Research On Health Monitoring Method Of Mine Belt Conveyor Idler

Due to the advantages of large transportation volume,long transportation distance,high transportation efficiency and continuous transportation,belt conveyors are widely applied and have become one of the key transportation equipments in the mining production process.According to the increasing requirements of output and productivity of coal mine,the mine belt conveyor is developing toward large-scaled and higher transmission speed.Because of its pivotal role in mine haulage,it's of great significance for coal mine safety to ensure the safe running of mine belt conveyor and avoid serious accidents.On the one hand,as the typical large-scale rotating machinery,its large amounts of carrier idlers are the key hazard sources of conveyor fire accidents.On the other hand,the traditional periodic maintenance and post-maintenance usually lead to the corresponding high maintenance costs,poor real-time performance,high security risks and service potential underutilization.Therefore,it is necessary to implement condition minotoring for the carrier idler of mine belt conveyor to diagnose its anomalies and predict its remaining useful lifes(RULs)via analyzing the monitoring data,which could contribute to developing appropriate maintenance strategies to eliminate hazards and maximize the idler's service time.Supported by the National Natural Science Foundation of China “Muti-time-scale Hybrid Signals based Health Management of Rotating Machinery Under Varying Operation Conditions” and the Natural Science Foundation of Jiangsu Province “Muti-time-scale Hybrid Signals based Health Monitoring of Large-scale Belt Conveyor in Coalmines”,this dissertation proposes a vibration signal analysis based approach for fault diagnosis,fault identification and RUL prediction of the carrier idler.This method is based on signal processing,feature extraction,fault diagnosis,health assessment and RUL prediction technology and theory.The corresponding research results can provide theoretical support and technical solutions to ensuring the safe operation of the whole machine and developing the optimal maintenance strategy for the carrier idlers of mine belt conveyor.The main contents include:(1)The fault evolution characteristics,fault characteristic frequencies and fault vibration model of mining idler are analyzed based on the special structure of mine belt conveyor and carrier idler.The time-and frequency-domain characteristics of the monitoring signal under different health states are also analyzed by means of simulation,which lays a foundation for the follow-up health monitoring study of carrier idler via spectrum structure mining technology.Meanwhile,some simulation experiments are also designed,and the corresponding data are obtained under different operating conditions.(2)In order to perform the health monitoring for the carrier idler of mine belt conveyor and diagnose its occurred faults,a fault diagnosis approach is studied for the carrier idler.In general,the monitoring signal of early fault is inevitably suffering from noise and the corresponding fault-related component is weak,so that it's difficult to determine the fault resonance frequency band.Considered the above-mentioned characteristics,a novel approach named structural information of spectrum(SIOS)algorithm is proposed for the fault diagnosis of carrier idler.In this method,the SIOS of monitoring signal is constructed to separate the fault component and the noise in spectrum level through utilizing their differences in the spectrum.Finally,the carrier idler fault is diagnosed through dentifying the dominant frequency components of the SIOS diagnostic parameters.(3)In order to recognize the idler fault rapidly with the accumulated fault samples in long-term production practice,a rapid fault pattern identification approach is also studied for the carrier idler.Firstly,according to the spectrum structure difference for the monitoring signals of carrier idler under different fault modes,a spectrum image based feature extraction method is proposed using computer vision technique.Secondly,considered the situation that the fault identification accuracy is commonly low due to the operating condition difference between the testing cases and the training ones,an adjusted spectrum image based feature extraction method is proposed by means of adjusting the spectrum with rotating frequency.Finally,the rapid fault identification of carrier idler is realized by employing two-dimensional principal component analysis(2DPCA)and nearest neighbor classifier(NNC)for feature dimension reduction and feature pattern classification,respectively.(4)The traditional periodic maintenance and post-maintenance could result in high-costs and service potential underutilization during the maintenance process of carrier idlers,hence the degradation assessment and RUL prediction approach is studied for conveyor idler as well.Firstly,utilized the excellent performance of SIOS algorithm on exploiting the fault information,a new degradation assessment approach is proposed for carrier idler by constructing a novel SIOS-based health indicator.Secondly,aimed at improving the low accuracy and reducing the difficulty in RUL prediction,a two-stage division strategy is introduced for describing the health degradation process of carrier idler.Meanwhile,an initial degenerate point(IDP)detection method is proposed for the determination of the boundary point between the divided two stages,where the IDP is detected by constructing an IDP indicator with the self-zero space observer of normal stage.Finally,the RULs of carrier idler are predicted with the SIOS-based health indicator and particle filtering algorithm on the basis of modeling its degradation process.At last,the work of this dissertation is summarized and the future research direction of the project is prospected as well.