Research On On-Line Coal Quality Detection Method Of Fully Mechanized Caving Coal Based On Multi-Mode Fusion

Accurate identification of coal and rock is helpful to remote mining face,unmanned/less human,and is an important part of intelligent mine construction and intelligent mining in the coal industry.In this paper,the coal and rock video of fully mechanized top coal caving working face is taken as the research object,the sound Meir spectrum map data set and the coal and rock image data set are constructed for analysis,and the convolutional neural network is used for classification and recognition to obtain the recognition rate required by the experiment.The D-s evidence theory can be used to integrate the decision level,improve the recognition rate and the stability of the coal-rock identification system.The main work is as follows: extract sound data and image data using Python’s own module Opencv,generate data sets and pre-process them separately.The sound pre-processing includes pre-weighting,framing,and windowing.The pre-processed sound data is generated into Meier spectrograms,which are divided into 3 categories:coal and coal impact sound dataset,coal and rock impact sound dataset,and rock and rock impact sound dataset;the VGG deep learning model of convolutional neural network is used to classify and recognize the 3 categories of datasets,and the recognition rate is derived.The image data preprocessing mainly works on rotating,intercepting,adding noise and other operations to expand the data volume,and the final dataset is divided into 3 categories,which are: coal image dataset,coal and rock image dataset,and rock image dataset.The VGG deep learning model of convolutional neural network is used to classify and recognize the 3 types of datasets,and the recognition rate is obtained.After two corresponding recognition rates are obtained,the two evidence sources are fused at decision level.It is experimentally demonstrated that the signal fusion can improve the recognition rate...