| Nowadays,mechanical equipment has been widely used,but it’s failure rate remains high.The failure of these mechanical equipment will affect the operation of the entire system and have a huge impact on production,and sometimes even the failure can be catastrophic.Rotating machinery is a relatively common mechanical equipment.It’s prone to failure during the special working environment for a long time,including wear,erosion,impact,vibration,fatigue and other effects.As the core component of rotating machinery equipment,bearings are closely related to the occurrence of rotating machinery failures.Therefore,bearing fault detection and diagnosis can not only ensure the safe operation of the equipment,but also maximize the working potential of the equipment.In summary,the research of bearing fault diagnosis technology will be important on theoretical and practical value.Traditional bearing fault diagnosis is generally divided into two types.One is offline diagnosis,a method of completely diagnosing the bearing locally,the other is online diagnosis,a method of completely diagnosing the bearing in the cloud.No matter which method is used,the impact of diagnosis resources and environment is not be considered,and this kind of research loses its practical value.the resources to support bearing fault diagnosis is different while various working conditions of bearing.In order to make the research results more generalizable,this paper researches the bearing fault diagnosis technology with the condition of limited diagnosis resources under cloud plus end frame.First of all,this study combines the characteristics of the edge computing framework and builds a new framework for bearing fault diagnosis——cloud plus end frame.For how to allocate the bearing fault diagnosis algorithm to cloud plus end frame,this paper proposes the cloud plus end bearing fault diagnosis mode.This mode has two methods of offline and online that can handle the disconnection or unblocking of the network,and also has a selfupdate function of the diagnostic model,so that cloud plus end fault diagnosis system will not be misdiagnosed due to wear and tear of the equipment for a long time.Then for the purpose of saving resources.Furthermore,an improved algorithm for online diagnosis in fault diagnosis of cloud plus end bearing is proposed.Experiments show that the algorithm has higher comprehensive advantages in diagnosis time and accuracy.Then,on the basis of the fault diagnosis mode of cloud plus end bearing,the condition of limited network resources is added for research.Considering the indirect relationship between the length of the uploaded fault feature,the network bandwidth and the diagnosis time,an adaptive feature extraction algorithm based on network bandwidth is proposed for the problem.The diagnosis time and diagnosis accuracy obtained by the algorithm can reach the theoretically optimal results.Through simulation experiments and comparison with several common upload methods,it can be found that the proposed algorithm has more advantages in the overall evaluation.Finally,under the condition of limited network and computing resource while diagnose on the cloud plus end,how to diagnose the bearing effectively and reliably has been studied.Aiming at this problem,this paper proposes a decision framework for bearing fault diagnosis of cloud plus end based on reinforcement learning.This framework judges the current global upload strategy of the system by synthesizing the fault sample information on the device side and the global resource allocation information on the edge side.Within the limited resources,results show that the method can ensure the device side that needs more diagnostic resources is priority given and the timeliness of the diagnosis,the reliability of the diagnostic system is improved.Cloud plus end bearing fault diagnosis mode proposed in this study makes full use of the characteristics of the device side,edge side,and cloud side,and can perform effective bearing fault diagnosis under various circumstances such as computing and network resource constraints.So that the safety work of bearing equipment is more guaranteed,has certain practical and research value in engineering applications. |
PDF Full Text Request