Pub Date : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187055
Hassan Hijry, Richard Olawoyin
Hospital emergency departments (EDs) in vital locations face high patient demand during peak events such as the annual Islamic pilgrimage (the Hajj event) in Mecca, Saudi Arabia, the New Year celebration ceremony in New York, and the World Cup, etc. Variable patient arrival rates and hospital conditions, particularly the availability of beds for inpatients, impacts long waiting times and length of stay (LOS), causing pain and dissatisfaction to patients. Patient length of stay is chosen to be a measure of ED overcrowding as a compliance measure set by most hospitals. Clinicians need to get an opportunity to be proactive in ED overcrowding crises, specifically in the case of peak days. For this purpose, the research aims to build a model to forecast Hajj patient LOS, using machine learning algorithms through predictive input factors such as patient age, mode of arrival, and patient’s type of condition in the ED. Therefore, using machine learning algorithms, such as artificial neural networks, linear and logistic regressions, to forecast ED LOS allows clinicians to prepare for high levels of congestion and provide insights to determine the LOS of patients during vital times.
{"title":"Application of Machine Learning Algorithms for Patient Length of Stay Prediction in Emergency Department During Hajj","authors":"Hassan Hijry, Richard Olawoyin","doi":"10.1109/ICPHM49022.2020.9187055","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187055","url":null,"abstract":"Hospital emergency departments (EDs) in vital locations face high patient demand during peak events such as the annual Islamic pilgrimage (the Hajj event) in Mecca, Saudi Arabia, the New Year celebration ceremony in New York, and the World Cup, etc. Variable patient arrival rates and hospital conditions, particularly the availability of beds for inpatients, impacts long waiting times and length of stay (LOS), causing pain and dissatisfaction to patients. Patient length of stay is chosen to be a measure of ED overcrowding as a compliance measure set by most hospitals. Clinicians need to get an opportunity to be proactive in ED overcrowding crises, specifically in the case of peak days. For this purpose, the research aims to build a model to forecast Hajj patient LOS, using machine learning algorithms through predictive input factors such as patient age, mode of arrival, and patient’s type of condition in the ED. Therefore, using machine learning algorithms, such as artificial neural networks, linear and logistic regressions, to forecast ED LOS allows clinicians to prepare for high levels of congestion and provide insights to determine the LOS of patients during vital times.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131324716","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187023
Nastaran Enshaei, Safwan Ahmad, F. Naderkhani
Over the recent years, developing a reliable auto-mated visual inspection system/approach for manufacturing and industry sectors which are moving toward smart manufacturing operations faces lots of significant challenges. Traditional visual inspection techniques which are developed based on manually extracted features, can rarely be generalized and have shown weak performance in real applications in different industries. In this paper, we propose a novel and automated visual inspection system which can outperform the statistical methods in terms of detection and the quantification of anomalies in image data for performing critical industrial tasks such as detecting micro scratches on product. In particular, an end-to-end UNet-based fully convolutional neural network for automated defect detection in industrial surfaces is designed and developed. The proposed network has the capability to accept raw images as input and the output is pixel-wise masks. In order to avoid overfitting and improve the model generalization, we use real-time data augmentation approach during our training phase. To evaluate the performance of the proposed model, we use a publicly available data set containing ten different types of textured-surfaces with their associated weakly annotated masks. The findings indicate that despite working with roughly annotated labels, our results are in agreement with previous works and show improvements regarding the detection time.
{"title":"Automated detection of textured-surface defects using UNet-based semantic segmentation network","authors":"Nastaran Enshaei, Safwan Ahmad, F. Naderkhani","doi":"10.1109/ICPHM49022.2020.9187023","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187023","url":null,"abstract":"Over the recent years, developing a reliable auto-mated visual inspection system/approach for manufacturing and industry sectors which are moving toward smart manufacturing operations faces lots of significant challenges. Traditional visual inspection techniques which are developed based on manually extracted features, can rarely be generalized and have shown weak performance in real applications in different industries. In this paper, we propose a novel and automated visual inspection system which can outperform the statistical methods in terms of detection and the quantification of anomalies in image data for performing critical industrial tasks such as detecting micro scratches on product. In particular, an end-to-end UNet-based fully convolutional neural network for automated defect detection in industrial surfaces is designed and developed. The proposed network has the capability to accept raw images as input and the output is pixel-wise masks. In order to avoid overfitting and improve the model generalization, we use real-time data augmentation approach during our training phase. To evaluate the performance of the proposed model, we use a publicly available data set containing ten different types of textured-surfaces with their associated weakly annotated masks. The findings indicate that despite working with roughly annotated labels, our results are in agreement with previous works and show improvements regarding the detection time.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115647720","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187047
Qiyao Wang, Ahmed K. Farahat, Chetan Gupta, Haiyan Wang
Prognostics is concerned with predicting the future health of the equipment and any potential failures. With the advances in the Internet of Things (IoT), data-driven approaches for prognostics that leverage the power of machine learning models are gaining popularity. One of the most important categories of data-driven approaches relies on a predefined or learned health indicator to characterize the equipment condition up to the present time and make inference on how it is likely to evolve in the future. In these approaches, health indicator forecasting that constructs the health indicator curve over the lifespan using partially observed measurements (i.e., health indicator values within an initial period) plays a key role. Existing health indicator forecasting algorithms, such as the functional Empirical Bayesian approach, the regression-based formulation, a naive scenario matching based on the nearest neighbor, have certain limitations. In this paper, we propose a new ‘generative + scenario matching' algorithm for health indicator forecasting. The key idea behind the proposed approach is to first non-parametrically fit the underlying health indicator curve with a continuous Gaussian Process using a sample of run-to-failure health indicator curves. The proposed approach then generates a rich set of random curves from the learned distribution, attempting to obtain all possible variations of the target health condition evolution process over the system's lifespan. The health indicator extrapolation for a piece of functioning equipment is inferred as the generated curve that has the highest matching level within the observed period. Our experimental results show the superiority of our algorithm over the other state-of-the-art methods.
{"title":"Health Indicator Forecasting for Improving Remaining Useful Life Estimation","authors":"Qiyao Wang, Ahmed K. Farahat, Chetan Gupta, Haiyan Wang","doi":"10.1109/ICPHM49022.2020.9187047","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187047","url":null,"abstract":"Prognostics is concerned with predicting the future health of the equipment and any potential failures. With the advances in the Internet of Things (IoT), data-driven approaches for prognostics that leverage the power of machine learning models are gaining popularity. One of the most important categories of data-driven approaches relies on a predefined or learned health indicator to characterize the equipment condition up to the present time and make inference on how it is likely to evolve in the future. In these approaches, health indicator forecasting that constructs the health indicator curve over the lifespan using partially observed measurements (i.e., health indicator values within an initial period) plays a key role. Existing health indicator forecasting algorithms, such as the functional Empirical Bayesian approach, the regression-based formulation, a naive scenario matching based on the nearest neighbor, have certain limitations. In this paper, we propose a new ‘generative + scenario matching' algorithm for health indicator forecasting. The key idea behind the proposed approach is to first non-parametrically fit the underlying health indicator curve with a continuous Gaussian Process using a sample of run-to-failure health indicator curves. The proposed approach then generates a rich set of random curves from the learned distribution, attempting to obtain all possible variations of the target health condition evolution process over the system's lifespan. The health indicator extrapolation for a piece of functioning equipment is inferred as the generated curve that has the highest matching level within the observed period. Our experimental results show the superiority of our algorithm over the other state-of-the-art methods.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128995630","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187028
Jiyu Zhang, M. Salman
Electric drive system is a key subsystem of battery electric vehicles (BEVs). Since the electric drive is a single point of failure in BEVs, any abnormality in the components comprising an electric drive system may have significant impact on the vehicle performance. This paper presents a model-based method for detecting and isolating the component faults in an electric drive system, considering electric motor faults and and various sensor faults that include three phase current sensor faults and rotor position sensor faults. The method presented in this paper can be used for developing a systematic diagnostic and prognostic system for electric drive systems of electrified vehicles.
{"title":"A Model-Based Method for Fault Detection and Isolation of Electric Drive Systems","authors":"Jiyu Zhang, M. Salman","doi":"10.1109/ICPHM49022.2020.9187028","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187028","url":null,"abstract":"Electric drive system is a key subsystem of battery electric vehicles (BEVs). Since the electric drive is a single point of failure in BEVs, any abnormality in the components comprising an electric drive system may have significant impact on the vehicle performance. This paper presents a model-based method for detecting and isolating the component faults in an electric drive system, considering electric motor faults and and various sensor faults that include three phase current sensor faults and rotor position sensor faults. The method presented in this paper can be used for developing a systematic diagnostic and prognostic system for electric drive systems of electrified vehicles.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130267693","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187057
M. Rautela, S. Gopalakrishnan, Karthik Gopalakrishnan, Y. Deng
Identification of elastic properties is crucial for nondestructive material characterization as well as for in-situ condition monitoring. In this paper, we have used ultrasonic guided waves for the identification of elastic properties of a unidirectional laminate with stacked transversely isotropic lamina. The forward problem is formulated and solved using the Spectral Finite Element Method. The data collected from the forward model is utilized to solve the inverse problem of property identification. A supervised regression-based 1D-Convolutional Neural Network is trained with ultrasonic guided wave modes as inputs and elastic properties as targets. The performance of the network is evaluated based on mean squared loss, mean absolute error, and coefficient of determination. It is seen that such deep networks can learn the unknown mappings and generalize well on unseen examples.
{"title":"Ultrasonic Guided Waves Based Identification of Elastic Properties Using 1D-Convolutional Neural Networks","authors":"M. Rautela, S. Gopalakrishnan, Karthik Gopalakrishnan, Y. Deng","doi":"10.1109/ICPHM49022.2020.9187057","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187057","url":null,"abstract":"Identification of elastic properties is crucial for nondestructive material characterization as well as for in-situ condition monitoring. In this paper, we have used ultrasonic guided waves for the identification of elastic properties of a unidirectional laminate with stacked transversely isotropic lamina. The forward problem is formulated and solved using the Spectral Finite Element Method. The data collected from the forward model is utilized to solve the inverse problem of property identification. A supervised regression-based 1D-Convolutional Neural Network is trained with ultrasonic guided wave modes as inputs and elastic properties as targets. The performance of the network is evaluated based on mean squared loss, mean absolute error, and coefficient of determination. It is seen that such deep networks can learn the unknown mappings and generalize well on unseen examples.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129010490","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187032
S. Sato, Yosuke Watanabe, H. Seki, Yoshinao Ishii, Shoji Yuen
This paper proposes a learning and diagnosis method that can be applied immediately after a distributed system starts cooperative operation. The proposed method first learns behavioral rules for individual systems from their time series data, which are collected under independent operations. Then, anomality is detected and the system is diagnosed following the cooperative specification. The proposed method learns rules for individual systems based on ACEDIA, which is a kind of inductive logic programming; the rules are either transition rules or relationship rules that hold among variables at the same transition time. In a diagnostic phase, inconsistent rules and inconsistent specifications are obtained with ranking information against the diagnostic data, where ranking is performed through evaluation in terms of the generality on each rule and specification. We demonstrate that the proposed method correctly outputs the rules and specifications that are violated by diagnostic data. Moreover, in a case study on a simplified automotive system consisting of multiple control systems, the rules essentially related to the error were ranked higher.
{"title":"Fault Diagnosis for Distributed Cooperative System Using Inductive Logic Programming","authors":"S. Sato, Yosuke Watanabe, H. Seki, Yoshinao Ishii, Shoji Yuen","doi":"10.1109/ICPHM49022.2020.9187032","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187032","url":null,"abstract":"This paper proposes a learning and diagnosis method that can be applied immediately after a distributed system starts cooperative operation. The proposed method first learns behavioral rules for individual systems from their time series data, which are collected under independent operations. Then, anomality is detected and the system is diagnosed following the cooperative specification. The proposed method learns rules for individual systems based on ACEDIA, which is a kind of inductive logic programming; the rules are either transition rules or relationship rules that hold among variables at the same transition time. In a diagnostic phase, inconsistent rules and inconsistent specifications are obtained with ranking information against the diagnostic data, where ranking is performed through evaluation in terms of the generality on each rule and specification. We demonstrate that the proposed method correctly outputs the rules and specifications that are violated by diagnostic data. Moreover, in a case study on a simplified automotive system consisting of multiple control systems, the rules essentially related to the error were ranked higher.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132320941","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187029
Sara Kohtz, Pingfeng Wang
Prognostics and health management has become a prominent field for the analyses of dynamic system degradation. Specifically, methods for forecasting remaining useful life have been studied extensively, including some hybrid approaches that have indicated successful results. Mainly, a combination of machine learning and filtering techniques have shown to be the most effective. Currently, there exists a need to determine an optimal general method for remaining useful life estimation in complex systems. This paper focuses on a comparison between successful hybrid approaches. The methods are applied to modeling capacity degradation in lithium-ion batteries, with the NASA dataset utilized for this study.
{"title":"A Comparison Study of Machine Learning Enabled Filtering Methods for Battery Management","authors":"Sara Kohtz, Pingfeng Wang","doi":"10.1109/ICPHM49022.2020.9187029","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187029","url":null,"abstract":"Prognostics and health management has become a prominent field for the analyses of dynamic system degradation. Specifically, methods for forecasting remaining useful life have been studied extensively, including some hybrid approaches that have indicated successful results. Mainly, a combination of machine learning and filtering techniques have shown to be the most effective. Currently, there exists a need to determine an optimal general method for remaining useful life estimation in complex systems. This paper focuses on a comparison between successful hybrid approaches. The methods are applied to modeling capacity degradation in lithium-ion batteries, with the NASA dataset utilized for this study.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"93 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133166465","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187038
Stephen C. Adams, Tyler Cody, P. Beling, Sherwood Polter, K. Farinholt
Data-driven prognostics and health management (PHM) models are generally trained on a set of data collected from the system under study. A standard assumption of this paradigm is that the training data contains all the normal operating conditions and fault conditions that are possible. If the training data does not contain all possible conditions, a single classifier approach will not be adequate because the PHM model could have difficulty classifying a new condition not previously seen during training. This study investigates the use of hierarchical classification in situations where the training data is incomplete in terms of the faults that are present in the testing set and characterizes the proposed problem as a transfer learning problem. The hierarchical classifier employs non-mandatory leaf node prediction where the model is not required to move to the lower levels of the hierarchy. It is hypothesized that this construction allows the classification to stop at a higher level when the fault is not present in the training data. The proposed method is demonstrated on a hydraulic actuator condition monitoring data set.
{"title":"Hierarchical Classification for Unknown Faults","authors":"Stephen C. Adams, Tyler Cody, P. Beling, Sherwood Polter, K. Farinholt","doi":"10.1109/ICPHM49022.2020.9187038","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187038","url":null,"abstract":"Data-driven prognostics and health management (PHM) models are generally trained on a set of data collected from the system under study. A standard assumption of this paradigm is that the training data contains all the normal operating conditions and fault conditions that are possible. If the training data does not contain all possible conditions, a single classifier approach will not be adequate because the PHM model could have difficulty classifying a new condition not previously seen during training. This study investigates the use of hierarchical classification in situations where the training data is incomplete in terms of the faults that are present in the testing set and characterizes the proposed problem as a transfer learning problem. The hierarchical classifier employs non-mandatory leaf node prediction where the model is not required to move to the lower levels of the hierarchy. It is hypothesized that this construction allows the classification to stop at a higher level when the fault is not present in the training data. The proposed method is demonstrated on a hydraulic actuator condition monitoring data set.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129982992","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187058
Tilman Krokotsch, M. Knaak, C. Gühmann
Unsupervised Domain Adaption (DA) is an approach for adapting a data-driven model to new data without labels. Recent work on Remaining Useful Lifetime (RUL) estimation of aero engines yielded promising results for this approach. However, the current evaluation framework for DA is of limited significance when used for RUL estimation. It assumes a use case where a large number of fully degraded systems are available for adaption, which makes unsupervised DA in itself unnecessary. It is shown that the current framework overestimates adaption performance and obscures potential, negative effects of DA on performance. We propose a novel evaluation framework for unsupervised DA, specialized in RUL estimation, that takes the number of available systems and their grade of degradation into account. It enables an informed performance comparison of DA methods. We detail the framework’s capabilities on two DA methods and show that unsupervised DA delivers improved RUL estimations under real-life scenarios, as well.
{"title":"A Novel Evaluation Framework for Unsupervised Domain Adaption on Remaining Useful Lifetime Estimation","authors":"Tilman Krokotsch, M. Knaak, C. Gühmann","doi":"10.1109/ICPHM49022.2020.9187058","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187058","url":null,"abstract":"Unsupervised Domain Adaption (DA) is an approach for adapting a data-driven model to new data without labels. Recent work on Remaining Useful Lifetime (RUL) estimation of aero engines yielded promising results for this approach. However, the current evaluation framework for DA is of limited significance when used for RUL estimation. It assumes a use case where a large number of fully degraded systems are available for adaption, which makes unsupervised DA in itself unnecessary. It is shown that the current framework overestimates adaption performance and obscures potential, negative effects of DA on performance. We propose a novel evaluation framework for unsupervised DA, specialized in RUL estimation, that takes the number of available systems and their grade of degradation into account. It enables an informed performance comparison of DA methods. We detail the framework’s capabilities on two DA methods and show that unsupervised DA delivers improved RUL estimations under real-life scenarios, as well.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114739190","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 : 2020-06-01DOI: 10.1109/ICPHM49022.2020.9187056
N. K. Vuong, Yang Xue, Shudong Liu, Yu Zhou, Min Wu
In advanced manufacturing, the tool quality is critical to ensure the product quality, and thus needs to be closely monitored. However, directly measuring the tool quality can be time-consuming and impractical for continuous manufacturing. In this work, leveraging on the Internet of Things (IoT) infrastructure and machine learning techniques, two important quality metrics for turning machinery tools, namely surface roughness and flank wear, are modeled using easily available sensory data from turning machines, including vibration data, force data, and cutting parameter information. A regression-based prediction model is proposed incorporating both time domain and frequency domain features as selected using sequential replacement feature selection algorithm. The regression model itself is selected using cross validation from multiple models including linear regression, quadratic regression, random forest, and Gradient Boosting Machine (GBM). Experiment using actual manufacturing data collected with a prototype IoT setup showed that the proposed model achieved high prediction accuracy of 0.860 and low variance as indicated by Adj-R2 of 0.722 for flank wear, and similarly for surface roughness the accuracy is 0.9525 on average and Adj-R2 is 0.7175 on average. This work demonstrated the prediction of the quality metrics of turning machinery tools, which are conventionally difficult to measure in continuous manufacturing, based on IoT sensory data and machine learning techniques.
{"title":"Predicting Surface Roughness and Flank Wear in Turning Processes","authors":"N. K. Vuong, Yang Xue, Shudong Liu, Yu Zhou, Min Wu","doi":"10.1109/ICPHM49022.2020.9187056","DOIUrl":"https://doi.org/10.1109/ICPHM49022.2020.9187056","url":null,"abstract":"In advanced manufacturing, the tool quality is critical to ensure the product quality, and thus needs to be closely monitored. However, directly measuring the tool quality can be time-consuming and impractical for continuous manufacturing. In this work, leveraging on the Internet of Things (IoT) infrastructure and machine learning techniques, two important quality metrics for turning machinery tools, namely surface roughness and flank wear, are modeled using easily available sensory data from turning machines, including vibration data, force data, and cutting parameter information. A regression-based prediction model is proposed incorporating both time domain and frequency domain features as selected using sequential replacement feature selection algorithm. The regression model itself is selected using cross validation from multiple models including linear regression, quadratic regression, random forest, and Gradient Boosting Machine (GBM). Experiment using actual manufacturing data collected with a prototype IoT setup showed that the proposed model achieved high prediction accuracy of 0.860 and low variance as indicated by Adj-R2 of 0.722 for flank wear, and similarly for surface roughness the accuracy is 0.9525 on average and Adj-R2 is 0.7175 on average. This work demonstrated the prediction of the quality metrics of turning machinery tools, which are conventionally difficult to measure in continuous manufacturing, based on IoT sensory data and machine learning techniques.","PeriodicalId":148899,"journal":{"name":"2020 IEEE International Conference on Prognostics and Health Management (ICPHM)","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123894011","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：