Pub Date : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8943017
{"title":"PHM-Qingdao 2019 Committee","authors":"","doi":"10.1109/phm-qingdao46334.2019.8943017","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8943017","url":null,"abstract":"","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"232 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121868605","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8943035
Ting Yu, Guicui Fu, Y. Qiu, Ye Wang
CMOS image sensors are extensively utilized in digital imaging systems for their excellent performance and low power consumption. As an essential components in the system, CMOS image sensors are expected with low noise levels. The images captured by CMOS image sensor contain random noise (RN), digital noise (DN), and fixed pattern noise (FPN). FPN of CMOS image sensors has a greater impact on the perceived image quality than random noise, which seriously restricts the development and application of CMOS image sensors. This paper proposed a noise power spectrum (NPS) method for estimating column FPN of CMOS image sensor based on AR model. First, dozens of images under uniform illumination are acquired by established test vehicle. Second, random noise of the images is restrained by using the multi-frame averaging method. Then, column FPN is modeled by an autoregressive (AR) random process subsequently. Ultimately, column FPN is estimated by calculating NPS of the image based on the AR model. A case application was proposed by using this method.
{"title":"Noise Power Spectrum Estimation of Column Fixed Pattern Noise in CMOS Image Sensors Based on AR Model","authors":"Ting Yu, Guicui Fu, Y. Qiu, Ye Wang","doi":"10.1109/phm-qingdao46334.2019.8943035","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8943035","url":null,"abstract":"CMOS image sensors are extensively utilized in digital imaging systems for their excellent performance and low power consumption. As an essential components in the system, CMOS image sensors are expected with low noise levels. The images captured by CMOS image sensor contain random noise (RN), digital noise (DN), and fixed pattern noise (FPN). FPN of CMOS image sensors has a greater impact on the perceived image quality than random noise, which seriously restricts the development and application of CMOS image sensors. This paper proposed a noise power spectrum (NPS) method for estimating column FPN of CMOS image sensor based on AR model. First, dozens of images under uniform illumination are acquired by established test vehicle. Second, random noise of the images is restrained by using the multi-frame averaging method. Then, column FPN is modeled by an autoregressive (AR) random process subsequently. Ultimately, column FPN is estimated by calculating NPS of the image based on the AR model. A case application was proposed by using this method.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125237454","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8942942
B. Yan, Yifan Zhou, D. Yu, Xianpeng Wang
With the further opening of the bidding market in China, the accuracy of electricity price prediction directly affects the operational decisions and profits of power producers. The core factor that affects electricity price is power flow. In the early stage of electric power reform, the data of electricity price was too insufficient to support the forecasting analysis. This paper assists electric power traders to fill in the appropriate amount of electricity during the transaction process by predicting the relevant cross-section power flow. Computational methods are complex and require data of many variables at present. Therefore, this paper uses autoregressive integrated moving average (ARIMA) model and long short-term memory (LSTM) model to predict the power flow. The prediction error of the model is less than 5%. Furthermore, the conclusion shows that there is no difference between weekdays and weekends, and the power flow is a stationary time series. Based on the result of this research, some decision-making suggestions that can maximize the profit of the manufacturer are given.
{"title":"Power Flow Prediction: A Case in Ningxia Electricity Market","authors":"B. Yan, Yifan Zhou, D. Yu, Xianpeng Wang","doi":"10.1109/phm-qingdao46334.2019.8942942","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8942942","url":null,"abstract":"With the further opening of the bidding market in China, the accuracy of electricity price prediction directly affects the operational decisions and profits of power producers. The core factor that affects electricity price is power flow. In the early stage of electric power reform, the data of electricity price was too insufficient to support the forecasting analysis. This paper assists electric power traders to fill in the appropriate amount of electricity during the transaction process by predicting the relevant cross-section power flow. Computational methods are complex and require data of many variables at present. Therefore, this paper uses autoregressive integrated moving average (ARIMA) model and long short-term memory (LSTM) model to predict the power flow. The prediction error of the model is less than 5%. Furthermore, the conclusion shows that there is no difference between weekdays and weekends, and the power flow is a stationary time series. Based on the result of this research, some decision-making suggestions that can maximize the profit of the manufacturer are given.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125330864","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8942861
Y. Yu, Yunqiang Wu, Lin Yue
The blade tip-timing has become the most promising technique in the field of rotating blade vibration monitoring with its advantages of non-contacting. However the signal can be disturbed by many factors, especially the noise and drift of the blade vibration displacement curve caused by the centrifugal force changed with rotating speed. The main difficulty to data zeroing is to prevent the peak amplitude from being attenuated or eliminated. In this paper, a method was developed using blade vibration displacement to identify the areas of resonance by calculating the correlation of the data over a number of assembly revolutions from the multi-probe. The blade vibration simulator is carried out to study the relationship between the number of probes and the window width in the correlation. Applying this method into the experimental data, and verify the superiority of the correlation method.
{"title":"Data Zeroing Based on Correlation and Linear Interpolation of the Blade Tip-Timing Data","authors":"Y. Yu, Yunqiang Wu, Lin Yue","doi":"10.1109/phm-qingdao46334.2019.8942861","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8942861","url":null,"abstract":"The blade tip-timing has become the most promising technique in the field of rotating blade vibration monitoring with its advantages of non-contacting. However the signal can be disturbed by many factors, especially the noise and drift of the blade vibration displacement curve caused by the centrifugal force changed with rotating speed. The main difficulty to data zeroing is to prevent the peak amplitude from being attenuated or eliminated. In this paper, a method was developed using blade vibration displacement to identify the areas of resonance by calculating the correlation of the data over a number of assembly revolutions from the multi-probe. The blade vibration simulator is carried out to study the relationship between the number of probes and the window width in the correlation. Applying this method into the experimental data, and verify the superiority of the correlation method.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"24 9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125672459","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8942914
Yaohua Chen, C. Zhang, Ning Zhang, Yiting Chen, Huan Wang
The traction motor is one of the key components that plays an important role in ensuring the safety and stability of the running EMU (Electric Multiple Units). The running state of the traction motor can be determined through monitoring and predicting the change of EMU bearing temperature. In this paper, we propose a Long Short-Term Memory Neural Network based on Multi-task Learning and Attention Mechanism for the bearing temperature prediction in view of the complex influencing factors of bearing temperature in train operation. The model learns the characteristics of temperature sensors in different positions jointly through multi-task learning. And the Long Short-Term Memory Neural Network based on Attention Mechanism is used to consider the influence of current operating conditions and previous train records on bearing temperature in different degrees. So the model takes various influencing factors and spatial-temporal correlation into consideration. The experimental results with actual EMU datasets show that our method outperforms the baseline approaches.
{"title":"Multi-Task Learning and Attention Mechanism Based Long Short-Term Memory for Temperature Prediction of EMU Bearing","authors":"Yaohua Chen, C. Zhang, Ning Zhang, Yiting Chen, Huan Wang","doi":"10.1109/phm-qingdao46334.2019.8942914","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8942914","url":null,"abstract":"The traction motor is one of the key components that plays an important role in ensuring the safety and stability of the running EMU (Electric Multiple Units). The running state of the traction motor can be determined through monitoring and predicting the change of EMU bearing temperature. In this paper, we propose a Long Short-Term Memory Neural Network based on Multi-task Learning and Attention Mechanism for the bearing temperature prediction in view of the complex influencing factors of bearing temperature in train operation. The model learns the characteristics of temperature sensors in different positions jointly through multi-task learning. And the Long Short-Term Memory Neural Network based on Attention Mechanism is used to consider the influence of current operating conditions and previous train records on bearing temperature in different degrees. So the model takes various influencing factors and spatial-temporal correlation into consideration. The experimental results with actual EMU datasets show that our method outperforms the baseline approaches.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125763311","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8943002
Xiaohang Jin, Z. Que, Yi Sun
Bearing failure can cause their host system shutdown, and even some catastrophic accidents. These will lead to a high maintenance cost and a huge economic loss. Thus, health monitoring and fault prognosis for bearings becomes increasingly important. Developing an effective health index (HI) will do help in these works. Hence, three different HIs are developed by using root mean square, Kolmogorov-Smirnov test, and Mahalanobis distance to reflect bearings’ online health conditions. Four degradation models are constructed to estimate bearings remaining useful life (RUL) by using particle filter algorithm. Bearing life data are used to test the performance of fault prognostic approaches. Results show that all HIs reflect the degradation process of bearing effectively, and the proposed degradation model has the best performance in bearing RUL prediction.
{"title":"Development of Vibration-Based Health Indexes for Bearing Remaining Useful Life Prediction","authors":"Xiaohang Jin, Z. Que, Yi Sun","doi":"10.1109/phm-qingdao46334.2019.8943002","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8943002","url":null,"abstract":"Bearing failure can cause their host system shutdown, and even some catastrophic accidents. These will lead to a high maintenance cost and a huge economic loss. Thus, health monitoring and fault prognosis for bearings becomes increasingly important. Developing an effective health index (HI) will do help in these works. Hence, three different HIs are developed by using root mean square, Kolmogorov-Smirnov test, and Mahalanobis distance to reflect bearings’ online health conditions. Four degradation models are constructed to estimate bearings remaining useful life (RUL) by using particle filter algorithm. Bearing life data are used to test the performance of fault prognostic approaches. Results show that all HIs reflect the degradation process of bearing effectively, and the proposed degradation model has the best performance in bearing RUL prediction.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124537915","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8943014
Ruihua Jiao, Kai-xiang Peng, Jie Dong, Kai Zhang, Chuang-jian Zhang
Remaining useful life (RUL) prediction is of great importance in a successful prognostics and health management system. The performance of RUL prediction is mainly decided by the development of an appropriate health indicator (HI), which can accurately indicate the degree of degradation of the equipment. Therefore, we proposed an unsupervised method for HI construction based on deep belief network (DBN) by using multisensory historical data. Firstly, DBN is employed to describe the hidden representation corresponding to the healthy state. With the running of the system, its performance will decrease over time and the corresponding potential characteristics tend to be different. The deviation degree of degraded state can be used to establish HI so as to estimate the RUL. Finally, a case study is conducted to validate the effectiveness of the new method, where it can be seen that the new approach achieves better performance compared to traditional methods.
{"title":"A Health Indicator Construction Method based on Deep Belief Network for Remaining Useful Life Prediction","authors":"Ruihua Jiao, Kai-xiang Peng, Jie Dong, Kai Zhang, Chuang-jian Zhang","doi":"10.1109/phm-qingdao46334.2019.8943014","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8943014","url":null,"abstract":"Remaining useful life (RUL) prediction is of great importance in a successful prognostics and health management system. The performance of RUL prediction is mainly decided by the development of an appropriate health indicator (HI), which can accurately indicate the degree of degradation of the equipment. Therefore, we proposed an unsupervised method for HI construction based on deep belief network (DBN) by using multisensory historical data. Firstly, DBN is employed to describe the hidden representation corresponding to the healthy state. With the running of the system, its performance will decrease over time and the corresponding potential characteristics tend to be different. The deviation degree of degraded state can be used to establish HI so as to estimate the RUL. Finally, a case study is conducted to validate the effectiveness of the new method, where it can be seen that the new approach achieves better performance compared to traditional methods.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"79 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124589487","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8943060
Shenghao Shi, Yongzhi Qu, Jinglin Wang, Liu Hong, J. Dhupia, Zude Zhou
The gearbox is one of the most common and important components in the drivetrains. Thus, the online monitoring of the dynamic behavior of geared system is crucial for the optimization, diagnosis and prognosis of the drivetrains. The conventional online monitoring system for gearboxes is to use the vibration sensor mounted on the gear housing. However, in the measured housing vibration signal, the dynamic response of the monitored geared pair is usually distorted, which is caused by the complex transfer path of the vibration. Therefore, to advance the art of online monitoring of gearboxes, this work proposes to employ the fiber Bragg grating as the strain sensor to mount near the gear mesh region. The experimental assessment of the feasibility of the fiber Bragg grating based online monitoring system is conducted in a laboratory fixed-axis spur gearbox. To validate and analyze the measurement from the fiber Bragg grating system, a gear mesh model is developed using the finite element method. The comparison between the measurement and theoretical simulation show the proposed fiber Bragg grating based online monitoring system is capable to capture the variation of the root strain during the gear mesh process. This result proves the proposed technique has a promising potential in developing a commercial online monitoring system to measure the subtle dynamic behavior of gearboxes.
{"title":"Feasibility Study of Online Monitoring Using the Fiber Bragg Grating Sensor for Geared System","authors":"Shenghao Shi, Yongzhi Qu, Jinglin Wang, Liu Hong, J. Dhupia, Zude Zhou","doi":"10.1109/phm-qingdao46334.2019.8943060","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8943060","url":null,"abstract":"The gearbox is one of the most common and important components in the drivetrains. Thus, the online monitoring of the dynamic behavior of geared system is crucial for the optimization, diagnosis and prognosis of the drivetrains. The conventional online monitoring system for gearboxes is to use the vibration sensor mounted on the gear housing. However, in the measured housing vibration signal, the dynamic response of the monitored geared pair is usually distorted, which is caused by the complex transfer path of the vibration. Therefore, to advance the art of online monitoring of gearboxes, this work proposes to employ the fiber Bragg grating as the strain sensor to mount near the gear mesh region. The experimental assessment of the feasibility of the fiber Bragg grating based online monitoring system is conducted in a laboratory fixed-axis spur gearbox. To validate and analyze the measurement from the fiber Bragg grating system, a gear mesh model is developed using the finite element method. The comparison between the measurement and theoretical simulation show the proposed fiber Bragg grating based online monitoring system is capable to capture the variation of the root strain during the gear mesh process. This result proves the proposed technique has a promising potential in developing a commercial online monitoring system to measure the subtle dynamic behavior of gearboxes.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130078687","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8943006
Xiaochuan Li, D. Mba, Tianran Lin
In this work, a hybrid prognostic framework which interfaces a model-based prognostic method, namely particle filter, with a similarity-based prognostic method is proposed. The proposed framework consists of automatic determination of predication start point, sensor fusion, and prognostics steps that lead to accurate remaining useful life (RUL) estimations. This approach first applies the canonical variate analysis (CVA) approach for determining the prediction start time and constructing the prognostic health indicators (HIs). The similarity-based method is then employed together with the model-based particle filter (PF) algorithm to improve the predictive performance in terms of reducing the uncertainty of RUL and improving the prediction accuracy. The proposed framework can automatically construct HIs that are suitable for RUL prediction and offer higher prediction accuracy and lower uncertainty boundaries than traditional model-based PF methods. Our proposed approach is successfully applied on aircraft turbofan engines RUL prediction.
{"title":"A Similarity-based and Model-based Fusion Prognostics Framework for Remaining Useful Life Prediction","authors":"Xiaochuan Li, D. Mba, Tianran Lin","doi":"10.1109/phm-qingdao46334.2019.8943006","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8943006","url":null,"abstract":"In this work, a hybrid prognostic framework which interfaces a model-based prognostic method, namely particle filter, with a similarity-based prognostic method is proposed. The proposed framework consists of automatic determination of predication start point, sensor fusion, and prognostics steps that lead to accurate remaining useful life (RUL) estimations. This approach first applies the canonical variate analysis (CVA) approach for determining the prediction start time and constructing the prognostic health indicators (HIs). The similarity-based method is then employed together with the model-based particle filter (PF) algorithm to improve the predictive performance in terms of reducing the uncertainty of RUL and improving the prediction accuracy. The proposed framework can automatically construct HIs that are suitable for RUL prediction and offer higher prediction accuracy and lower uncertainty boundaries than traditional model-based PF methods. Our proposed approach is successfully applied on aircraft turbofan engines RUL prediction.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121817283","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 : 2019-10-01DOI: 10.1109/phm-qingdao46334.2019.8942879
D. He, W. Guo, Mao He
Deep learning is the most attractive topic in the field of machine learning and relevant applications. Owing to the strong learning ability of the convolutional neural network (CNN), it integrates the feature extraction from raw data and classification as a complete learning process and makes the bearing fault diagnosis intelligent. In the published results, the inputs of the CNN may be the raw temporal waveform of vibration, its processed waveform or converted 2D images. In this paper, focusing on the diagnosis accuracy of rolling bearings, a comparative study is conducted among the inputs using the raw temporal waveform, the frequency spectrum, and the envelope spectrum of a vibration signal. First, an appropriate classification model based on the CNN is constructed. Then, experimental data from bearing with real damages are collected and then transformed and converted into some small gray pixel images for training and testing the CNN model. Finally, the classification accuracies using three signals are compared. The results indicate that the diagnosis performances using the above three signals are close when the trained CNN models are stable; among them the model using the frequency spectrum of the vibration signal is a little better than the models using the other two signals, which may be a reference for further investigating the deep learning used in the field of bearing diagnosis.
{"title":"Bearing Diagnosis Accuracy Comparison Using Convolutional Neural Network with Time/Frequency Domain Signals","authors":"D. He, W. Guo, Mao He","doi":"10.1109/phm-qingdao46334.2019.8942879","DOIUrl":"https://doi.org/10.1109/phm-qingdao46334.2019.8942879","url":null,"abstract":"Deep learning is the most attractive topic in the field of machine learning and relevant applications. Owing to the strong learning ability of the convolutional neural network (CNN), it integrates the feature extraction from raw data and classification as a complete learning process and makes the bearing fault diagnosis intelligent. In the published results, the inputs of the CNN may be the raw temporal waveform of vibration, its processed waveform or converted 2D images. In this paper, focusing on the diagnosis accuracy of rolling bearings, a comparative study is conducted among the inputs using the raw temporal waveform, the frequency spectrum, and the envelope spectrum of a vibration signal. First, an appropriate classification model based on the CNN is constructed. Then, experimental data from bearing with real damages are collected and then transformed and converted into some small gray pixel images for training and testing the CNN model. Finally, the classification accuracies using three signals are compared. The results indicate that the diagnosis performances using the above three signals are close when the trained CNN models are stable; among them the model using the frequency spectrum of the vibration signal is a little better than the models using the other two signals, which may be a reference for further investigating the deep learning used in the field of bearing diagnosis.","PeriodicalId":259179,"journal":{"name":"2019 Prognostics and System Health Management Conference (PHM-Qingdao)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121965145","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：