Research On The Method Of Detecting The Abnormal State Of Rotating Equipment

The fault diagnosis technology of rotating equipment has an important position and application value in engineering,and it has a high priority in establishing a set of feature extraction and state detection methods.Prevent major accidents in the equipment through early diagnosis.It has great value for national economy and security,and is also the focus of current research.Commonly feature extraction and detection methods are roughly divided into three categories:frequency domain analysis,time domain analysis and time-frequency domain analysis.However,the real signal components are complex and the noise interference is serious.Using the latest machine learning algorithms,the sample training cost is high and the parameter adjustment is unsuitable in the later stage,which has poor application in industrial production.Many diagnostic indicators for fault diagnosis of rotating equipment have different scopes of application,and encounter diverse signals with poor results and misjudgments.Because the early signal of the fault is very weak and difficult to diagnose,industrial diagnosis requires high accuracy.A system that can effectively extract features and make accurate diagnosis in most cases is hard all the time,and research in this direction is also very meaningful.Aiming at the above problems,this paper builds an effective feature extraction and diagnosis system,the main research is as follows:First,this paper proposes a method based on sparse filtering and parameterized time-frequency analysis(SF-PTFA).Select the corresponding atoms through sparse filtering to get the expected fault characteristics,and then use parametric time-frequency analysis to accurately extract these signals.In the simulation and real signal experiments,the bearing and gear signals are separately analyzed,and the fault signals can be well resolved and accurately extracted.It is proved that this method has the advantages of identifying fault components and accurate feature extraction in complex signals with low signal-to-noise ratio,which has a positive impact in future research.Secondly,after analyzing and highlighting some potential fault components of the signal through SF-PTFA,this paper uses principal component analysis to synthesize the results of each diagnostic index based on 16 practical diagnostic indexes,and obtains a more objective evaluation result.The actual bearing fault signal of testbed has fully verified the feasibility of this method.The monitoring and diagnosis method has the advantages of being sensitive to faults of complex rotating equipment and having good applicability to a large number of signals.Compared with traditional methods,the diagnostic effect has been significantly improved.Finally,according to the intelligent requirements of the development of Industry 4.0 standards,this article uses software design,which is the most effective way to achieve application today.Relied on the algorithm research,this paper designs the software process and interface function module through the KANO model requirement analysis method,and finally edits and completes a PC-side application software.The software finally passed the actual signal test,it can indicate that the software has good reliability and practicality.The software gathers the characteristics of the above research methods,it has a good diagnostic effect in complex high-noise signals,and the software operation is very simple and easy to use.The landing application of the abnormal monitoring method for rotating equipment is realized.