| The beam pumping unit is an important equipment for oil extraction.As the main bearing part of the beam pumping unit,the horse head often has failures during the operation,which affects production efficiency and even causes production accidents.This paper takes the horse head of the pumping unit as the research object,and develops a wireless monitoring system of the horse head state consisting of signal acquisition terminal,cloud server and local client,and deeply researches the dynamic characteristics and diagnosis methods of the horse head fault.The main research contents of this paper are as follows:(1)Developed a low-power wireless data acquisition terminal with STM32F767 as the processor,LSM6DSOX sensor and GM510 4G module as the main functional components,and wrote corresponding software.After experimental tests,the collection terminal can achieve accurate collection of triaxial acceleration signals and triaxial angular velocity signals with a sampling frequency of 6664 Hz,and has functions such as local data storage,network transmission,and feature statistics.(2)Based on C#,written cloud server software and local client software,and in conjunction with the signal acquisition terminal,constructed a three-dimensional and hierarchical structure of data collection,storage,and calculation of the horse head state.The terminal is responsible for signal collection and preliminary diagnosis,the cloud is responsible for data storage and instruction forwarding,and the client is responsible for data analysis and remote control,which realizes the remote monitoring and identification of the status of the horse head of the pumping unit.(3)This paper studies the failure characteristics of the horse head under four states:no failure,pin wear,installation misalignment and mixed failure.Based on the fast spectral kurtosis method,the shock component in the signal is extracted.The shock signal is superimposed and reconstructed according to the swing angle,which clearly expresses the characteristics of the shock signal.The measured signal found that the angular position of the impact signal will change with the change of the failure state of the horse’s head.The frequency spectrum analysis of the impact signals of different faults and different angular positions is carried out,and the obvious difference in the frequency structure is further found.(4)This paper proposes a fault identification method based on frequency domain features and principal component analysis.A variety of frequency domain characteristic parameters of 6-axis signals are calculated;an evaluation index for the effect of fault division is defined,and the characteristic parameters are expressed through specific numerical values.Fault division effect:After filtering out the parameters with poor division effect,after PCA principal component analysis,the rapid and high-precision identification of horse head faults is realized according to the projection map.The experimental verification shows that the recognition accuracy of this method reaches 98.5%.(5)Based on the ZCPA auditory model,the fault state of the horse head was identified,and the ZCPA auditory model was established.The auditory spectrum shape output by the model under different faults is obviously different;the auditory spectrum results are extracted by auditory features and principal component analysis,and the calculation results show This method can accurately distinguish different faults.According to the fault recognition effect of the projection map,the accuracy of the auditory spectrum is further verified.Verified by multiple sets of signals,the built model has a good fault recognition effect. |
PDF Full Text Request