Research On Compound Fault Diagnosis Of Fan Bearing And Blade Based On DSP

Fan equipment is a widely used rotating machinery,which plays an important role in venting and generating negative pressure in a cigarette factory,and guarantees the quality and speed of cigarettes in the package production line.According to the actual conditions of the fan failure in the cigarette factory,the vibration signal of blade and bearing failure in the fan rotor system was studied.The traditional vibration signal-based fault detection method usually performs Fourier transform directly on the non-stationary vibration signal collected by the sensor,and then checks the signal frequency distribution.Although this method has certain efficiency for fault diagnosis,it is difficult to analyze the types of faults contained in the signal more comprehensively.Based on this situation,the following research work has been carried out.Based on a thorough understanding of the work environment,working principle,and common faults of the cigarette factory fan,a fault simulation experiment platform was set up.The MEMS acceleration sensor was used to collect vibration signals at the bearing seat,and the single-channel blind source separation method was used to decompose this mixture signal.Firstly,the vibration signal is decomposed by using the pole symmetry mode,and the problem of underdetermined blind separation of single channel is converted into a multi-channel positive definite or overdetermined blind sepa ration problem.Then the principal component analysis method is used to extract the main component of the multi-dimensional signal.The experimental results show that the separation of vibration signals caused by different fault sources can be achieved to some extent by using this method.The drawbacks of the sample entropy and multiscale entropy when assessing the complexity of time series are firstly discussed,and then the fundamental notion and algorithm of hierarchical entropy are introduced.Th e hierarchical entropy calculates the sample entropy of lower frequency components of a time series and computes the sample entropy of the higher frequency components.Finally,the SVM is used to identify and classify different faults.The experimental results show that the accuracy of fault identification by using the above method can be very high when the number of samples is small.Based on the research and verification of the fault diagnosis algorithm,the software system was developed on the hardware circuit with TI's TMS320C6747 DSP as the core chip.With the help of Matlab software,the difficulty of program development is reduced,and the correctness of the program is also guaranteed.The program is written in C language,and thus the portability of the program can be improved.The software is based on a modular design,which is conducive to debugging the program and adjusting the program's functionality.