Pub Date : 2024-01-04DOI: 10.1088/1361-6501/ad197c
Yikun Liu, Song Fu, Lin Lin, Sihao Zhang, Shiwei Suo, Jianjun Xi
Conditional variational autoencoder (CVAE) has the potential for few-sample fault diagnosis of mechanical systems. Nevertheless, the scarcity of faulty samples leads the augmented samples generated using CVAE suffer from limited diversity. To address the issue, a novel CVAE variant namely CVAE with distribution augmentation (DECVAE) is developed, to generate a set of high-quality augmented samples that are different but share very similar characteristics and categories with the corresponding real samples. First, DECVAE add a new sample distribution distance loss into the optimization objective of traditional CVAE. Amplifying this loss in training process can make the augmented samples cover a larger space, thereby improving diversity. Second, DECVAE introduces an auxiliary classifier into traditional CVAE to enhance the sensitivity to category information, keeping the augmented samples class invariance. Furthermore, to ensure that the information of edge-distributed samples can be fully learned and make augmented samples representative and authentic, a novel multi-model independent fine-tuning strategy is designed to train the DECVAE, which utilizes multiple independent models to fairly focus on all samples of the minority class during DECVAE training. Finally, the effectiveness of the developed DECVAE in few-shot fault diagnosis of mechanical systems is verified on a series of comparative experiments.
{"title":"DECVAE: Data augmentation via conditional variational auto-encoder with distribution enhancement for few-shot fault diagnosis of mechanical system","authors":"Yikun Liu, Song Fu, Lin Lin, Sihao Zhang, Shiwei Suo, Jianjun Xi","doi":"10.1088/1361-6501/ad197c","DOIUrl":"https://doi.org/10.1088/1361-6501/ad197c","url":null,"abstract":"Conditional variational autoencoder (CVAE) has the potential for few-sample fault diagnosis of mechanical systems. Nevertheless, the scarcity of faulty samples leads the augmented samples generated using CVAE suffer from limited diversity. To address the issue, a novel CVAE variant namely CVAE with distribution augmentation (DECVAE) is developed, to generate a set of high-quality augmented samples that are different but share very similar characteristics and categories with the corresponding real samples. First, DECVAE add a new sample distribution distance loss into the optimization objective of traditional CVAE. Amplifying this loss in training process can make the augmented samples cover a larger space, thereby improving diversity. Second, DECVAE introduces an auxiliary classifier into traditional CVAE to enhance the sensitivity to category information, keeping the augmented samples class invariance. Furthermore, to ensure that the information of edge-distributed samples can be fully learned and make augmented samples representative and authentic, a novel multi-model independent fine-tuning strategy is designed to train the DECVAE, which utilizes multiple independent models to fairly focus on all samples of the minority class during DECVAE training. Finally, the effectiveness of the developed DECVAE in few-shot fault diagnosis of mechanical systems is verified on a series of comparative experiments.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"2 1","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139385905","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-04DOI: 10.1088/1361-6501/ad1917
Yikai Zong, Shujing Su, Yuhong Gao, Lili Zhang
This paper proposes an improved attitude estimation algorithm based on the extended Kalman filter (EKF), and it is applied to suppress the accuracy reduction in attitude estimation caused by fusing magnetometer data under large angular motion. In the proposed attitude estimation structure, the approximate variance of the estimated horizontal northbound magnetic vector is used to dynamically adjust the participation of magnetometer data in attitude estimation, as the approximate variance increases significantly under large angular motion and fusing magnetometer data will reduce estimation accuracy. A three-axis position-velocity controlled turntable is used to conduct rocking experiments for validating the proposed attitude estimation algorithm. The results show a significant improvement in yaw angle estimation accuracy with the proposed attitude estimation algorithm and correspondingly enhance the distribution of pitch and roll angle errors.
{"title":"An improved attitude estimation algorithm for suppressing magnetic vector disturbance based on extended Kalman filter","authors":"Yikai Zong, Shujing Su, Yuhong Gao, Lili Zhang","doi":"10.1088/1361-6501/ad1917","DOIUrl":"https://doi.org/10.1088/1361-6501/ad1917","url":null,"abstract":"This paper proposes an improved attitude estimation algorithm based on the extended Kalman filter (EKF), and it is applied to suppress the accuracy reduction in attitude estimation caused by fusing magnetometer data under large angular motion. In the proposed attitude estimation structure, the approximate variance of the estimated horizontal northbound magnetic vector is used to dynamically adjust the participation of magnetometer data in attitude estimation, as the approximate variance increases significantly under large angular motion and fusing magnetometer data will reduce estimation accuracy. A three-axis position-velocity controlled turntable is used to conduct rocking experiments for validating the proposed attitude estimation algorithm. The results show a significant improvement in yaw angle estimation accuracy with the proposed attitude estimation algorithm and correspondingly enhance the distribution of pitch and roll angle errors.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"18 6","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139385712","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-04DOI: 10.1088/1361-6501/ad1914
Yuzhang Wang, Kanru Cheng, Fan Liu, Jiao Li, Kunyu Zhang
Correct and reliable measurement data are crucial for state monitoring, safe operations, health assessment, and life prediction of integrated energy systems (IESs). Sensors are often installed in harsh environments and prone to all kinds of faults; therefore, it is necessary to diagnose sensor faults. A diagnostic method for sensor faults based on gradient histogram distribution (GHD) combined with light gradient boosting machine (LightGBM) is presented in this paper. This proposed method effectively utilizes the coupling information between the relevant parameters. The GHD efficiently extracted the time-domain characteristics of sensor faults and reduced the dimension of eigenvectors. This is beneficial to increasing the diagnostic speed. The kernel density estimation distributions of the gradient and eigenvectors for the sensor with strong correlation are similar, but that for the sensor with weak correlation are completely different. A LightGBM classifier trained based on the feature vectors was utilized to diagnose and classify the sensor faults. The diagnosis accuracy and the diagnosis time of this developed method were examined using the multiple-condition practical operation data of gas turbines in the IES. The experiment results demonstrate that the diagnostic accuracy of five sensor faults using this developed method is all above 90%. The diagnostic time is about 0.47–1.34 s, and is less than 2 s for the gradual faults.
{"title":"Study of the fault diagnosis method for gas turbine sensors based on inter-parameter coupling information","authors":"Yuzhang Wang, Kanru Cheng, Fan Liu, Jiao Li, Kunyu Zhang","doi":"10.1088/1361-6501/ad1914","DOIUrl":"https://doi.org/10.1088/1361-6501/ad1914","url":null,"abstract":"Correct and reliable measurement data are crucial for state monitoring, safe operations, health assessment, and life prediction of integrated energy systems (IESs). Sensors are often installed in harsh environments and prone to all kinds of faults; therefore, it is necessary to diagnose sensor faults. A diagnostic method for sensor faults based on gradient histogram distribution (GHD) combined with light gradient boosting machine (LightGBM) is presented in this paper. This proposed method effectively utilizes the coupling information between the relevant parameters. The GHD efficiently extracted the time-domain characteristics of sensor faults and reduced the dimension of eigenvectors. This is beneficial to increasing the diagnostic speed. The kernel density estimation distributions of the gradient and eigenvectors for the sensor with strong correlation are similar, but that for the sensor with weak correlation are completely different. A LightGBM classifier trained based on the feature vectors was utilized to diagnose and classify the sensor faults. The diagnosis accuracy and the diagnosis time of this developed method were examined using the multiple-condition practical operation data of gas turbines in the IES. The experiment results demonstrate that the diagnostic accuracy of five sensor faults using this developed method is all above 90%. The diagnostic time is about 0.47–1.34 s, and is less than 2 s for the gradual faults.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"23 9","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139450777","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-04DOI: 10.1088/1361-6501/ad1b33
Kai Wei, Min Li, Tianhe Xu, Dixing Wang, Yali Shi, Honglei Yang, Xiaoji Dai
The precise orbit determination (POD) of scientific low Earth orbit (LEO) satellites is a prerequisite for the successful implementation of scientific missions. In recent years, global navigation satellite systems have become the main means of determining the orbits of LEO satellites. The global navigation satellite system receiver onboard the Tianjin University No. 1 (TJU-01) satellite receives both GPS and BDS-2/3 signals, with the addition of BDS-2/3 observations playing an important role in improving the POD of LEO satellites. This study comprehensively analyzes the spaceborne GPS/BDS data quality, including BDS-2/3 and GPS code multipath errors. Appreciable code multipath errors are found for the B1I signal of BDS-2 medium Earth orbit (MEO) satellites at elevations higher than 40°, whereas slight near-field relevant multipath errors of both frequencies are found for GPS and BDS-3 MEO satellites. The GPS and BDS-2/3 code multipath errors are estimated through elevation/azimuth-relevant piece-wise modeling and applied in the POD calculations. Several schemes, namely GPS-based, BDS-based, BDS-based without geo-synchronous (GEO) satellites, and GPS/BDS combined schemes, are designed to evaluate the POD performance. Fourteen days of data are calculated and the average three-dimensional (3D) orbital root mean square (RMS) of orbit overlapping differences obtained from GPS-based and BDS-based POD (without GEO satellites) solutions are 37.4 and 27.1 mm, respectively. The BDS-based solutions are obviously better than the GPS-based solutions, mainly owing to better data availability. The GPS/BDS combined solutions have the best accuracy, with a 3D RMS value of 20.6 mm. In addition, when BDS GEO satellites are included, the 3D RMS of the overlapping orbit differences reduces to 32.9 and 27.4 mm for BDS-based and GPS/BDS combined solutions, respectively. Double-difference (DD) and single-difference (SD) integer ambiguity resolution (IAR) are adopted to further improve the POD performance. The fixed orbit of the TJU-01 satellite is solved through DD IAR and SD IAR, and the contribution of the TJU-01 satellite to ambiguity fixing is analyzed. Relative to the float solution, the improvements made using the two ambiguity fixing approaches are equivalent, both being approximately 13%. The importance of this research is not only the precise determination of the orbit of TJU-01 for occultation service but also the demonstration of the contribution of BDS observations to the performance of the POD of LEO satellites.
低地球轨道（LEO）科学卫星的精确轨道测定（POD）是成功执行科学飞行任务的先决条件。近年来，全球卫星导航系统已成为确定低地轨道卫星轨道的主要手段。天津大学一号卫星（TJU-01）上搭载的全球导航卫星系统接收机同时接收 GPS 和 BDS-2/3 信号，其中 BDS-2/3 观测数据的加入对提高低地轨道卫星的 POD 起到了重要作用。这项研究全面分析了空间 GPS/BDS 数据质量，包括 BDS-2/3 和 GPS 代码多径误差。发现 BDS-2 中地球轨道（MEO）卫星的 B1I 信号在海拔高于 40° 时存在明显的代码多径误差，而 GPS 和 BDS-3 中地球轨道卫星的两个频率都存在轻微的近场相关多径误差。全球定位系统和 BDS-2/3 代码多径误差是通过与海拔/方位相关的片断建模估算出来的，并应用于 POD 计算。为评估 POD 性能，设计了几种方案，即基于 GPS 的方案、基于 BDS 的方案、基于 BDS（不含地球同步（GEO）卫星）的方案和基于 GPS/BDS 的组合方案。计算了 14 天的数据，基于 GPS 的 POD 方案和基于 BDS 的 POD 方案（不含地球同步轨道卫星）得到的轨道重叠差的平均三维（3D）轨道均方根（RMS）分别为 37.4 毫米和 27.1 毫米。基于 BDS 的解法明显优于基于 GPS 的解法，这主要是由于数据可用性更好。GPS/BDS 组合方案的精度最高，三维均方根值为 20.6 毫米。此外，当包括 BDS 地球同步轨道卫星时，基于 BDS 和 GPS/BDS 组合解法的重叠轨道差的三维均方根值分别降低到 32.9 毫米和 27.4 毫米。采用双差分（DD）和单差分（SD）整数模糊分辨率（IAR）进一步提高了 POD 性能。通过 DD IAR 和 SD IAR 求解了 TJU-01 卫星的固定轨道，并分析了 TJU-01 卫星对模糊固定的贡献。与浮动解法相比，两种模糊性修正方法的改进效果相当，都约为 13%。这项研究的重要性不仅在于精确确定了用于掩星服务的 TJU-01 的轨道，还在于证明了 BDS 观测对低地球轨道卫星 POD 性能的贡献。
{"title":"Contribution of BDS-3 observations to the precise orbit determination of LEO satellites: A case study of TJU-01","authors":"Kai Wei, Min Li, Tianhe Xu, Dixing Wang, Yali Shi, Honglei Yang, Xiaoji Dai","doi":"10.1088/1361-6501/ad1b33","DOIUrl":"https://doi.org/10.1088/1361-6501/ad1b33","url":null,"abstract":"\u0000 The precise orbit determination (POD) of scientific low Earth orbit (LEO) satellites is a prerequisite for the successful implementation of scientific missions. In recent years, global navigation satellite systems have become the main means of determining the orbits of LEO satellites. The global navigation satellite system receiver onboard the Tianjin University No. 1 (TJU-01) satellite receives both GPS and BDS-2/3 signals, with the addition of BDS-2/3 observations playing an important role in improving the POD of LEO satellites. This study comprehensively analyzes the spaceborne GPS/BDS data quality, including BDS-2/3 and GPS code multipath errors. Appreciable code multipath errors are found for the B1I signal of BDS-2 medium Earth orbit (MEO) satellites at elevations higher than 40°, whereas slight near-field relevant multipath errors of both frequencies are found for GPS and BDS-3 MEO satellites. The GPS and BDS-2/3 code multipath errors are estimated through elevation/azimuth-relevant piece-wise modeling and applied in the POD calculations. Several schemes, namely GPS-based, BDS-based, BDS-based without geo-synchronous (GEO) satellites, and GPS/BDS combined schemes, are designed to evaluate the POD performance. Fourteen days of data are calculated and the average three-dimensional (3D) orbital root mean square (RMS) of orbit overlapping differences obtained from GPS-based and BDS-based POD (without GEO satellites) solutions are 37.4 and 27.1 mm, respectively. The BDS-based solutions are obviously better than the GPS-based solutions, mainly owing to better data availability. The GPS/BDS combined solutions have the best accuracy, with a 3D RMS value of 20.6 mm. In addition, when BDS GEO satellites are included, the 3D RMS of the overlapping orbit differences reduces to 32.9 and 27.4 mm for BDS-based and GPS/BDS combined solutions, respectively. Double-difference (DD) and single-difference (SD) integer ambiguity resolution (IAR) are adopted to further improve the POD performance. The fixed orbit of the TJU-01 satellite is solved through DD IAR and SD IAR, and the contribution of the TJU-01 satellite to ambiguity fixing is analyzed. Relative to the float solution, the improvements made using the two ambiguity fixing approaches are equivalent, both being approximately 13%. The importance of this research is not only the precise determination of the orbit of TJU-01 for occultation service but also the demonstration of the contribution of BDS observations to the performance of the POD of LEO satellites.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"64 4","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139385538","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-04DOI: 10.1088/1361-6501/ad191c
Xiaoyu Wen, Juxiang Zhou, Jianhou Gan, Sen Luo
Driven by advancements in deep learning technologies, substantial progress has been achieved in the field of facial expression recognition over the past decade, while challenges remain brought about by occlusions, pose variations and subtle expression differences in unconstrained (wild) scenarios. Therefore, a novel multiscale feature extraction method is proposed in this paper, that leverages convolutional neural networks to simultaneously extract deep semantic features and shallow geometric features. Through the mechanism of channel-wise self-attention, prominent features are further extracted and compressed, preserving advantageous features for distinction and thereby reducing the impact of occlusions and pose variations on expression recognition. Meanwhile, inspired by the large cosine margin concept used in face recognition, a center cosine loss function is proposed to avoid the misclassification caused by the underlying interclass similarity and substantial intra-class feature variations in the task of expression recognition. This function is designed to enhance the classification performance of the network through making the distribution of samples within the same class more compact and that between different classes sparser. The proposed method is benchmarked against several advanced baseline models on three mainstream wild datasets and two datasets that present realistic occlusion and pose variation challenges. Accuracies of 89.63%, 61.82%, and 91.15% are achieved on RAF-DB, AffectNet and FERPlus, respectively, demonstrating the greater robustness and reliability of this method compared to the state-of-the-art alternatives in the real world.
{"title":"A discriminative multiscale feature extraction network for facial expression recognition in the wild","authors":"Xiaoyu Wen, Juxiang Zhou, Jianhou Gan, Sen Luo","doi":"10.1088/1361-6501/ad191c","DOIUrl":"https://doi.org/10.1088/1361-6501/ad191c","url":null,"abstract":"Driven by advancements in deep learning technologies, substantial progress has been achieved in the field of facial expression recognition over the past decade, while challenges remain brought about by occlusions, pose variations and subtle expression differences in unconstrained (wild) scenarios. Therefore, a novel multiscale feature extraction method is proposed in this paper, that leverages convolutional neural networks to simultaneously extract deep semantic features and shallow geometric features. Through the mechanism of channel-wise self-attention, prominent features are further extracted and compressed, preserving advantageous features for distinction and thereby reducing the impact of occlusions and pose variations on expression recognition. Meanwhile, inspired by the large cosine margin concept used in face recognition, a center cosine loss function is proposed to avoid the misclassification caused by the underlying interclass similarity and substantial intra-class feature variations in the task of expression recognition. This function is designed to enhance the classification performance of the network through making the distribution of samples within the same class more compact and that between different classes sparser. The proposed method is benchmarked against several advanced baseline models on three mainstream wild datasets and two datasets that present realistic occlusion and pose variation challenges. Accuracies of 89.63%, 61.82%, and 91.15% are achieved on RAF-DB, AffectNet and FERPlus, respectively, demonstrating the greater robustness and reliability of this method compared to the state-of-the-art alternatives in the real world.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"42 11","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139384971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-03DOI: 10.1088/1361-6501/ad1a87
Zhichao Li, Mingxue Shen, Li Tian, Xue-feng Yan
Modern industrial processes are increasingly complex, where multiple characteristics usually coexist in process data. Therefore, traditional monitoring methods based on a single model may ignore other data characteristics and obtain poor monitoring performance. Aiming at this problem, a novel monitoring method based on multi-model information extraction and fusion is proposed in this paper. Firstly, several methods are used to extract different characteristics from process data. For example, principal component analysis, independent component analysis and slow features analysis can be used to extract Gaussian, non-Gaussian and dynamic characteristics respectively. Secondly, features extracted from multiple models are combined into new potential features. Then, Lasso regression models between potential features and process variables are established. In this way, not only are multiple characteristics in process data considered during the reconstruction, but key potential features (KPFs) can be selected for each process variable. The KPFs for each process variable can form a monitoring subspace to enhance the sensitivity for fault detection. Furthermore, cluster analysis is used to reduce the redundancy of monitoring subspaces based on the similarity of each subspace. Process monitoring can be achieved by fusing the monitoring results of finally determined multiple subspaces and residual space. Case studies on three simulation processes and a real industrial process demonstrate the effectiveness and better performance.
{"title":"A novel monitoring method based on multi-model information extraction and fusion","authors":"Zhichao Li, Mingxue Shen, Li Tian, Xue-feng Yan","doi":"10.1088/1361-6501/ad1a87","DOIUrl":"https://doi.org/10.1088/1361-6501/ad1a87","url":null,"abstract":"\u0000 Modern industrial processes are increasingly complex, where multiple characteristics usually coexist in process data. Therefore, traditional monitoring methods based on a single model may ignore other data characteristics and obtain poor monitoring performance. Aiming at this problem, a novel monitoring method based on multi-model information extraction and fusion is proposed in this paper. Firstly, several methods are used to extract different characteristics from process data. For example, principal component analysis, independent component analysis and slow features analysis can be used to extract Gaussian, non-Gaussian and dynamic characteristics respectively. Secondly, features extracted from multiple models are combined into new potential features. Then, Lasso regression models between potential features and process variables are established. In this way, not only are multiple characteristics in process data considered during the reconstruction, but key potential features (KPFs) can be selected for each process variable. The KPFs for each process variable can form a monitoring subspace to enhance the sensitivity for fault detection. Furthermore, cluster analysis is used to reduce the redundancy of monitoring subspaces based on the similarity of each subspace. Process monitoring can be achieved by fusing the monitoring results of finally determined multiple subspaces and residual space. Case studies on three simulation processes and a real industrial process demonstrate the effectiveness and better performance.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"20 4","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139389428","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-03DOI: 10.1088/1361-6501/ad1a68
C. Gao, Xiaolin Zhao, Shuqi Zhang, Ke Wang, Bo Qi, Chengrong Li
The design of insulating structures for transformers under impulse voltage relies predominantly on simulation software due to the absence of experimental validation. This underscores the pressing need for comprehensive research into the spatial electric field and charge properties of oil-paper/pressboard insulation systems. In response to this imperative, a suite of specialized instruments leveraging the Kerr electro-optic effect to meticulously measure the spatial electric field within oil-pressboard structures under impulse voltage was established. As the precision of measurements hinges upon a multitude of influencing factors, this study embarks on a multifaceted examination, centering its focus on four pivotal dimensions: incident laser beam angle, electrical noise, temperature and non-ideal optical elements. A quantitative calculation method for electric field measurement errors was presented, and on the basis of which, suppression methods are proposed for the error sources having the largest impacts on the experimental results. Finally, the overall measurement uncertainty of the device is systematically evaluated.
{"title":"Influencing factors and uncertainty analysis for Kerr electro-optic effect based electric field measurements in transformer oil under impulse voltage","authors":"C. Gao, Xiaolin Zhao, Shuqi Zhang, Ke Wang, Bo Qi, Chengrong Li","doi":"10.1088/1361-6501/ad1a68","DOIUrl":"https://doi.org/10.1088/1361-6501/ad1a68","url":null,"abstract":"\u0000 The design of insulating structures for transformers under impulse voltage relies predominantly on simulation software due to the absence of experimental validation. This underscores the pressing need for comprehensive research into the spatial electric field and charge properties of oil-paper/pressboard insulation systems. In response to this imperative, a suite of specialized instruments leveraging the Kerr electro-optic effect to meticulously measure the spatial electric field within oil-pressboard structures under impulse voltage was established. As the precision of measurements hinges upon a multitude of influencing factors, this study embarks on a multifaceted examination, centering its focus on four pivotal dimensions: incident laser beam angle, electrical noise, temperature and non-ideal optical elements. A quantitative calculation method for electric field measurement errors was presented, and on the basis of which, suppression methods are proposed for the error sources having the largest impacts on the experimental results. Finally, the overall measurement uncertainty of the device is systematically evaluated.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"29 24","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139388741","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-03DOI: 10.1088/1361-6501/ad1a69
Junling Zhang, Lixiang Duan, Ke Li, Shilong Luo
he working environment of reciprocating machinery is complex, characterized by nonlinear and non-stationary signals. Deep learning can solve the above problems, but it has its own problems such as complex model and large amount of parameters. Additionally, privacy considerations among enterprises prevent data sharing, leading to the emergence of "data islands" and inadequate training of deep learning models. Based on the above analysis, this paper proposes a reciprocating mechanical feature extraction method based on an improved federated lightweight network. A lightweight network SqueezeNet model is used to solve the problems such as long training time of deep learning. By establishing a federated learning framework, the reciprocating mechanical data can be collectively diagnosed across various enterprises, thereby addressing the problem of limited model training caused by insufficient data. Furthermore, to enhance the accuracy of network training and diagnosis, modifications are made to the SqueezeNet network to reduce the number of model parameters while increasing the number and variety of feature extractions. Experimental results demonstrate that when the number of 1×1 and 3×3 channels is 1 to 7, the fault diagnosis accuracy is the highest, about 97.96%, which enriches the categories of feature extraction. The number of parameters in In-SqueezeNet is 56% of that in SqueezeNet network model, and the training time is reduced by nearly 15%. The fault diagnosis accuracy is increased from 95.1% to 97.3%, and the diversity of extracted features is increased. Compared with other network models such as ResNet, the improved lightweight federated learning network has a fault diagnosis accuracy of 96.6%, an improvement of 10.6%. At the same time, the training time was reduced to 1982s, a reduction of about 41.5%. The validity of the proposed model is further verified.
{"title":"Improved lightweight federated learning network for fault feature extraction of reciprocating machinery","authors":"Junling Zhang, Lixiang Duan, Ke Li, Shilong Luo","doi":"10.1088/1361-6501/ad1a69","DOIUrl":"https://doi.org/10.1088/1361-6501/ad1a69","url":null,"abstract":"\u0000 he working environment of reciprocating machinery is complex, characterized by nonlinear and non-stationary signals. Deep learning can solve the above problems, but it has its own problems such as complex model and large amount of parameters. Additionally, privacy considerations among enterprises prevent data sharing, leading to the emergence of \"data islands\" and inadequate training of deep learning models. Based on the above analysis, this paper proposes a reciprocating mechanical feature extraction method based on an improved federated lightweight network. A lightweight network SqueezeNet model is used to solve the problems such as long training time of deep learning. By establishing a federated learning framework, the reciprocating mechanical data can be collectively diagnosed across various enterprises, thereby addressing the problem of limited model training caused by insufficient data. Furthermore, to enhance the accuracy of network training and diagnosis, modifications are made to the SqueezeNet network to reduce the number of model parameters while increasing the number and variety of feature extractions. Experimental results demonstrate that when the number of 1×1 and 3×3 channels is 1 to 7, the fault diagnosis accuracy is the highest, about 97.96%, which enriches the categories of feature extraction. The number of parameters in In-SqueezeNet is 56% of that in SqueezeNet network model, and the training time is reduced by nearly 15%. The fault diagnosis accuracy is increased from 95.1% to 97.3%, and the diversity of extracted features is increased. Compared with other network models such as ResNet, the improved lightweight federated learning network has a fault diagnosis accuracy of 96.6%, an improvement of 10.6%. At the same time, the training time was reduced to 1982s, a reduction of about 41.5%. The validity of the proposed model is further verified.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"44 21","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139451990","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper proposes a normal tracking differential confocal measurement method for the inside and outside surface profiles, shell thickness uniformity, and central asymmetry of inside and outside surfaces of the hemispherical shell resonator (HSR). A differential confocal technique with high-transmittance focusing ability is used to measure a single point on the inside and outside surfaces of the HSR. The normal alignment measurement technique is used to accurately measure the inside and outside surfaces and shell thickness of the HSR with a common reference in one measurement process. The HSR is step-rotated to synchronously collect information on the inside and outside surfaces, and using the differential confocal sensor to measure the different normal-section profiles. The experimental results indicate successful measurement of HSR central asymmetry. The repeated measurement accuracy for the inside and outside surface profiles and thickness uniformity is better than 30 nm.
{"title":"A normal tracking differential confocal measurement method for multiple geometric parameters of hemispherical shell resonator with a common reference","authors":"Yuhan Liu, Xiaocheng Zhang, Yuan Fu, Yun Wang, Zhuxian Yao, Weiqian Zhao","doi":"10.1088/1361-6501/ad1a85","DOIUrl":"https://doi.org/10.1088/1361-6501/ad1a85","url":null,"abstract":"\u0000 This paper proposes a normal tracking differential confocal measurement method for the inside and outside surface profiles, shell thickness uniformity, and central asymmetry of inside and outside surfaces of the hemispherical shell resonator (HSR). A differential confocal technique with high-transmittance focusing ability is used to measure a single point on the inside and outside surfaces of the HSR. The normal alignment measurement technique is used to accurately measure the inside and outside surfaces and shell thickness of the HSR with a common reference in one measurement process. The HSR is step-rotated to synchronously collect information on the inside and outside surfaces, and using the differential confocal sensor to measure the different normal-section profiles. The experimental results indicate successful measurement of HSR central asymmetry. The repeated measurement accuracy for the inside and outside surface profiles and thickness uniformity is better than 30 nm.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"59 3","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139451094","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-03DOI: 10.1088/1361-6501/ad1a86
Beiyan He, Chunli Zhu, Zhongxiang Li, Chun Hu, Dezhi Zheng
Sensors equipped on the high-speed train provide large amounts of data which contributes to its state monitoring. However, it is challenging to distinguish whether the fault originates from the mechanical component or the sensors themselves. The main difficulties lie in the biased amount of normal and fault data as well as the deficiency of multi-source data’s inherent correlation. In this paper, we propose a Bayesian Convolutional neural networks (CNN)-based fusion framework to enhance the ability to identify sensor errors. The framework utilizes wavelet time-frequency maps to extract abnormal features, employs a Bayesian CNN to obtain spatial features from a single sensor, integrates multi-source features via Bidirectional Long Short-Term Memory Network (Bi-LSTM) and enhances the acquired spatial and temporal features using an attention mechanism. The enhanced information finally generated leads to precise identification of the sensor faults. The proposed feature-level fusion framework and the associated attention mechanism facilitate discovering the inherent correlation and filtering of irrelevant information. Results indicate that our proposed method achieves 95.4% in terms of accuracy, which outperforms methods relying on feature extraction with single-source sensors by 7.8%.
{"title":"A Bayesian CNN-based Fusion Framework of Sensor Fault Diagnosis","authors":"Beiyan He, Chunli Zhu, Zhongxiang Li, Chun Hu, Dezhi Zheng","doi":"10.1088/1361-6501/ad1a86","DOIUrl":"https://doi.org/10.1088/1361-6501/ad1a86","url":null,"abstract":"\u0000 Sensors equipped on the high-speed train provide large amounts of data which contributes to its state monitoring. However, it is challenging to distinguish whether the fault originates from the mechanical component or the sensors themselves. The main difficulties lie in the biased amount of normal and fault data as well as the deficiency of multi-source data’s inherent correlation. In this paper, we propose a Bayesian Convolutional neural networks (CNN)-based fusion framework to enhance the ability to identify sensor errors. The framework utilizes wavelet time-frequency maps to extract abnormal features, employs a Bayesian CNN to obtain spatial features from a single sensor, integrates multi-source features via Bidirectional Long Short-Term Memory Network (Bi-LSTM) and enhances the acquired spatial and temporal features using an attention mechanism. The enhanced information finally generated leads to precise identification of the sensor faults. The proposed feature-level fusion framework and the associated attention mechanism facilitate discovering the inherent correlation and filtering of irrelevant information. Results indicate that our proposed method achieves 95.4% in terms of accuracy, which outperforms methods relying on feature extraction with single-source sensors by 7.8%.","PeriodicalId":18526,"journal":{"name":"Measurement Science and Technology","volume":"97 23","pages":""},"PeriodicalIF":2.4,"publicationDate":"2024-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139387927","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","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：