As a type of reciprocating machine, the reciprocating compressor has a compact structure and many excitation sources.Once the small end bearing of the connecting rod is worn, it is easy to cause the sintering of the b...As a type of reciprocating machine, the reciprocating compressor has a compact structure and many excitation sources.Once the small end bearing of the connecting rod is worn, it is easy to cause the sintering of the bearing and the abnormal vibration of the body.Based on the characteristics of poor lubrication state and complex force of connecting rod small head bearing, a mixed lubrication model considering oil groove feed was established, and the dynamic simulation of the reciprocating compressor model with lubricated bearings was carried out;considering different speeds and gas load conditions, the law of the impact of the eigenvalues changing with working conditions was explored.The fault simulation experiment was carried out by selecting representative working conditions, which verified the correctness of the simulation method.The study found that two contact collisions between the pin and the bearing bush occurred in one cycle, the collision impact was more severe under the wear fault, and the existence of the gap made the dynamic response more sensitive to the change of working conditions.This research provides ideas for the location and feature extraction of fault symptom signal angular segments in the process of complex measured signal processing.展开更多
The change of working conditions not only makes the data distribution inconsistent,but also increases the diagnosis difficulty of fuzzy samples at the fault boundary.The traditional distance-based deep metric learning...The change of working conditions not only makes the data distribution inconsistent,but also increases the diagnosis difficulty of fuzzy samples at the fault boundary.The traditional distance-based deep metric learning cannot effectively classify the fuzzy samples at the fault boundary.In the traditional transfer learning models,the maximum mean discrepancy(MMD)and joint maximum mean discrepancy only increase the transferability of same-class samples,and neglect the discriminability of different-class samples across different domains.The discriminative joint probability MMD(DJP-MMD)increases the transferability of same-class samples and the discriminability of different-class samples across different domains,but it only considers the global transferability of all fault classes,ignoring the different transferability of each same fault class.Therefore,a Yu norm-based deep transfer metric learning based on weighted DJP-MMD is proposed to further improve the diagnosis accuracy of bearings under variable working conditions.The deep transfer metric learning model adopts the Yu norm-based similarity instead of the distance-based similarity to effectively classify the data samples,especially those at the fault boundary,and uses the weighted DJP-MMD to measure the data distribution discrepancy between the source and target domains to increase the transferability of each same-class samples and discriminability of different-class samples across different domains.Through the fault diagnosis analysis on bearings under variable working conditions,the diagnosis results demonstrate that the proposed deep transfer metric learning model can diagnose bearing faults with higher accuracy,stronger generalization and anti-noise capabilities compared with other fault diagnosis methods based on transfer learning.展开更多
To address the issue of scarce labeled samples and operational condition variations that degrade the accuracy of fault diagnosis models in variable-condition gearbox fault diagnosis,this paper proposes a semi-supervis...To address the issue of scarce labeled samples and operational condition variations that degrade the accuracy of fault diagnosis models in variable-condition gearbox fault diagnosis,this paper proposes a semi-supervised masked contrastive learning and domain adaptation(SSMCL-DA)method for gearbox fault diagnosis under variable conditions.Initially,during the unsupervised pre-training phase,a dual signal augmentation strategy is devised,which simultaneously applies random masking in the time domain and random scaling in the frequency domain to unlabeled samples,thereby constructing more challenging positive sample pairs to guide the encoder in learning intrinsic features robust to condition variations.Subsequently,a ConvNeXt-Transformer hybrid architecture is employed,integrating the superior local detail modeling capacity of ConvNeXt with the robust global perception capability of Transformer to enhance feature extraction in complex scenarios.Thereafter,a contrastive learning model is constructed with the optimization objective of maximizing feature similarity across different masked instances of the same sample,enabling the extraction of consistent features from multiple masked perspectives and reducing reliance on labeled data.In the final supervised fine-tuning phase,a multi-scale attention mechanism is incorporated for feature rectification,and a domain adaptation module combining Local Maximum Mean Discrepancy(LMMD)with adversarial learning is proposed.This module embodies a dual mechanism:LMMD facilitates fine-grained class-conditional alignment,compelling features of identical fault classes to converge across varying conditions,while the domain discriminator utilizes adversarial training to guide the feature extractor toward learning domain-invariant features.Working in concert,they markedly diminish feature distribution discrepancies induced by changes in load,rotational speed,and other factors,thereby boosting the model’s adaptability to cross-condition scenarios.Experimental evaluations on the WT planetary gearbox dataset and the Case Western Reserve University(CWRU)bearing dataset demonstrate that the SSMCL-DA model effectively identifies multiple fault classes in gearboxes,with diagnostic performance substantially surpassing that of conventional methods.Under cross-condition scenarios,the model attains fault diagnosis accuracies of 99.21%for the WT planetary gearbox and 99.86%for the bearings,respectively.Furthermore,the model exhibits stable generalization capability in cross-device settings.展开更多
基金Supported by the National Natural Science Foundation of China (No.52101343)。
文摘As a type of reciprocating machine, the reciprocating compressor has a compact structure and many excitation sources.Once the small end bearing of the connecting rod is worn, it is easy to cause the sintering of the bearing and the abnormal vibration of the body.Based on the characteristics of poor lubrication state and complex force of connecting rod small head bearing, a mixed lubrication model considering oil groove feed was established, and the dynamic simulation of the reciprocating compressor model with lubricated bearings was carried out;considering different speeds and gas load conditions, the law of the impact of the eigenvalues changing with working conditions was explored.The fault simulation experiment was carried out by selecting representative working conditions, which verified the correctness of the simulation method.The study found that two contact collisions between the pin and the bearing bush occurred in one cycle, the collision impact was more severe under the wear fault, and the existence of the gap made the dynamic response more sensitive to the change of working conditions.This research provides ideas for the location and feature extraction of fault symptom signal angular segments in the process of complex measured signal processing.
基金funded by the National Natural Science Foundation of China(Grant No.51775391).
文摘The change of working conditions not only makes the data distribution inconsistent,but also increases the diagnosis difficulty of fuzzy samples at the fault boundary.The traditional distance-based deep metric learning cannot effectively classify the fuzzy samples at the fault boundary.In the traditional transfer learning models,the maximum mean discrepancy(MMD)and joint maximum mean discrepancy only increase the transferability of same-class samples,and neglect the discriminability of different-class samples across different domains.The discriminative joint probability MMD(DJP-MMD)increases the transferability of same-class samples and the discriminability of different-class samples across different domains,but it only considers the global transferability of all fault classes,ignoring the different transferability of each same fault class.Therefore,a Yu norm-based deep transfer metric learning based on weighted DJP-MMD is proposed to further improve the diagnosis accuracy of bearings under variable working conditions.The deep transfer metric learning model adopts the Yu norm-based similarity instead of the distance-based similarity to effectively classify the data samples,especially those at the fault boundary,and uses the weighted DJP-MMD to measure the data distribution discrepancy between the source and target domains to increase the transferability of each same-class samples and discriminability of different-class samples across different domains.Through the fault diagnosis analysis on bearings under variable working conditions,the diagnosis results demonstrate that the proposed deep transfer metric learning model can diagnose bearing faults with higher accuracy,stronger generalization and anti-noise capabilities compared with other fault diagnosis methods based on transfer learning.
基金supported by the National Natural Science Foundation of China Funded Project(Project Name:Research on Robust Adaptive Allocation Mechanism of Human Machine Co-Driving System Based on NMS Features,Project Approval Number:52172381).
文摘To address the issue of scarce labeled samples and operational condition variations that degrade the accuracy of fault diagnosis models in variable-condition gearbox fault diagnosis,this paper proposes a semi-supervised masked contrastive learning and domain adaptation(SSMCL-DA)method for gearbox fault diagnosis under variable conditions.Initially,during the unsupervised pre-training phase,a dual signal augmentation strategy is devised,which simultaneously applies random masking in the time domain and random scaling in the frequency domain to unlabeled samples,thereby constructing more challenging positive sample pairs to guide the encoder in learning intrinsic features robust to condition variations.Subsequently,a ConvNeXt-Transformer hybrid architecture is employed,integrating the superior local detail modeling capacity of ConvNeXt with the robust global perception capability of Transformer to enhance feature extraction in complex scenarios.Thereafter,a contrastive learning model is constructed with the optimization objective of maximizing feature similarity across different masked instances of the same sample,enabling the extraction of consistent features from multiple masked perspectives and reducing reliance on labeled data.In the final supervised fine-tuning phase,a multi-scale attention mechanism is incorporated for feature rectification,and a domain adaptation module combining Local Maximum Mean Discrepancy(LMMD)with adversarial learning is proposed.This module embodies a dual mechanism:LMMD facilitates fine-grained class-conditional alignment,compelling features of identical fault classes to converge across varying conditions,while the domain discriminator utilizes adversarial training to guide the feature extractor toward learning domain-invariant features.Working in concert,they markedly diminish feature distribution discrepancies induced by changes in load,rotational speed,and other factors,thereby boosting the model’s adaptability to cross-condition scenarios.Experimental evaluations on the WT planetary gearbox dataset and the Case Western Reserve University(CWRU)bearing dataset demonstrate that the SSMCL-DA model effectively identifies multiple fault classes in gearboxes,with diagnostic performance substantially surpassing that of conventional methods.Under cross-condition scenarios,the model attains fault diagnosis accuracies of 99.21%for the WT planetary gearbox and 99.86%for the bearings,respectively.Furthermore,the model exhibits stable generalization capability in cross-device settings.
