Pub Date : 2022-09-01DOI: 10.1784/insi.2022.64.9.520
Kai-Shang Lin, Zhiran Zhou, D. Pan, Yu Zhang
Valves and other actuators may fail and cause economic losses or safety accidents. To ensure the stable operation of a control system, it is necessary to identify the failures of various valves and carry out the corresponding maintenance. Several methods are designed and implemented for valve fault diagnosis in this paper. In particular, a novel fault diagnosis method based on a time-series convolution network (FDM-TSCN) is proposed, which is built on a time-series data feature extracting and convolutional neural network. FDM-TSCN can classify 18 out of 19 types of fault, while many other methods cannot. This algorithm is presented in detail and implemented as a prototype system. Comprehensive simulations are performed on valve fault datasets that are generated by the development and application of methods for actuator fault diagnosis in industrial systems (DAMADICS). The simulation results prove the effectiveness and superiority of the proposed FDM-TSCN method. All of the source codes and related data in the paper are made available, which enables other researchers to verify the work easily and may inspire them to carry out more informed research.
{"title":"Fault diagnosis methods based on a time-series convolution and the comparison of multiple methods","authors":"Kai-Shang Lin, Zhiran Zhou, D. Pan, Yu Zhang","doi":"10.1784/insi.2022.64.9.520","DOIUrl":"https://doi.org/10.1784/insi.2022.64.9.520","url":null,"abstract":"Valves and other actuators may fail and cause economic losses or safety accidents. To ensure the stable operation of a control system, it is necessary to identify the failures of various valves and carry out the corresponding maintenance. Several methods are designed and implemented\u0000 for valve fault diagnosis in this paper. In particular, a novel fault diagnosis method based on a time-series convolution network (FDM-TSCN) is proposed, which is built on a time-series data feature extracting and convolutional neural network. FDM-TSCN can classify 18 out of 19 types of fault,\u0000 while many other methods cannot. This algorithm is presented in detail and implemented as a prototype system. Comprehensive simulations are performed on valve fault datasets that are generated by the development and application of methods for actuator fault diagnosis in industrial systems\u0000 (DAMADICS). The simulation results prove the effectiveness and superiority of the proposed FDM-TSCN method. All of the source codes and related data in the paper are made available, which enables other researchers to verify the work easily and may inspire them to carry out more informed research.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117271397","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}
In the quality analysis of contemporary industrial production of praseodymium-neodymium (Pr-Nd) alloys, the amount of carbon content is mainly determined using chemical analysis methods. To overcome the shortcomings of the long durations and high costs of quality inspection cycles, this study proposes a non-destructive model for determining the carbon content of Pr-Nd alloys using acoustic emission signals collected using a mel frequency cepstral coefficient (MFCC) long short-term memory (LSTM) network (MFCC-LSTM) model and a data acquisition system. The MFCC ensures accurate signal feature extraction and data dimensionality reduction and the LSTM enables learning of the extracted features. The recognition rate of the MFCC-LSTM model reaches up to 97.53%, which can satisfy the quality inspection requirements for the industrial production of Pr-Nd alloys. In model evaluation, the receiver operating characteristic (ROC) curve shows good performance indices, indicating that the model is robust. Real-time verification of the model shows that the proposed method greatly shortens the time of each quality inspection link; the quality inspection time for a single piece of Pr-Nd alloy is only 0.3-0.65 s, which is a good real-time parameter.
{"title":"Acoustic emission signal-based non-destructive testing of carbon content of Pr-Nd alloys","authors":"Xinyu Chen, Xin-yu Wu, Feifei Liu, Bo-hua Zeng, Yuan-min Tu, Le-le Cao","doi":"10.1784/insi.2022.64.9.503","DOIUrl":"https://doi.org/10.1784/insi.2022.64.9.503","url":null,"abstract":"In the quality analysis of contemporary industrial production of praseodymium-neodymium (Pr-Nd) alloys, the amount of carbon content is mainly determined using chemical analysis methods. To overcome the shortcomings of the long durations and high costs of quality inspection cycles,\u0000 this study proposes a non-destructive model for determining the carbon content of Pr-Nd alloys using acoustic emission signals collected using a mel frequency cepstral coefficient (MFCC) long short-term memory (LSTM) network (MFCC-LSTM) model and a data acquisition system. The MFCC ensures\u0000 accurate signal feature extraction and data dimensionality reduction and the LSTM enables learning of the extracted features. The recognition rate of the MFCC-LSTM model reaches up to 97.53%, which can satisfy the quality inspection requirements for the industrial production of Pr-Nd alloys.\u0000 In model evaluation, the receiver operating characteristic (ROC) curve shows good performance indices, indicating that the model is robust. Real-time verification of the model shows that the proposed method greatly shortens the time of each quality inspection link; the quality inspection time\u0000 for a single piece of Pr-Nd alloy is only 0.3-0.65 s, which is a good real-time parameter.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123446366","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 : 2022-09-01DOI: 10.1784/insi.2022.64.9.496
G. Shen, Yongna Shen, Yilin Yuan
For the purpose of rapidly detecting and evaluating active cracks in the weld seam and the heat-affected zones of carbon steel pressure vessels, acoustic emission (AE) tests are conducted on a pressure vessel with natural cracks. The AE locations and parametric distribution characteristics of these cracks are investigated and presented. It is shown that surface cracks propagate and generate AE location sources at lower pressures than internal cracks. For both internal and surface cracks, more than 85% of the AE location sources are generated during the initial pressurisation steps. During the depressurisation steps, surface cracks may also generate AE location sources due to crack closure. Compared with the parametric distribution of hits, the peaks in the distribution graph of AE location sources in the main parametric scale shift to higher parametric values. In this study, the amplitude of the AE location sources is approximately 53 dBAE , the energy is around 32 eu, the rise time is in the range of 20-100 μs and the count is in the range of 1-200. The distribution of corresponding parameters of AE hits is similar for both surface and internal weld cracks. The AE location and parametric distribution features can be used directly to identify weld cracks.
{"title":"Study of the acoustic emission characteristics of weld cracks in carbon steel pressure vessels","authors":"G. Shen, Yongna Shen, Yilin Yuan","doi":"10.1784/insi.2022.64.9.496","DOIUrl":"https://doi.org/10.1784/insi.2022.64.9.496","url":null,"abstract":"For the purpose of rapidly detecting and evaluating active cracks in the weld seam and the heat-affected zones of carbon steel pressure vessels, acoustic emission (AE) tests are conducted on a pressure vessel with natural cracks. The AE locations and parametric distribution characteristics\u0000 of these cracks are investigated and presented. It is shown that surface cracks propagate and generate AE location sources at lower pressures than internal cracks. For both internal and surface cracks, more than 85% of the AE location sources are generated during the initial pressurisation\u0000 steps. During the depressurisation steps, surface cracks may also generate AE location sources due to crack closure. Compared with the parametric distribution of hits, the peaks in the distribution graph of AE location sources in the main parametric scale shift to higher parametric values.\u0000 In this study, the amplitude of the AE location sources is approximately 53 dBAE , the energy is around 32 eu, the rise time is in the range of 20-100 μs and the count is in the range of 1-200. The distribution of corresponding parameters of AE hits is similar for both surface\u0000 and internal weld cracks. The AE location and parametric distribution features can be used directly to identify weld cracks.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128426844","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 : 2022-09-01DOI: 10.1784/insi.2022.64.9.511
Deyan Yang, Hongquan Jiang, Z. Liu, Yonghong Wang, Huyue Cheng
Radiographic testing (RT) images of complex components are affected by several factors, including low greyscale levels, low contrast and blur. These factors can significantly restrict the accuracy and effectiveness of defect recognition. To address this issue, this paper proposes a radiographic image enhancement method based on a triple constraint U-Net network. Firstly, a radiographic image preprocessing target dataset is constructed based on conventional image preprocessing technology and previous experience. The U-Net model is then used to design a model loss function, including the parameters of image consistency, texture consistency and structural similarity, in order to achieve structure preservation and noise removal in the images. Finally, radiographic images of actual complex components are used to illustrate and verify the effectiveness of the proposed method. The results show that the proposed method can effectively convert an original image to a target image, enhance the details of the defect area and improve the accuracy of defect recognition by 5.2%.
{"title":"Radiographic image enhancement based on a triple constraint U-Net network","authors":"Deyan Yang, Hongquan Jiang, Z. Liu, Yonghong Wang, Huyue Cheng","doi":"10.1784/insi.2022.64.9.511","DOIUrl":"https://doi.org/10.1784/insi.2022.64.9.511","url":null,"abstract":"Radiographic testing (RT) images of complex components are affected by several factors, including low greyscale levels, low contrast and blur. These factors can significantly restrict the accuracy and effectiveness of defect recognition. To address this issue, this paper proposes a\u0000 radiographic image enhancement method based on a triple constraint U-Net network. Firstly, a radiographic image preprocessing target dataset is constructed based on conventional image preprocessing technology and previous experience. The U-Net model is then used to design a model loss function,\u0000 including the parameters of image consistency, texture consistency and structural similarity, in order to achieve structure preservation and noise removal in the images. Finally, radiographic images of actual complex components are used to illustrate and verify the effectiveness of the proposed\u0000 method. The results show that the proposed method can effectively convert an original image to a target image, enhance the details of the defect area and improve the accuracy of defect recognition by 5.2%.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114778141","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 : 2022-09-01DOI: 10.1784/insi.2022.64.9.528
Genshang Wu, Xinyao Sun, S. Hao, Xianfeng Yan, Yitao Zhao
Loosening of bolts, which is a common form of failure in bolted connections, causes relative slippage between the connected surfaces. The bolts fail under the action of external shear forces due to fatigue and breakage, thereby affecting the service performance and connection strength of the equipment, potentially resulting in major accidents. At present, condition monitoring, which is used to detect the tightness of bolt connections, has obtained acceptable results; however, most of them are still carried out under laboratory conditions and cannot be applied to engineering. In addition, effective remedial measures should be implemented after detecting bolt looseness. On the basis of such problems, a multi-bolt looseness monitoring method based on machine vision and deep learning is proposed. At the same time, shape memory alloy is used in the design of a structure that actively compensates for loose bolts. This method realises bolt recognition of the bolt connection structure through video monitoring and looseness monitoring of multi-target bolts at the same time. When the system detects that the bolts are loosened, an alarm signal is issued and, at the same time, the control device is activated to compensate, to increase the time available for repair time and to ensure the service performance of major equipment.
{"title":"Deep learning recognition of bolt looseness and axial force compensation of shape memory alloy","authors":"Genshang Wu, Xinyao Sun, S. Hao, Xianfeng Yan, Yitao Zhao","doi":"10.1784/insi.2022.64.9.528","DOIUrl":"https://doi.org/10.1784/insi.2022.64.9.528","url":null,"abstract":"Loosening of bolts, which is a common form of failure in bolted connections, causes relative slippage between the connected surfaces. The bolts fail under the action of external shear forces due to fatigue and breakage, thereby affecting the service performance and connection strength\u0000 of the equipment, potentially resulting in major accidents. At present, condition monitoring, which is used to detect the tightness of bolt connections, has obtained acceptable results; however, most of them are still carried out under laboratory conditions and cannot be applied to engineering.\u0000 In addition, effective remedial measures should be implemented after detecting bolt looseness. On the basis of such problems, a multi-bolt looseness monitoring method based on machine vision and deep learning is proposed. At the same time, shape memory alloy is used in the design of a structure\u0000 that actively compensates for loose bolts. This method realises bolt recognition of the bolt connection structure through video monitoring and looseness monitoring of multi-target bolts at the same time. When the system detects that the bolts are loosened, an alarm signal is issued and, at\u0000 the same time, the control device is activated to compensate, to increase the time available for repair time and to ensure the service performance of major equipment.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128823525","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 : 2022-08-01DOI: 10.1784/insi.2022.64.8.437
J. Liška, J. Jakl, S. Kunkel
Turbine generator torsional vibration is becoming a major concern in modern power grids with a high level of changeability due to the operation of renewable energy sources. The traditional absence of standard torsional vibration monitoring and a lack of experience with the operation of torsional vibration monitoring systems opens up a wide range of opportunities for the design of torsional vibration monitoring systems and the possibility of their installation in power plants. As the measured signals are adversely affected by noise, proper filtering is essential for capturing the torsional vibration information. The benefits of the designed Kalman filtering method are the computational efficiency and the possibility of tackling two different types of noise: the state noise and the measurement noise. The feasibility of the proposed method is demonstrated by case studies based on practical signals measured on steam turbine generators.
{"title":"Advanced turbine generator torsional vibration evaluation method using Kalman filtering","authors":"J. Liška, J. Jakl, S. Kunkel","doi":"10.1784/insi.2022.64.8.437","DOIUrl":"https://doi.org/10.1784/insi.2022.64.8.437","url":null,"abstract":"Turbine generator torsional vibration is becoming a major concern in modern power grids with a high level of changeability due to the operation of renewable energy sources. The traditional absence of standard torsional vibration monitoring and a lack of experience with the operation\u0000 of torsional vibration monitoring systems opens up a wide range of opportunities for the design of torsional vibration monitoring systems and the possibility of their installation in power plants. As the measured signals are adversely affected by noise, proper filtering is essential for capturing\u0000 the torsional vibration information. The benefits of the designed Kalman filtering method are the computational efficiency and the possibility of tackling two different types of noise: the state noise and the measurement noise. The feasibility of the proposed method is demonstrated by case\u0000 studies based on practical signals measured on steam turbine generators.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129188566","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 : 2022-08-01DOI: 10.1784/insi.2022.64.8.442
M. Moslem, A. Zippo, G. Iarriccio, L. Bergamini, F. Pellicano
Bevel gear pairs are employed extensively in transmission systems, such as vehicle transmissions (rear axle drive), aircraft engines/turbines and helicopter gears, to transfer power between non-parallel shafts at high speed or high torque. The most complex form of bevel gear is the spiral bevel gear (SBG). SBG pairs are commonly used in applications that require high load capacity at higher operating speeds than are typically possible with other types of bevel gear. When manufactured in a metal-cutting process, spiral bevel gears can either be produced using single indexing (a face-milling method, which is considered in this study) or continuous indexing (a face-hobbing method). Due to manufacturing imperfections and the flexibility of components, the system might experience misalignments that intensify or exert a destructive effect on the gear vibration, which causes disruption in the stress distribution, thereby decreasing the lifetime of the gearbox. The main purpose of this study is to carry out loaded tooth contact analysis (LTCA) and unloaded tooth contact analysis (UTCA) for an SBG pair in the presence of two types of misalignment, axial and radial misalignment, and represent their effects on the mesh stiffness (MS). To calculate the MS, it is essential to determine the geometrical mismatch between two mating tooth profiles by means of UTCA. To conduct LTCA, three main approaches can be utilised: the finite element method (FEM) and experimental and analytical approaches. Due to the development of software packages during the last decade, Transmission3D-Calyx, an FEM-based software, is used in this study to carry out LTCA and UTCA. Finally, the MS for different misalignment cases is compared to represent the effect of misalignment on the SBG pair.
{"title":"Loaded and unloaded tooth contact analysis of spiral bevel gears in consideration of misalignments","authors":"M. Moslem, A. Zippo, G. Iarriccio, L. Bergamini, F. Pellicano","doi":"10.1784/insi.2022.64.8.442","DOIUrl":"https://doi.org/10.1784/insi.2022.64.8.442","url":null,"abstract":"Bevel gear pairs are employed extensively in transmission systems, such as vehicle transmissions (rear axle drive), aircraft engines/turbines and helicopter gears, to transfer power between non-parallel shafts at high speed or high torque. The most complex form of bevel gear is the\u0000 spiral bevel gear (SBG). SBG pairs are commonly used in applications that require high load capacity at higher operating speeds than are typically possible with other types of bevel gear. When manufactured in a metal-cutting process, spiral bevel gears can either be produced using single indexing\u0000 (a face-milling method, which is considered in this study) or continuous indexing (a face-hobbing method). Due to manufacturing imperfections and the flexibility of components, the system might experience misalignments that intensify or exert a destructive effect on the gear vibration, which\u0000 causes disruption in the stress distribution, thereby decreasing the lifetime of the gearbox. The main purpose of this study is to carry out loaded tooth contact analysis (LTCA) and unloaded tooth contact analysis (UTCA) for an SBG pair in the presence of two types of misalignment, axial and\u0000 radial misalignment, and represent their effects on the mesh stiffness (MS). To calculate the MS, it is essential to determine the geometrical mismatch between two mating tooth profiles by means of UTCA. To conduct LTCA, three main approaches can be utilised: the finite element method (FEM)\u0000 and experimental and analytical approaches. Due to the development of software packages during the last decade, Transmission3D-Calyx, an FEM-based software, is used in this study to carry out LTCA and UTCA. Finally, the MS for different misalignment cases is compared to represent the effect\u0000 of misalignment on the SBG pair.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"130 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123450587","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 : 2022-08-01DOI: 10.1784/insi.2022.64.8.432
G. Capobianco, N. Bohun, M. Gratton, R. Serra, A. Zinbi, N. Rigollet
This paper presents a method for determining the axial and radial load distribution of the moment generated in a V-band clamp and is validated experimentally using finite element analysis (FEA). The method comprises a slotted flange, which is distinguished by having three different profiles for different levels of load symmetrically divided among eight sectors. Each sector is characterised and calibrated. The load cell is analysed using finite element Abaqus software to predict and corroborate the system. In the experimental test, the axial and radial loads are measured using strain gauges for each sector and the total axial load is validated by three button sensors. Tests on the V-band clamp were successfully carried out and indicated a non-uniform distribution of axial and radial loads, with three highlights relating to existing papers: improved results for axial loads, new results for radial loads and an analysis of the moment and its direction, which is consistent with finite element studies.
{"title":"Both radial and axial load distribution measurement on a V-band clamp by a new load cell design","authors":"G. Capobianco, N. Bohun, M. Gratton, R. Serra, A. Zinbi, N. Rigollet","doi":"10.1784/insi.2022.64.8.432","DOIUrl":"https://doi.org/10.1784/insi.2022.64.8.432","url":null,"abstract":"This paper presents a method for determining the axial and radial load distribution of the moment generated in a V-band clamp and is validated experimentally using finite element analysis (FEA). The method comprises a slotted flange, which is distinguished by having three different\u0000 profiles for different levels of load symmetrically divided among eight sectors. Each sector is characterised and calibrated. The load cell is analysed using finite element Abaqus software to predict and corroborate the system. In the experimental test, the axial and radial loads are measured\u0000 using strain gauges for each sector and the total axial load is validated by three button sensors. Tests on the V-band clamp were successfully carried out and indicated a non-uniform distribution of axial and radial loads, with three highlights relating to existing papers: improved results\u0000 for axial loads, new results for radial loads and an analysis of the moment and its direction, which is consistent with finite element studies.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116654143","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 : 2022-08-01DOI: 10.1784/insi.2022.64.8.447
A. Kiakojouri, Z. Lu, P. Mirring, H. Powrie, Ling Wang
Intelligent fault classification of rolling element bearings (REBs) using machine learning (ML) techniques increases the reliability of industrial assets. One of the main issues associated with ML model development is the lack of training data and, most importantly, the ability of models to be used for applications without specific training data, ie the generalisation capability of models. This study investigates the feasibility of using multinomial logistic regression (MLR) as generalised ML models for rolling element bearing fault classification without the requirement of training data for new bearing designs and varied machine operations. This has been achieved by using bearing characteristic frequencies (BCFs) as inputs to the MLR models extracted by a newly developed hybrid method. The new method combines cepstrum pre-whitening (CPW) and full-band enveloping, which can effectively identify the BCFs in vibration data from various machines. This paper presents the methods of the feature extraction and the development of generalised ML models for REBs based on data from the EU Clean Sky 2 I2BS project1. This model is then validated by data from Case Western Reserve University (CWRU) and the Society for Machinery Failure Prevention Technology (MFPT), available in the public domain without further training.
{"title":"A generalised machine learning model based on multinomial logistic regression and frequency features for rolling bearing fault classification","authors":"A. Kiakojouri, Z. Lu, P. Mirring, H. Powrie, Ling Wang","doi":"10.1784/insi.2022.64.8.447","DOIUrl":"https://doi.org/10.1784/insi.2022.64.8.447","url":null,"abstract":"Intelligent fault classification of rolling element bearings (REBs) using machine learning (ML) techniques increases the reliability of industrial assets. One of the main issues associated with ML model development is the lack of training data and, most importantly, the ability of models\u0000 to be used for applications without specific training data, ie the generalisation capability of models. This study investigates the feasibility of using multinomial logistic regression (MLR) as generalised ML models for rolling element bearing fault classification without the requirement of\u0000 training data for new bearing designs and varied machine operations. This has been achieved by using bearing characteristic frequencies (BCFs) as inputs to the MLR models extracted by a newly developed hybrid method. The new method combines cepstrum pre-whitening (CPW) and full-band enveloping,\u0000 which can effectively identify the BCFs in vibration data from various machines. This paper presents the methods of the feature extraction and the development of generalised ML models for REBs based on data from the EU Clean Sky 2 I2BS project1. This model is then validated\u0000 by data from Case Western Reserve University (CWRU) and the Society for Machinery Failure Prevention Technology (MFPT), available in the public domain without further training.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"332 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134454589","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 : 2022-08-01DOI: 10.1784/insi.2022.64.8.453
M. Káš, F. F. Wamba
The impact of an anomaly is domain-dependent. In a dataset of network activities, an anomaly can imply an intrusion attack. Other objectives of anomaly detection are industrial damage detection, data leak prevention, identifying security vulnerabilities or military surveillance. Anomalies are observations or a sequence of observations in which the distribution deviates remarkably from the general distribution of the whole dataset. A large majority of the dataset consists of normal (healthy) data points. The anomalies form only a very small part of the dataset. Anomaly detection is the technique used to find these observations and its methods are specific to the type of data. While there is a wide spectrum of anomaly detection approaches available today, it becomes increasingly difficult to keep track of all the techniques. In fact, it is not clear which of the three categories of detection methods, ie statistical approaches, machine learning approaches or deep learning approaches, is more appropriate to detect anomalies in time-series data, which are mainly used in industry. A typical industrial device has multi-dimensional characteristics. It is possible to measure voltage, current, active power, vibrations, rotational speed, temperature, pressure difference, etc, on such a device. Early detection of the anomalous behaviour of industrial devices can help reduce or prevent serious damage, which could lead to significant financial loss. This paper presents a summary of the methods used to detect anomalies in condition monitoring applications.
{"title":"Anomaly detection-based condition monitoring","authors":"M. Káš, F. F. Wamba","doi":"10.1784/insi.2022.64.8.453","DOIUrl":"https://doi.org/10.1784/insi.2022.64.8.453","url":null,"abstract":"The impact of an anomaly is domain-dependent. In a dataset of network activities, an anomaly can imply an intrusion attack. Other objectives of anomaly detection are industrial damage detection, data leak prevention, identifying security vulnerabilities or military surveillance. Anomalies\u0000 are observations or a sequence of observations in which the distribution deviates remarkably from the general distribution of the whole dataset. A large majority of the dataset consists of normal (healthy) data points. The anomalies form only a very small part of the dataset. Anomaly detection\u0000 is the technique used to find these observations and its methods are specific to the type of data. While there is a wide spectrum of anomaly detection approaches available today, it becomes increasingly difficult to keep track of all the techniques. In fact, it is not clear which of the three\u0000 categories of detection methods, ie statistical approaches, machine learning approaches or deep learning approaches, is more appropriate to detect anomalies in time-series data, which are mainly used in industry. A typical industrial device has multi-dimensional characteristics. It is possible\u0000 to measure voltage, current, active power, vibrations, rotational speed, temperature, pressure difference, etc, on such a device. Early detection of the anomalous behaviour of industrial devices can help reduce or prevent serious damage, which could lead to significant financial loss. This\u0000 paper presents a summary of the methods used to detect anomalies in condition monitoring applications.","PeriodicalId":344397,"journal":{"name":"Insight - Non-Destructive Testing and Condition Monitoring","volume":"65 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134275310","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：