The Fault Recognition Of The Remote Belt Conveyor Idler Based On Support Vector Machine

As the key equipment of bulk material transportation the remote belt conveyor has been used widely.But idlers have a high probability of failure,which leads to various accidents.In order to overcome the shortcomings of traditional manual inspection,this paper studies the inspection method and proposes the automatic inspection scheme of idlers.By analyzing the running environment and characteristics of the remote belt conveyor,the scheme of automatic inspection device is proposed realize the inspection of idlers.And the device run on the H-shaped steel.In the meanwhile,the device can collect the idlers running data by sound pick-up,infrared thermal imager and other equipment.Then data is sent to the industrial computer which could deal with all kinds of data.By analyzing the running environment of remote belt conveyor,the key functional requirements and performance indicators is determined.And the system composition and driving plans of the device is analyzed.What's more,it is important that the overall structure of the automatic inspection device is designed,including the actuating device and the equipment carrying box.In addition,some equipments carried by the inspection device are selected.By analyzing the common fault types of idlers,the fault identification method of idlers based on Mel Cepstrum Coefficient,Principal Component Analysis and Support Vector Machine is proposed in this paper.Firstly,the four sets of sound signals are combined and normalized.Secondly,the Mel Cepstrum Coefficients are used as the 12 dimensional characteristics of the idlers.The next,the characteristics are reduced by using the Principal Component Analysis.Finally,by constructing the classification model and selecting the appropriate kernel function its parameters are optimized through the use of Grid Search Method,7 dimensional characteristics are classified by Support Vector Machine.And the results show that the accuracy of four groups' simulation experiments is more than 95%,which is proved to be effective for fault identification of idlers.All in all,the scheme of automatic inspection device and the method of idlers' fault identification based on Mel Cepstrum Coefficient,Principal Component Analysis and Support Vector Machine is proposed in this paper,which is of great significance in perfecting the rules of walk-around inspection of remote belt conveyor,improving the inspection efficiency of idler and avoiding all kinds accidents.