Development Of Fault Prediction And Health Assessment System For Mine Main Drainage Equipment

This subject is the main research content of the major scientific and technological development plan project "establishment of mine automatic drainage system" of Shanxi Jinmei group.It is proposed for the lack of fault prediction and health evaluation function of underground main drainage equipment in coal mine.Taking the central main drainage system of Changping Mine as the research object,this paper establishes the intelligent analysis model of fault prediction and health evaluation of main equipment of main drainage system,and develops the Internet of things system of fault prediction and health evaluation technology of mine main drainage system,which integrates the underlying equipment information sensing network,information centralized acquisition communication network and ground intelligent analysis platform.The bottom equipment information sensor network is the comprehensive sensing part of the automatic drainage system of Changping central water tank,and it is also the bottom layer of the Internet of things.According to its transformation plan,vibration sensors are installed at the bearings,bases and volutes of centrifugal pumps,current and vibration sensors are installed on the pump motors,and these signals are collected with high-frequency cards to meet the requirements of spectrum analysis.In addition to vibration and current signals,the system also collects signals such as liquid level,temperature,negative pressure,positive pressure,working days,cumulative flow,pump motor temperature,gate valve opening,pump motor current(for protection and early warning).Information communication network is the reliable transmission part of the Internet of things system.The communication between underground and ground is based on Industrial Ethernet ring network,namely underground ring network.The high-frequency signal realizes the sharing of underground and aboveground resources in the form of network database.The signals such as liquid level are uploaded and issued by underground PLC and aboveground dispatching room with TCP/IP protocol.Ground intelligent analysis platform is the intelligent processing part of the Internet of things system.It is the key technology of fault early warning and health assessment.Firstly,the paper analyzes the fault mechanism of centrifugal pump and pump motor,the main equipment of water pump system,and finds its embodiment in the signal.Based on artificial intelligence technology such as neural network,Matlab is used as the analysis tool,FFT and Wavelet are used as the signal feature extraction method,WinCC configuration software is used as the front-end development platform,and SQLServer is used as the network database,the health management of underground main drainage system integrating data display,fault early warning and health evaluation is realized.This topic uses the data-driven method to evaluate the health status of the water pump system,and the acquisition of fault data can not be obtained overnight.Therefore,on the drainage system experimental platform established by Shanxi Key Laboratory of coal mine electrical equipment and intelligent control,the research group collected physical quantities such as vibration and current on centrifugal pump model equipment and specially made fault motor.Since the failure mechanism analysis of the main equipment and components of the well water pump system is not very comprehensive,and the signal eigenvalues are also obtained in the laboratory,some characteristic frequencies are different from the mechanism analysis.In order to obtain sufficient information and improve the accuracy of evaluation,This topic not only selects a total of 335 data,that is 293 segments of wavelet packet energy of high-frequency signal vibration and current of centrifugal pump and pump motor,42 low-frequency thresholds,as eigenvalues,but also selects the real-time signals of radial horizontal vibration at the driving end of centrifugal pump and radial horizontal vibration of pump motor body to form a 4D sample set with 64×32×2×2000,used as the input sample data of neural network intelligent analysis.On the basis of multi information samples,a multi-channel packet convolution neural network is designed,and the selection of various parameters of the whole network is analyzed and explained.Through the deep learning of multi-information and multi-channel grouping convolution network,the normal state of water pump system,centrifugal pump bearing fault,centrifugal pump blade fault,centrifugal pump bearing and blade mixed fault,pump motor broken bar fault,pump motor eccentric fault,pump motor broken bar and eccentric mixed fault,as well as pipeline pressure fault,pump motor overtemperature fault and electric valve fault are classified,and the accuracy and classification ability are significantly improved,and has good generalization ability,which realizes the effective evaluation of the health status of the water pump.