On-line Detection System For Rock And Soil Moisture Based On Image Recognition

Gas drilling is a new type of drilling technology that uses gas as a circulating medium and crushes rock energy.It has the advantages of protecting reservoirs and high drilling efficiency.However,when it encounters water from the formation,it may cause stuck drilling and wall collapse,which may cause major safety accidents.With economic losses.The humidity of the rock and soil brought out by the drill bit during the drilling process can be used as one of the indicators for detecting the water output of the formation.In gas drilling,geotechnical soil is usually collected manually at a fixed sampling point,and then the geotechnical humidity is measured by the dry weighing method,which has the disadvantages of slow detection speed and automatic detection.In order to realize the geotechnical humidity detection in gas drilling,this paper designs an image recognition method based on image processing and machine learning algorithm to detect geotechnical humidity.The detection principle is as follows: firstly,the filtering algorithm is used to denoise the collected geotechnical image,then the mixed Gaussian model is used to segment the image and find the pixels belonging to the geotechnical,and the segmented image is transformed from the RGB color space to the LAB color space.The DBSCAN clustering algorithm finds the center point of the LAB color space,and inputs the center point value as the eigenvalue into the ridge regression algorithm,and outputs the predicted geotechnical humidity value.This paper compares the denoising effect of mean filtering,median filtering,Gaussian low-pass filtering,mixed Gaussian model and image segmentation effect using a and b eigenvalues in LAB space,RGB space and LAB space clustering effect,in regression algorithm The least square method,ridge regression,lasso and elastic network algorithms are analyzed and the feasibility of the ridge regression algorithm is verified by reference data.The segmented ridge regression algorithm is designed to meet the high precision of gas drilling and low humidity.Measurement needs.The geotechnical humidity online detection system of this paper consists of three parts: lower computer,upper computer and WIFI communication.The lower computer uses the open source hardware Raspberry Pi as the controller to connect the industrial camera through the USB interface,realizes the camera's acquisition image size,white balance,exposure and other parameter settings,real-time image data acquisition,image compression and transmission using Huffman algorithm;The machine visualization interface is developed by Qt framework,which realizes the functions of geotechnical image data reception,image algorithm processing,data visualization and data storage.The WIFI communication is used between the upper computer and the lower computer,and the transmission layer uses the TCP protocol.In order to verify the effect and repeatability of the above-mentioned geotechnical humidity detection method,this paper built a simulation environment and conducted experiments.The cross-validation method was used to analyze the experimental results,and the functions of each module of the geotechnical humidity online detection system were tested.The feasibility of the system is proved.The geotechnical humidity detection method in this paper can meet the complexity of gas drilling scene,the need for more accurate detection under low humidity of rock and soil,and meet the real-time performance of drilling speed,which has certain reference value.