Design And Realization Of Cuttings Recognition System Based On Linear CCD

In the process of oil exploration and development, the application of PDC (Polycrystalline Diamond Compact) drill causes a greater degree of rock fragmentation, produce tiny cuttings particles, and make it difficult to accurately determine the true lithology of cuttings artificially. And this seriously affected the logging cuttings naming and description, resulting in significantly reducing of coincidence rate of some exploratory profile. Therefore, it is urgent to develop a set of PDC lithology identification system for logging on site. Although domestic companies have developed cuttings logging equipment, the accuracy and applicability of cuttings recognition remain to be improved. In recent years, digital image acquisition and processing technology is widely used in all scale of society for its high flexibility, wide area of adaptation and reliable processing accuracy, but the application in the oil exploration is rarely reported. To meet the requirements of current cuttings logging field, in this thesis, a set of high-resolution cuttings identification system based on linear CCD is specially designed and also improved the system function through extensive testing experiment, achieving the stability and adaptability standards of field application.The author's main work is divided into two parts:hardware and software, in which hardware devices successfully achieved the clear recording of digital images of the cuttings through the rational selection of devices and optimize design of structure. In this system, the images were recorded by a linear CCD(BASLER L304KC) with 4080 pixels. A PIXCI(?) ELI frame grabber is used to transfer images to PC. The information of image texture details can be displayed more clearly by using macro lens whose largest multiple is 1:1; under the uniform motion of transmission plate which is driven by screw, the image size is not distorted. Then we completed the controlling part of linear CCD by invoking the library functions package attached to the CCD, and also completed the design of mechanical devices plate and its control programming, so as to ensure that PC software can send real-time commands and operates accurately. An optimum parameter configuration was determined through adjusting and testing parameters of each hardware device, including aperture of macro lens, zoom multiples and depth of field, exposure time of linear CCD, transmission rate and so on. Clear images of cuttings could be accurately displayed on PC finally.The second part of this thesis is software updating and testing, which are based on images captured by this linear CCD. It includes database design, software design and improvement of recognition system, cuttings identification scheme, and as well as the testing part of system stability and adaptivity. An image data storage system, based on Oracle database, was designed to save the datas and testing results of cuttings sample on local disks through planning and setting up the database. The upgrade work on software interface were as follows. Scroll bar processes was added and so as the picture's viewing functions frame, making the image processing and operating process more simple and clear. An identifying experimental of cuttings identifying scheme was made to obtain the best parameter settings of this cuttings recognition system, and Gabor was determined as the feature extraction method. SVM(Support Vector Machine) was determined as the classifier and its parameters were configured. Finally, Through 12 groups'testing of cuttings samples with single lithology, a conclusion was made that the recognition accuracy rate for 10 groups reached 95%; then after each 5 tests of six groups of mixed lithology samples, a least recognition accuracy rate was got as 95% for 5 groups, which shows that the system can ensure a perfect adaptivity in logging work site.In summary, High-resolution images of PDC cuttings can be captured by this cuttings recognition system based on linear CCD, and high recognition accuracy rate is achieved using digital image processing technology:the accurate rate of identification of single-lithology cuttings and cuttings with mixed lithology both reach 95%. This result meet the requirements of high intelligence, high-resolution, high accuracy of logging site for the system, but the image processing of this system takes a bit long time, so the direction of our further work is to improve the software algorithms in order to reduce the overall uptime.