Research On Key Technologies Of Rib Spalling Recognition On The Fully Mechanized Coal Mining Face

On the fully mechanized coal mining face,the rib spalling may cause the leakage and the structural damage of roof supports.Large-scale rib spalling may lead to the load-mutation of scraper conveyer,which may damage the driving motors,threaten the stability of power in working field and worsen the security of the fully mechanized coal mining face.To recognize the rib spalling,the key technologies of recognizing the rib spalling based on machine vision have been studied in this thesis.In order to assess the hazard level of rib spalling,the mixed algorithm for enhancing the monitoring image and the feature analysis method of rib spalling were proposed.The main researches are as follows:(1)To solve the problem that the low-quality monitoring images had on fully mechanized coal mining face,the mixed algorithm based on single-scale Retinex and bilateral filtering was proposed.The experimental results showed that,the proposed method was more suitable for enhancing the quality of rib spalling monitoring image in aspects of subjective effect,contrast and comentropy than common methods.(2)The feature analysis structure of rib spalling was built,four features: time,area,height of rib spalling and rib spalling center height were analyzed,the analysis method was proposed too.Simulation results showed that,the error of analysed value was small enough to support the assessment of rib spalling hazard level.(3)Four hazard level of rib falling: safe,slight,medium and heavy were chosen.The assessment of rib spalling hazard level based on support vector machine(SVM)was studied,and the BP neural network and artificial immunity algorithm were taken as the comparison.The simulation results showed that SVM was better than the other two methods in the case that the feature samples are few.(4)The simulation experiment platform was built and experiments were executed.The results showed that,the mixed image enhancement algorithm could improve the quality of interrupted rib spalling image,decrease the average error of feature analysis and increase the accuracy of rib spalling assessment 13.3%,but consumed more time.