| Coal mine accident statistics show that the proportion of roadway roof accidents is the first of all types of accidents.And with the increasing depth and intensity of coal mining,the safety of roadway roof is particularly prominent.Real-time optimization of support scheme and parameters according to the characteristics of roadway roof rock layers is an important way to prevent roof accidents.However,the traditional method for detecting the roof strata of coal mine roadway is slow and expensive,and can no longer meet the needs of safe,efficient and intelligent mining.In view of the above problems,this thesis further improves the rock-breaking mechanism of the rotary two-wing PDC bit;comprehensively analyzes the response characteristics of various drilling parameters when drilling in different rock formations through ANSYS numerical simulation software and field tests;independently developed the device to collect drilling parameters.The identification method of rock type,rock interface and fracture while drilling is proposed through laboratory experiments.Based on the existing borehole imaging technology,an intelligent identification program for rock fracture has been developed,which promotes the development of intelligent identification of rock strata characteristics,and provides geological guarantee for the intelligent development of coal mines.The main conclusions of the study are as follows:(1)The rock breaking mechanism and drilling response characteristics of the two-wing PDC bit were analyzed through theoretical analysis,numerical simulation and field test.The shape of the borehole bottom of the two-wing PDC drill is "boss-shaped",and the borehole bottom is asymmetrically damaged during actual drilling.For rock formations with uniaxial compressive strength of 25 to 98 MPa,the reaming range is between 0.273 to 2.056 mm.With the increase of rock strength,the degree of hole expansion has a decreasing trend.The cuttings size is closely related to the drilling efficiency.When the cuttings size is larger,the drilling efficiency is relatively higher.Thrust and torque have similar trends,and as the thrust and torque increase,the ROP also increases.Both thrust and torque are quadratic functions with ROP.There is a significant correlation between rotational speed and sound pressure level,as well as cutting depth and ROP.(2)Based on the self-developed parameter acquisition device while drilling,the relationship between drilling parameters and rock mechanical properties was analyzed.The experimental device can realize real-time acquisition of parameters such as displacement,torque,rotational speed and sound pressure level.Based on this experimental device,drilling experiment was carried out in the laboratory.The relationship between 8 parameters while drilling including ROP,rotational speed,torque,sound pressure level,mechanical specific energy,torque work,torque work ratio,and cutting depth and 6 rock mechanics parameters including uniaxial compressive strength,tensile strength,cohesion,internal friction angle,elastic modulus and Poisson’s ratio is comprehensively analyzed.Finally,the relationship model between each drilling index and each rock mechanical parameter is obtained,which provides a reference for the identification of rock layer characteristics while drilling on the roadway roof and the determination of rock mechanical parameters.(3)An identification method while drilling is proposed for the rock type and rock interface of the roadway roof.Through the laboratory simulation of the drilling process of different rock formations,it is found that under stable drilling conditions,with ROP and sound pressure level as input parameters,the recognition rates of BP neural network and support vector machine method for rock formation types are89.53% and 89.13% respectively.Compared with other drilling parameters,using the ROP as the input parameter,the change point detection method can effectively identify the location of the rock interface.When the ROP and the sound pressure level show similar trends,the drilled rock layer is the same rock layer;when the ROP increases and the sound pressure level has no obvious change or a downward trend,the drill bit encounters a rock layer with lower strength;When the sound pressure level decreases and the sound pressure level does not change significantly or has a downward trend,the drill bit encounters a rock formation with higher strength.(4)The response characteristics of the drill bit when passing through the fracture were analyzed,and a multi-parameter voting method is proposed to identify rock fractures while drilling.With the increase of the fracture width,when the drill bit enters the fracture from the rock layer,the rock damage area formed around the borehole tends to increase.When the bit drilled through the fracture into the rock,there was little visible fractured area around the borehole.In addition,the cuttings size in the fractures was larger when the drill bit passed through the larger-width fractures compared to the smaller-width fractures.A multi-parameter voting method for fracture identification under unstable drilling conditions is proposed.Through experimental data verification,it is found that the multi-parameter voting method has a low recognition rate(37.5%)for 1mm wide fractures,but the average recognition rate for2 mm and 3mm fractures reaches 96.88%.With the increase of rock strength,the false alarm rate gradually increased.(5)The gray value description method of rock formation characteristics is proposed,which realizes the automatic identification of rock formation interface under the condition of borehole imaging.The method 1 proposed in this dissertation can be used to identify the position and dip of the rock interface.But it is mainly suitable for rock formations with homogeneous,clear interface and no fracture interference.Compared with method 1,method 2 cannot determine the dip angle of the rock interface.However,the method 2 has stronger anti-interference ability and can more accurately identify the location of the near-horizontal rock interface.(6)An intelligent identification program of rock fractures based on borehole imaging was developed.The borehole imaging was performed in 6 coal mines,and454 effective borehole histograms were obtained.These image databases are divided into training set,validation set and test set,and the shape of fractures is statistically divided into structural plane,longitudinal fracture and broken area.On this basis,a post-processing program is designed and developed.The test results show that the average recognition rate of three different fractures is 88.71%,and the average false alarm rate is 16.67%.Among them,the recognition rate of structural plane is 86.67%,and the false alarm rate is 13.33%;the recognition rate of longitudinal fracture is88.06%,and the false alarm rate is 23.88%;the recognition rate of broken area is91.53%,and the false alarm rate is 11.86%.(7)Based on the above research conclusions,conduct field tests in Wangjiazhai Coal Mine,Youzhong Coal Mine and Laomupo Coal Mine.The study found that the decision tree algorithm can quickly and accurately identify the position of the coal-rock interface,and the average identification error of the coal-rock interface in the two coal mines is 0.04 m and 0.06 m,respectively.For rock layers with similar strength and color,it is difficult to accurately determine the interface position of each rock layer by manual observation.In this complex situation,compared with the decision tree algorithm,the Strucchange model can better identify the position of the rock interface.For Wangjiazhai Mine,the coal rock type recognition rates of BP neural network model and SVM are 93.69% and 92.73%,respectively;for Youzhong Mine,the coal rock type recognition rates of BP neural network model and SVM are93.40% and 85.97%,respectively;for the rock layers with similar strength in Laomupo Mine,the recognition rates of rock types by BP neural network model and support vector machine are 62.85% and 61.71%,respectively.When the difference of rock stratum strength is large,the recognition rate of rock stratum type is higher.This dissertation totally has 143 figures,39 tables,and 190 references,respectively. |
