Research On Equipment Fault Full Vector Prediction Model Based On Big Data

Rotating machinery is an important part of mechanical equipment,it will cause the equipment or production process to stop running once the accident happened,which even causing a serious accident and significant economic losses.Therefore,the fault prediction of mechanical equipment is paid more attention and research.The traditional frequency spectrum analyzing method depends on vibration information coming from single channel,which loses the integrity of the information,while the full vector spectrum technology adopts the idea of the information fusion of the two channels to ensure that the spectrum contains complete and comprehensive vibration information;A large number of signals are usually collected in the modern equipment monitoring system,however,it cannot be fully leveraged in the process of fault prediction,that resulting in low credibility of mid-long-term prediction.Time series clustering method can cluster the approximate state and simplify the amount of monitoring information,that makes it possible to use more historical information in the prediction process;In order to overcome the defects of FV-ARIMA and FV-SVR prediction model,an improved FV-SVR prediction model is proposed.Therefore,this paper takes big data of equipment monitoring as the research object and the full vector spectrum technique and time series clustering as theoretical support and researches on equipment fault prediction combined with improved FV-SVR prediction model.The main research works are as follows:(1)The theory and algorithm of full vector spectrum technology are researched in detail,the algorithm of Hilbert-full vector spectrum is given and applied to the analysis of rolling bearing deterioration,It is verified that the Hilbert-full vector spectrum has a good envelope demodulation effect,and the characteristic principal vibration vector can be used to represent the vibration intensity and distinguish the fault type.(2)Characteristics of equipment monitoring data are researched,and the concept of big data of monitoring equipment are proposed;Data smoothing method and time series clustering analysis method are studied,and applied to the real time series,obtaining a good effect of smoothing effect and clustering.(3)The basic theory and algorithm of ARIMA model and SVR prediction method are studied;The basic process of the full vector prediction model is presented;rolling bearing state prediction,The advantages and disadvantages of full vector ARIMA and full vector SVR prediction model are analyzed and summarized through applying to the prediction of rolling bearing fault.(4)To overcome the shortcomings of full vector prediction model,propose an improved FV-SVR prediction model;Combining with the time series clustering method and the improved full vector SVR prediction model,mid-long term full vector prediction model of equipment fault are constructed based on big data;all the historical data in the rolling operation are applied to validate the improved full vector SVR prediction model and mid-long term full vector prediction model of equipment fault,the result shows that,the two prediction methods have achieved good results.