Some ideas in the development of fault diagnosis system for spacecraft are introduced. Firstly, the architecture of spacecraft fault diagnosis is proposed hierarchically with four diagnosis frames, i.e., system level,...Some ideas in the development of fault diagnosis system for spacecraft are introduced. Firstly, the architecture of spacecraft fault diagnosis is proposed hierarchically with four diagnosis frames, i.e., system level, subsystem level, component level and element level. Secondly, a hierarchical diagnosis model is expressed with four layers, i.e., sensors layer, function layer, behavior layer and structure layer. These layers are used to work together to accomplish the fault alarm, diagnosis and localization. Thirdly, a fault-tree-oriented hybrid knowledge representation based on frame and generalized rule and its relevant reasoning strategy is put forward. Finally, a diagnosis case for spacecraft power system is exemplified combining the above with a powerful expert system development tool G2.展开更多
The fault diagnosis of bearings is crucial in ensuring the reliability of rotating machinery.Deep neural networks have provided unprecedented opportunities to condition monitoring from a new perspective due to the pow...The fault diagnosis of bearings is crucial in ensuring the reliability of rotating machinery.Deep neural networks have provided unprecedented opportunities to condition monitoring from a new perspective due to the powerful ability in learning fault-related knowledge.However,the inexplicability and low generalization ability of fault diagnosis models still bar them from the application.To address this issue,this paper explores a decision-tree-structured neural network,that is,the deep convolutional tree-inspired network(DCTN),for the hierarchical fault diagnosis of bearings.The proposed model effectively integrates the advantages of convolutional neural network(CNN)and decision tree methods by rebuilding the output decision layer of CNN according to the hierarchical structural characteristics of the decision tree,which is by no means a simple combination of the two models.The proposed DCTN model has unique advantages in 1)the hierarchical structure that can support more accuracy and comprehensive fault diagnosis,2)the better interpretability of the model output with hierarchical decision making,and 3)more powerful generalization capabilities for the samples across fault severities.The multiclass fault diagnosis case and cross-severity fault diagnosis case are executed on a multicondition aeronautical bearing test rig.Experimental results can fully demonstrate the feasibility and superiority of the proposed method.展开更多
文摘Some ideas in the development of fault diagnosis system for spacecraft are introduced. Firstly, the architecture of spacecraft fault diagnosis is proposed hierarchically with four diagnosis frames, i.e., system level, subsystem level, component level and element level. Secondly, a hierarchical diagnosis model is expressed with four layers, i.e., sensors layer, function layer, behavior layer and structure layer. These layers are used to work together to accomplish the fault alarm, diagnosis and localization. Thirdly, a fault-tree-oriented hybrid knowledge representation based on frame and generalized rule and its relevant reasoning strategy is put forward. Finally, a diagnosis case for spacecraft power system is exemplified combining the above with a powerful expert system development tool G2.
基金The work was supported by the National Key R&D Program of China(Grant No.2020YFB2007700)the National Natural Science Foundation of China(Grant No.51975309)+1 种基金the State Key Laboratory of Tribology Initiative Research Program,China(Grant No.SKLT2020D21)the Natural Science Foundation of Shaanxi Province,China(Grant No_2019JQ-712).
文摘The fault diagnosis of bearings is crucial in ensuring the reliability of rotating machinery.Deep neural networks have provided unprecedented opportunities to condition monitoring from a new perspective due to the powerful ability in learning fault-related knowledge.However,the inexplicability and low generalization ability of fault diagnosis models still bar them from the application.To address this issue,this paper explores a decision-tree-structured neural network,that is,the deep convolutional tree-inspired network(DCTN),for the hierarchical fault diagnosis of bearings.The proposed model effectively integrates the advantages of convolutional neural network(CNN)and decision tree methods by rebuilding the output decision layer of CNN according to the hierarchical structural characteristics of the decision tree,which is by no means a simple combination of the two models.The proposed DCTN model has unique advantages in 1)the hierarchical structure that can support more accuracy and comprehensive fault diagnosis,2)the better interpretability of the model output with hierarchical decision making,and 3)more powerful generalization capabilities for the samples across fault severities.The multiclass fault diagnosis case and cross-severity fault diagnosis case are executed on a multicondition aeronautical bearing test rig.Experimental results can fully demonstrate the feasibility and superiority of the proposed method.
