susceptible to fail. For this purpose, prognostics and health management (PHM) is deployed to effectively monitor these components through the detection, diagnostics as well as prognostics of faults. In the literature, there exist numerous methods to ensure the above monitoring activities. However, few of them consider different failure types using heterogeneous data and various operating conditions. Also, there are no dominant methods that can be generalized for monitoring. For this reason, the genericity of these methods and their applicability in several systems is a crucial issue. To help researchers to achieve the above challenges, this paper presents a detailed description of data sources from experimental test benches. These data-sets correspond to different case studies that monitor the health states of multiple critical components in various operating conditions using numerous sensors.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":2.1,"publicationDate":"2023-08-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49011037","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-17DOI: 10.36001/ijphm.2023.v14i2.3533
Xuemei Yuan, Fei-long Liu, Yong-jun Qie, Shuai Sun, Jie Ren
Equipment reliability is the key feature to ensure the equipment operation for a long time. It is difficult to determine the overall reliability of industrial equipment due to the different reliability states of different subsystems. A device abnormality identification method based on JS (Jenson's Shannon) divergence and a health status assessment technology based on FMECA (failure mode, effect and criticality analysis) are proposed. This method enables an accurate assessment of the current health status of the device. First, the historical operation data is preprocessed according to the characteristics of the equipment to improve the data quality. The JS divergence method is reused to extract the similarity between the key feature data distribution and the benchmark data distribution. Then, the FMECA report is established using the real running data of the device combined with expert experience. Gray theory was used to determine the degree of association between one-way health state membership vector and different health state rank vector. Finally, the health status level was comprehensively evaluated by the fuzzy membership method. Taking the mechanical arm component of a 100-ton crane as an example, the results show that this method can effectively evaluate the current health state of the equipment, and provide power for the abnormal advance disposal and auxiliary management decisions.
{"title":"Device Health Status Assessment Under the Influence of Multiple Exception Modes","authors":"Xuemei Yuan, Fei-long Liu, Yong-jun Qie, Shuai Sun, Jie Ren","doi":"10.36001/ijphm.2023.v14i2.3533","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i2.3533","url":null,"abstract":"Equipment reliability is the key feature to ensure the equipment operation for a long time. It is difficult to determine the overall reliability of industrial equipment due to the different reliability states of different subsystems. A device abnormality identification method based on JS (Jenson's Shannon) divergence and a health status assessment technology based on FMECA (failure mode, effect and criticality analysis) are proposed. This method enables an accurate assessment of the current health status of the device. First, the historical operation data is preprocessed according to the characteristics of the equipment to improve the data quality. The JS divergence method is reused to extract the similarity between the key feature data distribution and the benchmark data distribution. Then, the FMECA report is established using the real running data of the device combined with expert experience. Gray theory was used to determine the degree of association between one-way health state membership vector and different health state rank vector. Finally, the health status level was comprehensively evaluated by the fuzzy membership method. Taking the mechanical arm component of a 100-ton crane as an example, the results show that this method can effectively evaluate the current health state of the equipment, and provide power for the abnormal advance disposal and auxiliary management decisions.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":2.1,"publicationDate":"2023-08-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41544446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-08-08DOI: 10.36001/ijphm.2023.v14i2.3486
Joseph Cohen, Xun Huan, Jun Ni
In the era of industrial big data, prognostics and health management is essential to improve the prediction of future failures to minimize inventory, maintenance, and human costs. Used for the 2021 PHM Data Challenge, the new Commercial Modular Aero-Propulsion System Simulation dataset from NASA is an open-source benchmark containing simulated turbofan engine units flown under realistic flight conditions. Deep learning approaches implemented previously for this application attempt to predict the remaining useful life of the engine units, but have not utilized labeled failure mode information, impeding practical usage and explainability. To address these limitations, a new prognostics approach is formulated with a customized loss function to simultaneously predict the current health state, the eventual failing component(s), and the remaining useful life. The proposed method incorporates principal component analysis to orthogonalize statistical time-domain features, which are inputs into supervised regressors such as random forests, extreme random forests, XGBoost, and artificial neural networks. The highest performing algorithm, ANN–Flux with PCA augmentation, achieves AUROC and AUPR scores exceeding 0.94 for each classification on average. In addition to predicting eventual failures with high accuracy, ANN–Flux achieves comparable remaining useful life RMSE for the same test split of the dataset when benchmarked against past work, with significantly less computational cost.
{"title":"Fault Prognosis of Turbofan Engines","authors":"Joseph Cohen, Xun Huan, Jun Ni","doi":"10.36001/ijphm.2023.v14i2.3486","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i2.3486","url":null,"abstract":"In the era of industrial big data, prognostics and health management is essential to improve the prediction of future failures to minimize inventory, maintenance, and human costs. Used for the 2021 PHM Data Challenge, the new Commercial Modular Aero-Propulsion System Simulation dataset from NASA is an open-source benchmark containing simulated turbofan engine units flown under realistic flight conditions. Deep learning approaches implemented previously for this application attempt to predict the remaining useful life of the engine units, but have not utilized labeled failure mode information, impeding practical usage and explainability. To address these limitations, a new prognostics approach is formulated with a customized loss function to simultaneously predict the current health state, the eventual failing component(s), and the remaining useful life. The proposed method incorporates principal component analysis to orthogonalize statistical time-domain features, which are inputs into supervised regressors such as random forests, extreme random forests, XGBoost, and artificial neural networks. The highest performing algorithm, ANN–Flux with PCA augmentation, achieves AUROC and AUPR scores exceeding 0.94 for each classification on average. In addition to predicting eventual failures with high accuracy, ANN–Flux achieves comparable remaining useful life RMSE for the same test split of the dataset when benchmarked against past work, with significantly less computational cost.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-08-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135793577","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-23DOI: 10.36001/ijphm.2023.v14i2.3425
Yong Chae Kim, Taehun Kim, J. U. Ko, Jinwook Lee, Keon Kim
Data-driven fault diagnosis is an essential technology for the safety and maintenance of rock drills. However, since the signals acquired from a rock drill have different distributions, which arise due to their variable operating conditions, the classification performance of any data-driven method is diminished; this is called the domain-shift issue. This paper proposes a new domain-adaptation-based fault diagnosis scheme to solve the domain-shift problem. The proposed method introduces a data-cropping technique to mitigate the difference in the length of the data measured from a rock drill for each impact cycle. To extract invariant features for all operating conditions, the proposed method combines two methods: a domain adversarial neural network and minimization of the maximum mean discrepancy (MMD) between the features from different domains. In addition, a soft voting ensemble is used to reduce the model uncertainty. The proposed method shows superior performance when validated with a rock drill dataset; the proposed approach was ranked in 2nd place in the 2022 PHM Conference Data Challenge.
{"title":"Domain Adaptation based Fault Diagnosis under Variable Operating Conditions of a Rock Drill","authors":"Yong Chae Kim, Taehun Kim, J. U. Ko, Jinwook Lee, Keon Kim","doi":"10.36001/ijphm.2023.v14i2.3425","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i2.3425","url":null,"abstract":"Data-driven fault diagnosis is an essential technology for the safety and maintenance of rock drills. However, since the signals acquired from a rock drill have different distributions, which arise due to their variable operating conditions, the classification performance of any data-driven method is diminished; this is called the domain-shift issue. This paper proposes a new domain-adaptation-based fault diagnosis scheme to solve the domain-shift problem. The proposed method introduces a data-cropping technique to mitigate the difference in the length of the data measured from a rock drill for each impact cycle. To extract invariant features for all operating conditions, the proposed method combines two methods: a domain adversarial neural network and minimization of the maximum mean discrepancy (MMD) between the features from different domains. In addition, a soft voting ensemble is used to reduce the model uncertainty. The proposed method shows superior performance when validated with a rock drill dataset; the proposed approach was ranked in 2nd place in the 2022 PHM Conference Data Challenge.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":2.1,"publicationDate":"2023-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46785169","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-07-10DOI: 10.36001/ijphm.2023.v14i1.3458
Hye Jun Oh, Jinoh Yoo, Sangkyung Lee, Minseok Chae, Jongmin Park, B. Youn
This study presents a novel method for fault diagnosis of a hydrostatic rock drill. Hydraulic rock drills suffer from both domain discrepancy issues that arise due to their harsh working environment and indivisible difference. As a result, fault diagnosis is very challenging. To overcome these problems, we propose a novel diagnosis method that combines both data-driven and signal-process-based methods. In the proposed approach, data-driven methods are employed for overall fault classification, using domain adaptation, metric learning, and pseudo-label-based deep learning methods. Next, a signal-process-based method is used to diagnose the specific fault by generating a reference signal. Using the combined approach, the fault-diagnosis performance was 100%; the proposed method was able to perform well even in cases with domain discrepancy.
{"title":"A Hybrid Approach Combining Data-Driven and Signal-Processing-Based Methods for Fault Diagnosis of a Hydraulic Rock Drill","authors":"Hye Jun Oh, Jinoh Yoo, Sangkyung Lee, Minseok Chae, Jongmin Park, B. Youn","doi":"10.36001/ijphm.2023.v14i1.3458","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i1.3458","url":null,"abstract":"This study presents a novel method for fault diagnosis of a hydrostatic rock drill. Hydraulic rock drills suffer from both domain discrepancy issues that arise due to their harsh working environment and indivisible difference. As a result, fault diagnosis is very challenging. To overcome these problems, we propose a novel diagnosis method that combines both data-driven and signal-process-based methods. In the proposed approach, data-driven methods are employed for overall fault classification, using domain adaptation, metric learning, and pseudo-label-based deep learning methods. Next, a signal-process-based method is used to diagnose the specific fault by generating a reference signal. Using the combined approach, the fault-diagnosis performance was 100%; the proposed method was able to perform well even in cases with domain discrepancy.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":2.1,"publicationDate":"2023-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49667590","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-06-11DOI: 10.36001/ijphm.2023.v14i1.3419
M. Radaideh, C. Pappas, M. Wezensky, P. Ramuhalli, Sarah Cousineau
Early fault detection and fault prognosis are crucial to ensure efficient and safe operations of complex engineering systems such as the Spallation Neutron Source (SNS) and its power electronics (high voltage converter modulators). Following an advanced experimental facility setup that mimics SNS operating conditions, the authors successfully conducted 21 early fault detection experiments, where fault precursors are introduced in the system to a degree enough to cause degradation in the waveform signals, but not enough to reach a real fault. Nine different machine learning techniques based on ensemble trees, convolutional neural networks, support vector machines, and hierarchical voting ensembles are proposed to detect the fault precursors. Although all 9 models have shown a perfect and identical performance during the training and testing phase, the performance of most models has decreased in the next test phase once they got exposed to realworld data from the 21 experiments. The hierarchical voting ensemble, which features multiple layers of diverse models, maintains a distinguished performance in early detection of the fault precursors with 95% success rate (20/21 tests), followed by adaboost and extremely randomized trees with 52% and 48% success rates, respectively. The support vector machine models were the worst with only 24% success rate (5/21 tests). The study concluded that a successful implementation of machine learning in the SNS or particle accelerator power systems would require a major upgrade in the controller and the data acquisition system to facilitate streaming and handling big data for the machine learning models. In addition, this study shows that the best performing models were diverse and based on the ensemble concept to reduce the bias and hyperparameter sensitivity of individual models.
{"title":"Early Fault Detection in Particle Accelerator Power Electronics Using Ensemble Learning","authors":"M. Radaideh, C. Pappas, M. Wezensky, P. Ramuhalli, Sarah Cousineau","doi":"10.36001/ijphm.2023.v14i1.3419","DOIUrl":"https://doi.org/10.36001/ijphm.2023.v14i1.3419","url":null,"abstract":"Early fault detection and fault prognosis are crucial to ensure efficient and safe operations of complex engineering systems such as the Spallation Neutron Source (SNS) and its power electronics (high voltage converter modulators). Following an advanced experimental facility setup that mimics SNS operating conditions, the authors successfully conducted 21 early fault detection experiments, where fault precursors are introduced in the system to a degree enough to cause degradation in the waveform signals, but not enough to reach a real fault. Nine different machine learning techniques based on ensemble trees, convolutional neural networks, support vector machines, and hierarchical voting ensembles are proposed to detect the fault precursors. Although all 9 models have shown a perfect and identical performance during the training and testing phase, the performance of most models has decreased in the next test phase once they got exposed to realworld data from the 21 experiments. The hierarchical voting ensemble, which features multiple layers of diverse models, maintains a distinguished performance in early detection of the fault precursors with 95% success rate (20/21 tests), followed by adaboost and extremely randomized trees with 52% and 48% success rates, respectively. The support vector machine models were the worst with only 24% success rate (5/21 tests). The study concluded that a successful implementation of machine learning in the SNS or particle accelerator power systems would require a major upgrade in the controller and the data acquisition system to facilitate streaming and handling big data for the machine learning models. In addition, this study shows that the best performing models were diverse and based on the ensemble concept to reduce the bias and hyperparameter sensitivity of individual models.","PeriodicalId":42100,"journal":{"name":"International Journal of Prognostics and Health Management","volume":null,"pages":null},"PeriodicalIF":2.1,"publicationDate":"2023-06-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46016459","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：