Research On Sensor Fault Diagnosis Of Gas Drainage Monitoring System

Gas drainage monitoring system is mainly used for monitoring and measuring CMM to ensure the safety of coal mine,which guarantees to implement the12-character principle of the production: "first draw after mining,monitoring and control to the wind production quotas".Due to the harsh environment,loss of precision and data exception caused by malfunction of the sensor or transmission malfunction is a typical problem in monitoring system,which causes inaccurate measurement,even accidents in production.Therefore,it’s practically significant to diagnose sensor data abnormality for gas drainage monitoring system.As the sub topic of “Yu xi ’three soft’ protruding seam with potential outburst into a High efficiency Gas Drainage Technology”,which is a project called Collaborative Innovation Center of Work Safety,Henan Province key scientific and technological,based on the correlation data for different regions,this paper takes gas drainage monitoring system of Zhengzhou Coal Group as the research object,and the main work of this paper is shown as follows:Aiming at unicity and low detection efficiency of correlation analysis method,which is used in expert subsystem of the existed pipe network monitoring system,this paper puts forward a method for diagnosing sensor fault based on principal component analysis(Principle Component Analysis,PCA),and focuses on the PCA modeling,sensitivity of three types statistic indicators and fault source localization method.What’s more,the validity of the proposed method,which is applied to the pipe network system to diagnose sensor fault,has been verified in the final experiments.According to the nonlinear relationship of drainage pump monitoring variables in the pump room monitoring system,a sensor fault detection method based on kernel principal component analysis(Kernel PCA,KPCA)is proposed and the corresponding detection model is constructed.Compared with PCA and KPCA by simulation experiments,the validity of KPCA has been verified.Aiming at the complexity of fault source identification and low efficiency in KPCA application program,this paper improves and proposes a recognition method ofmean square contribution value,which is used to locate the fault source and it’s suitable for online operation.For the instabilities of position indicating,the first-order inertial filter is used to obtain stable characterize indexes so that it can improve the efficiency of fault identification.The effectiveness of the proposed scheme has been verified by the simulation data and the field history data.