Development Of IoT Based Monitoring And Life Management System For Drainage Equipment In Mine

This paper is one of the important sub-topics of the major science and technology projects " Monitoring of Major Production Equipment for Coal Mine and Life Cycle Management System Development Based on Internet of Things(No:20131101029)" in Shanxi province.It is also the continuation of the project " Development of Automation Monitoring System of Main Drainage Used in Coal Mine(No:20100819-2)".It is proposed on the current situation that the conditions of poor environment,the fault of frequent operation and mine main drainage equipments monitored and forecasted hardly.The drainage system in mine bears the drainage task of the underground coal mine,and the operation and health situation of its equipments affect the safety in production of the underground coal mine.At present,the main drainage equipment in most coal mining enterprises has realized automatic control.It can control centrifugal pump and solenoid valve throught monitoring the water level and pipeline pressure,and can realize the remote control of the main drainage equipments in mine through connecting with the ground monitoring center.However,prognostic and health management monitoring of the main drainage equipment in mine are not comprehensive enough.The main manifestations are the small fault diagnosis range,the inaccurate fault location,the high false alarm rate,fault prediction based on the threshold prediction,and the comprehensive diagnosis and health state management are not realized.In view of the above problems,fault diagnosis and life prediction of the mine main drainage equipment was studied.It is through the theoretical analysis,experiment and data collection,intelligent treatment to realize the effective monitoring and reliable diagnosis of the life nodes and different fault states of the mine main drainage pump,and develop monitoring and life management system for the drainage equipment in mine.It is of great practical significance to ensure the safe and reliable operation of the equipment and avoid heavy casualties and economic losses caused by the coal mine water disaster.Through analysing the characteristic curve of the main drainage pump in mine,obtain the factors that affect the change of the characteristic curve of the pump and determine the life prediction indexs of the main drainage pump in mine are the head,the power,the efficiency and the NPSH with combining the composition,working principle and working environment of the main drainage water pump in mine.At the same time,we completed the experiment on the centrifugal pump accelerated life test platform,collected data and got the life prediction index value of centrifugal pump.Then,the residual life prediction method of centrifugal pump based on Grey-Kalman model is proposed.The validity of the Grey-Kalman model is verified.The prediction accuracy is 50% higher than that of the grey model,and the convergence speed is greatly improved.The life prediction indexs is input into the Grey-Kalman model to realize the residual life prediction of the main drainage pump in mine.The fault feature extraction method of centrifugal pump in mine is studied.The denoising method of vibration signal based on the blind source separation algorithm(BSS)is proposed,avoiding the lossing of partial fault information,retaining the original signal information at the greatest degree.The multi-scale decomposition method based on local mode decomposition(EMD)and feature vector construct method combining energy and sample entropy is proposed.EMD can suppress the endpoint effect and decreases the cumulative error,Characteristic quantity,energy and sample entropy,makes polymerization degrees of the similar fault samples higher,and discrimination of different samples better;Finally,data dimensionality reduction method based on expectation maximization-principal component analysis(EM-PCA)is proposed,taping the low dimensional characteristics of high dimensional data fully and improving the sensitivity of the characteristics further.The overall plan of fault diagnosis of centrifugal pump in the mine is made.The fault diagnosis experiment of centrifugal pump is completed on the accelerated life test platform,and vibration signal in each state is collected.Reduction noise using BSS algorithm for vibration signals of centrifugal pump and decompose multi-scale components using EMD algorithm for vibration signals of centrifugal pump,the feature vector of fault diagnosis of centrifugal pump based on energy and sample entropy was obtained.Considering that different data reduction methods and reduce dimension has different influences on mine main drainage pump fault diagnosis accuracy and convergence rate,it determine that EM-PCA is the optimal data reduction method and the 2 dimension is the optimal dimensionality.Feature vector after dimensionality reduction were input into BP neural network to achieving identification of different faults of mine main drainage pump,and the recognition accuracy is 98.75%.The life management platform of mine main drainage equipment based on the Internet of things is developed.First,complete the selection of sensor,data acquisition card and industrial control machine and the configuration of hardware and software,collect the real time signal of the mine main drainage equipment,realize the comprehensive perception of the state information of the main drainage system of the mine.Then,with Ethernet as the medium,the state information collected is written to the database remotely by Lab VIEW and the reliable transmission of the state information of the mine main drainage system is realized.Finally,through analysing of state information,intelligent processing and visual display,the fault diagnosis and life prediction of mine main drainage pump is realized.