Pub Date : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3322
Eli Gildish, M. Grebshtein, Y. Aperstein, Alex Kushnirski, Igor Makienko
The existing helicopter Health and Usage Management Systems (HUMS) collect and process flight operational parameters and sensors data such as vibrations to provide health monitoring of the helicopter dynamic assemblies and engines. So far, structure-related mechanical faults, such as looseness in bolted structures, have not been addressed by vibration-based condition monitoring in existing HUMS systems. Bolt loosening was identified as a potential risk to flight safety demanding periodical visual monitoring, and increased maintenance and repair expenses. Its automatic identification in helicopters by using vibration measurements is challenging due to the limited number of known events and the presence of high-energy vibrations originating in rotating parts, which shadow the low-level signals generated by the bolt loosening. New developed bolt loosening monitoring approach was tested on HUMS vibrations data recorded from the IAF AH-64 Apache helicopters fleet. ML-based unsupervised anomaly detection was utilized in order to address the limited number of faulty cases. The predictive power of health features was significantly improved by applying the Harmonic filtering differentiating between the high-energy vibrations generated by rotating parts compared with the low-energy structural vibrations. Different unsupervised anomaly detection techniques were examined on the dataset. The experimental results demonstrate that the developed approach enable successful bolt loosening monitoring in helicopters and can potentially be used in other health monitoring applications.
{"title":"Helicopter Bolt Loosening Monitoring using Vibrations and Machine Learning","authors":"Eli Gildish, M. Grebshtein, Y. Aperstein, Alex Kushnirski, Igor Makienko","doi":"10.36001/phme.2022.v7i1.3322","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3322","url":null,"abstract":"The existing helicopter Health and Usage Management Systems (HUMS) collect and process flight operational parameters and sensors data such as vibrations to provide health monitoring of the helicopter dynamic assemblies and engines. So far, structure-related mechanical faults, such as looseness in bolted structures, have not been addressed by vibration-based condition monitoring in existing HUMS systems. Bolt loosening was identified as a potential risk to flight safety demanding periodical visual monitoring, and increased maintenance and repair expenses. Its automatic identification in helicopters by using vibration measurements is challenging due to the limited number of known events and the presence of high-energy vibrations originating in rotating parts, which shadow the low-level signals generated by the bolt loosening. \u0000New developed bolt loosening monitoring approach was tested on HUMS vibrations data recorded from the IAF AH-64 Apache helicopters fleet. ML-based unsupervised anomaly detection was utilized in order to address the limited number of faulty cases. The predictive power of health features was significantly improved by applying the Harmonic filtering differentiating between the high-energy vibrations generated by rotating parts compared with the low-energy structural vibrations. Different unsupervised anomaly detection techniques were examined on the dataset. The experimental results demonstrate that the developed approach enable successful bolt loosening monitoring in helicopters and can potentially be used in other health monitoring applications.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133625868","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3319
Van-Thai Nguyen, P. Do, A. Voisin, B. Iung
It is well-known that maintenance decision optimization for multi-component systems faces the curse of dimensionality. Specifically, the number of decision variables needed to be optimized grows exponentially in the number of components causing computational expensive for optimization algorithms. To address this issue, we customize a multi-agent deep reinforcement learning algorithm, namely Weighted QMIX, in the case where system states can be fully observed to obtain cost-effective policies. A case study is conducted on a 13- component system to examine the effectiveness of the customized algorithm. The obtained results confirmed its performance.
{"title":"Weighted-QMIX-based Optimization for Maintenance Decision-making of Multi-component Systems","authors":"Van-Thai Nguyen, P. Do, A. Voisin, B. Iung","doi":"10.36001/phme.2022.v7i1.3319","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3319","url":null,"abstract":"It is well-known that maintenance decision optimization for multi-component systems faces the curse of dimensionality. Specifically, the number of decision variables needed to be optimized grows exponentially in the number of components causing computational expensive for optimization algorithms. To address this issue, we customize a multi-agent deep reinforcement learning algorithm, namely Weighted QMIX, in the case where system states can be fully observed to obtain cost-effective policies. A case study is conducted on a 13- component system to examine the effectiveness of the customized algorithm. The obtained results confirmed its performance.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121214514","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3336
Omnia Amin, Blair Brown, B. Stephen, S. Mcarthur
Civil nuclear generation plant must maximise it’s operational uptime in order to maintain it’s viability. With aging plant and heavily regulated operating constraints, monitoring is commonplace, but identifying health indicators to pre-empt disruptive faults is challenging owing to the volumes of data involved. Machine learning (ML) models are increasingly deployed in prognostics and health management (PHM) systems in various industrial applications, however, many of these are black box models that provide good performance but little or no insight into how predictions are reached. In nuclear generation, there is significant regulatory oversight and therefore a necessity to explain decisions based on outputs from predictive models. These explanations can then enable stakeholders to trust these outputs, satisfy regulatory bodies and subsequently make more effective operational decisions. How ML model outputs convey explanations to stakeholders is important, so these explanations must be in human (and technical domain related) understandable terms. Consequently, stakeholders can rapidly interpret, then trust predictions better, and will be able to act on them more effectively. The main contributions of this paper are: 1. introduce XAI into the PHM of industrial assets and provide a novel set of algorithms that translate the explanations produced by SHAP to text-based human-interpretable explanations; and 2. consider the context of these explanations as intended for application to prognostics of critical assets in industrial applications. The use of XAI will not only help in understanding how these ML models work, but also describe the most important features contributing to predicted degradation of the nuclear generation asset.
{"title":"Case-study Led Investigation of Explainable AI (XAI) to Support Deployment of Prognostics in the industry","authors":"Omnia Amin, Blair Brown, B. Stephen, S. Mcarthur","doi":"10.36001/phme.2022.v7i1.3336","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3336","url":null,"abstract":"Civil nuclear generation plant must maximise it’s operational uptime in order to maintain it’s viability. With aging plant and heavily regulated operating constraints, monitoring is commonplace, but identifying health indicators to pre-empt disruptive faults is challenging owing to the volumes of data involved. Machine learning (ML) models are increasingly deployed in prognostics and health management (PHM) systems in various industrial applications, however, many of these are black box models that provide good performance but little or no insight into how predictions are reached. In nuclear generation, there is significant regulatory oversight and therefore a necessity to explain decisions based on outputs from predictive models. These explanations can then enable stakeholders to trust these outputs, satisfy regulatory bodies and subsequently make more effective operational decisions. How ML model outputs convey explanations to stakeholders is important, so these explanations must be in human (and technical domain related) understandable terms. Consequently, stakeholders can rapidly interpret, then trust predictions better, and will be able to act on them more effectively. The main contributions of this paper are: 1. introduce XAI into the PHM of industrial assets and provide a novel set of algorithms that translate the explanations produced by SHAP to text-based human-interpretable explanations; and 2. consider the context of these explanations as intended for application to prognostics of critical assets in industrial applications. The use of XAI will not only help in understanding how these ML models work, but also describe the most important features contributing to predicted degradation of the nuclear generation asset.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126632469","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3330
P. Korkos, J. Kleemola, M. Linjama, A. Lehtovaara
A wind turbine is equipped with lots of sensors whose measurements are recorded by the supervisory control and data acquisition (SCADA) system and stored every 10 minutes. The pitch subsystem of a wind turbine is of critical importance as it presents the highest failure rate. Thus, selecting the most essential features from the SCADA system is performed in order to detect faults efficiently. In this study, a feature space of 49 features is available, referring to the condition of a hydraulic pitch system. The dimensionality of this feature space (original input space) is reduced using a Deep Autoencoder in order to extract latent information. The architecture of the Autoencoder is investigated regarding its efficiency on fault detection task. This way, effect of new extracted features on the performance of the classifier is presented. A Support Vector Machine (SVM) classifier is trained using a set of healthy (fault free) and faulty data, representing different kind of pitch system failures. The data are acquired from a wind farm of five 2.3MW fixed-speed wind turbines. The performance metric used to evaluate their effect on data is F1-score. Results show that SVM using new extracted feature by Autoencoder outperforms SVM classifier using the original feature set, underlining the power of Autoencoders to unveil latent information.
{"title":"Fault Detection in a Wind Turbine Hydraulic Pitch System Using Deep Autoencoder Extracted Features","authors":"P. Korkos, J. Kleemola, M. Linjama, A. Lehtovaara","doi":"10.36001/phme.2022.v7i1.3330","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3330","url":null,"abstract":"A wind turbine is equipped with lots of sensors whose measurements are recorded by the supervisory control and data acquisition (SCADA) system and stored every 10 minutes. The pitch subsystem of a wind turbine is of critical importance as it presents the highest failure rate. Thus, selecting the most essential features from the SCADA system is performed in order to detect faults efficiently. In this study, a feature space of 49 features is available, referring to the condition of a hydraulic pitch system. The dimensionality of this feature space (original input space) is reduced using a Deep Autoencoder in order to extract latent information. The architecture of the Autoencoder is investigated regarding its efficiency on fault detection task. This way, effect of new extracted features on the performance of the classifier is presented. A Support Vector Machine (SVM) classifier is trained using a set of healthy (fault free) and faulty data, representing different kind of pitch system failures. The data are acquired from a wind farm of five 2.3MW fixed-speed wind turbines. The performance metric used to evaluate their effect on data is F1-score. Results show that SVM using new extracted feature by Autoencoder outperforms SVM classifier using the original feature set, underlining the power of Autoencoders to unveil latent information.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129896310","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3302
Maximilian-Peter Radtke, Jürgen Bock
In the recent past deep learning approaches have achieved remarkable results in the area of Prognostics and Health Management (PHM). These algorithms rely on large amounts of data, which is often not available, and produce outputs, which are hard to interpret. Before the broad success of deep learning machine faults were often classified using domain expert knowledge based on experience and physical models. In comparison, these approaches only require small amounts of data and produce highly interpretable results. On the downside, however, they struggle to predict unexpected patterns hidden in data. This research aims to combine knowledge and deep learning to increase accuracy, robustness and interpretability of current models.
{"title":"Combining Knowledge and Deep Learning for Prognostics and Health Management","authors":"Maximilian-Peter Radtke, Jürgen Bock","doi":"10.36001/phme.2022.v7i1.3302","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3302","url":null,"abstract":"In the recent past deep learning approaches have achieved remarkable results in the area of Prognostics and Health Management (PHM). These algorithms rely on large amounts of data, which is often not available, and produce outputs, which are hard to interpret. Before the broad success of deep learning machine faults were often classified using domain expert knowledge based on experience and physical models. In comparison, these approaches only require small amounts of data and produce highly interpretable results. On the downside, however, they struggle to predict unexpected patterns hidden in data. This research aims to combine knowledge and deep learning to increase accuracy, robustness and interpretability of current models.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130101481","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3351
Rob Salaets, Valentin Sturm, T. Ooijevaar, V. Putz, Julia Mayer, A. Bey-Temsamani
Cutting tool wear needs to be monitored closely to ensure good quality of machined parts. However, manual inspection is both expensive and time consuming, therefore there is a need for automated monitoring methods. We present a technique that can reconstruct the cutting tool surface in 3D, allowing a spatial estimation of the tool wear with high accuracy. The reconstruction allows an automated direct monitoring method that estimates at any time the cutting tool condition, avoiding conversion work and major quality issues. The optical measurement setup consists of a hardware triggered line scan camera that registers the spinning cutting tool’s shadow inflicted by a collimated backlight. We show how to leverage the 1D line scan signal acquired at varying cutting heights of the tool into a full 3D reconstruction. The progression of tool wear may thus be monitored by comparing the reconstructed shape to previous measurements. To this end we show a methodology for tool wear quantification. Additionally, to assess the measurement technique, an accuracy analysis with ground truth geometry was performed. The technique was applied to multiple degrading drilling tools. By automation of the cutting tool health monitoring, retrofitting this technology on a conventional machining center would transform it into an Industry 4.0 compatible (smart) machining center utilizing off-the-shelf optical equipment with moderate costs.
{"title":"Optical Cutting Tool Wear Monitoring by 3D Geometry Reconstruction","authors":"Rob Salaets, Valentin Sturm, T. Ooijevaar, V. Putz, Julia Mayer, A. Bey-Temsamani","doi":"10.36001/phme.2022.v7i1.3351","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3351","url":null,"abstract":"Cutting tool wear needs to be monitored closely to ensure good quality of machined parts. However, manual inspection is both expensive and time consuming, therefore there is a need for automated monitoring methods. We present a technique that can reconstruct the cutting tool surface in 3D, allowing a spatial estimation of the tool wear with high accuracy. The reconstruction allows an automated direct monitoring method that estimates at any time the cutting tool condition, avoiding conversion work and major quality issues. The optical measurement setup consists of a hardware triggered line scan camera that registers the spinning cutting tool’s shadow inflicted by a collimated backlight. We show how to leverage the 1D line scan signal acquired at varying cutting heights of the tool into a full 3D reconstruction. The progression of tool wear may thus be monitored by comparing the reconstructed shape to previous measurements. To this end we show a methodology for tool wear quantification. Additionally, to assess the measurement technique, an accuracy analysis with ground truth geometry was performed. The technique was applied to multiple degrading drilling tools. By automation of the cutting tool health monitoring, retrofitting this technology on a conventional machining center would transform it into an Industry 4.0 compatible (smart) machining center utilizing off-the-shelf optical equipment with moderate costs.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"193 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122107878","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3345
Gautam Kumar Vadisala, A. Rawat, Abhishek Dubey, Gareth Yen Ket Chin, Fabio Abreu
With an increased collection of data, assessing the health of an asset and designing recommendations or executing response actions via prognostics and health management (PHM) has made great advances. These actions can be corrective or preventive depending upon the risk of failure or the cost of repair. As downhole testing tools operate in extreme environments, they are subjected to conditions like elevated temperature, shocks, vibrations, and pressures. The dump mandrels used in the process are prone to wear and tear like scratches, pits, and corrosion, which may cause operational failure. If these damages and their degree goes undetected and no remedial actions are taken, possibilities of non-productive time (NPT) and financial losses increase drastically. This paper aims to develop a fitness inspector which uses Computer Vision and Deep Learning to identify critical surfaces of these tools and the damage within them. This will help the Subject Matter Experts (SMEs) by replacing the qualified workforce provided by them and reducing the time consumed to gauge the health status of all the tools as the diagnosis can be made in real-time.
{"title":"Automating Critical Surface Identification and Damage Detection Using Deep Learning and Perspective Projection Methods","authors":"Gautam Kumar Vadisala, A. Rawat, Abhishek Dubey, Gareth Yen Ket Chin, Fabio Abreu","doi":"10.36001/phme.2022.v7i1.3345","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3345","url":null,"abstract":"With an increased collection of data, assessing the health of an asset and designing recommendations or executing response actions via prognostics and health management (PHM) has made great advances. These actions can be corrective or preventive depending upon the risk of failure or the cost of repair. As downhole testing tools operate in extreme environments, they are subjected to conditions like elevated temperature, shocks, vibrations, and pressures. The dump mandrels used in the process are prone to wear and tear like scratches, pits, and corrosion, which may cause operational failure. If these damages and their degree goes undetected and no remedial actions are taken, possibilities of non-productive time (NPT) and financial losses increase drastically. This paper aims to develop a fitness inspector which uses Computer Vision and Deep Learning to identify critical surfaces of these tools and the damage within them. This will help the Subject Matter Experts (SMEs) by replacing the qualified workforce provided by them and reducing the time consumed to gauge the health status of all the tools as the diagnosis can be made in real-time.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"221 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122526085","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3317
M. Kefalas, Bas van Stein, Mitra Baratchi, A. Apostolidis, T. Baeck
Estimating the remaining useful life (RUL) of an asset lies at the heart of prognostics and health management (PHM) of many operations-critical industries such as aviation. Modern methods of RUL estimation adopt techniques from deep learning (DL). However, most of these contemporary techniques deliver only single-point estimates for the RUL without reporting on the confidence of the prediction. This practice usually provides overly confident predictions that can have severe consequences in operational disruptions or even safety. To address this issue, we propose a technique for uncertainty quantification (UQ) based on Bayesian deep learning (BDL). The hyperparameters of the framework are tuned using a novel bi-objective Bayesian optimization method with objectives the predictive performance and predictive uncertainty. The method also integrates the data pre-processing steps into the hyperparameter optimization (HPO) stage, models the RUL as a Weibull distribution, and returns the survival curves of the monitored assets to allow informed decision-making. We validate this method on the widely used C-MAPSS dataset against a single-objective HPO baseline that aggregates the two objectives through the harmonic mean (HM). We demonstrate the existence of trade-offs between the predictive performance and the predictive uncertainty and observe that the bi-objective HPO returns a larger number of hyperparameter configurations compared to the single-objective baseline. Furthermore, we see that with the proposed approach, it is possible to configure models for RUL estimation that exhibit better or comparable performance to the single-objective baseline when validated on the test sets.
{"title":"End-to-End Pipeline for Uncertainty Quantification and Remaining Useful Life Estimation: An Application on Aircraft Engines","authors":"M. Kefalas, Bas van Stein, Mitra Baratchi, A. Apostolidis, T. Baeck","doi":"10.36001/phme.2022.v7i1.3317","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3317","url":null,"abstract":"Estimating the remaining useful life (RUL) of an asset lies at the heart of prognostics and health management (PHM) of many operations-critical industries such as aviation. Modern methods of RUL estimation adopt techniques from deep learning (DL). However, most of these contemporary techniques deliver only single-point estimates for the RUL without reporting on the confidence of the prediction. This practice usually provides overly confident predictions that can have severe consequences in operational disruptions or even safety. To address this issue, we propose a technique for uncertainty quantification (UQ) based on Bayesian deep learning (BDL). The hyperparameters of the framework are tuned using a novel bi-objective Bayesian optimization method with objectives the predictive performance and predictive uncertainty. The method also integrates the data pre-processing steps into the hyperparameter optimization (HPO) stage, models the RUL as a Weibull distribution, and returns the survival curves of the monitored assets to allow informed decision-making. We validate this method on the widely used C-MAPSS dataset against a single-objective HPO baseline that aggregates the two objectives through the harmonic mean (HM). We demonstrate the existence of trade-offs between the predictive performance and the predictive uncertainty and observe that the bi-objective HPO returns a larger number of hyperparameter configurations compared to the single-objective baseline. Furthermore, we see that with the proposed approach, it is possible to configure models for RUL estimation that exhibit better or comparable performance to the single-objective baseline when validated on the test sets.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124791170","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3299
Lilin Jia, Cordelia Mattuvarkuzhali Ezhilarasu, I. Jennions
In recent years, transfer learning as a method that solves many issues limiting the real-world application of conventional machine learning methods has received dramatically increasing attention in the field of machine fault diagnosis. One major finding from an initial literature review shows that the majority of the existing research only focus on the transfer of diagnostic knowledge between various conditions of the same machine or different representation of similar machines. The primary goal of the current work is to seek a way to apply transfer learning to distinct domains, thereby expanding the boundary of transfer learning in the fault diagnosis field. In particular, attempts will be made to explore ways of transferring knowledge between diagnostic tasks of different aircraft systems. One promising method to help achieving this goal is transfer learning by structural analogy, since this method is capable of extracting high-level structural knowledge to apply transfer learning between seemingly unrelated domains, similar to the scenarios of transfer between different aircraft systems.
{"title":"Novel Way to Apply Transfer Learning to Aircraft System Fault Diagnosis","authors":"Lilin Jia, Cordelia Mattuvarkuzhali Ezhilarasu, I. Jennions","doi":"10.36001/phme.2022.v7i1.3299","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3299","url":null,"abstract":"In recent years, transfer learning as a method that solves many issues limiting the real-world application of conventional machine learning methods has received dramatically increasing attention in the field of machine fault diagnosis. One major finding from an initial literature review shows that the majority of the existing research only focus on the transfer of diagnostic knowledge between various conditions of the same machine or different representation of similar machines. The primary goal of the current work is to seek a way to apply transfer learning to distinct domains, thereby expanding the boundary of transfer learning in the fault diagnosis field. In particular, attempts will be made to explore ways of transferring knowledge between diagnostic tasks of different aircraft systems. One promising method to help achieving this goal is transfer learning by structural analogy, since this method is capable of extracting high-level structural knowledge to apply transfer learning between seemingly unrelated domains, similar to the scenarios of transfer between different aircraft systems.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121553207","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 : 2022-06-29DOI: 10.36001/phme.2022.v7i1.3335
Kurçat Ince, G. Koçak, Yakup Genç
There has been an increasing demand on marine transportation and traveling, since the voyage of the ships are more economical and efficient than air or land-based alternatives. The propulsion of a ship is provided by a main engine system which includes the shaft, the propellers, and other auxiliary equipment. Marine diesel engine is a complex structure that the faults within these machines can cause malfunction of the whole system, which in turn inhibits the ship’s mission. It is crucial to monitor the engine and other auxiliary systems during the operation and infer their condition from their diagnostic data. In this study, we analyze monitoring data of a crude oil tanker for different ship loads and conditions. Our primary analysis includes main engine fault detection and classification for which we propose an end-to-end joint autoencoder-classifier model that contains a convolutional autoencoder, and a long-short term memory regressor connected to the latent space. Genetic algorithms optimized models gave us 93.61% accuracy for fault classification task. Further investigation on feature’s contributions to the model, we increased the accuracy up to 96%. One concern about marine transportation is the pollution of the air with greenhouse effect gases. In this study, we have developed NOx and SOx emission estimators for different faults and working conditions. Leveraging ship load, working conditions and engine faults in the models helped us to achieve 50% better estimation performance. Although there are other studies regarding gases emissions in the literature, this is the first study that took engine faults into account. We believe that the joint autoencoder-classifier model will be useful for other time series estimation task on other domains, especially where the operating condition plays a role in the process.
{"title":"Joint Autoencoder-Classifier Model for Malfunction Identification and Classification on Marine Diesel Engine Diagnostics Data","authors":"Kurçat Ince, G. Koçak, Yakup Genç","doi":"10.36001/phme.2022.v7i1.3335","DOIUrl":"https://doi.org/10.36001/phme.2022.v7i1.3335","url":null,"abstract":"There has been an increasing demand on marine transportation and traveling, since the voyage of the ships are more economical and efficient than air or land-based alternatives. The propulsion of a ship is provided by a main engine system which includes the shaft, the propellers, and other auxiliary equipment. Marine diesel engine is a complex structure that the faults within these machines can cause malfunction of the whole system, which in turn inhibits the ship’s mission. It is crucial to monitor the engine and other auxiliary systems during the operation and infer their condition from their diagnostic data. In this study, we analyze monitoring data of a crude oil tanker for different ship loads and conditions. Our primary analysis includes main engine fault detection and classification for which we propose an end-to-end joint autoencoder-classifier model that contains a convolutional autoencoder, and a long-short term memory regressor connected to the latent space. Genetic algorithms optimized models gave us 93.61% accuracy for fault classification task. Further investigation on feature’s contributions to the model, we increased the accuracy up to 96%. One concern about marine transportation is the pollution of the air with greenhouse effect gases. In this study, we have developed NOx and SOx emission estimators for different faults and working conditions. Leveraging ship load, working conditions and engine faults in the models helped us to achieve 50% better estimation performance. Although there are other studies regarding gases emissions in the literature, this is the first study that took engine faults into account. We believe that the joint autoencoder-classifier model will be useful for other time series estimation task on other domains, especially where the operating condition plays a role in the process.","PeriodicalId":422825,"journal":{"name":"PHM Society European Conference","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125645244","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：