| Ultrasonic imaging logging,one of the leading logging technologies in the world,is able to reliably and visibly present formation and well wall condition information in the form of images,significantly enhancing the effectiveness of oil well exploration and development.However,in actual engineering applications,the eccentricity of the instrument,borehole form,external environment,and transducer properties might make it challenging to analyze and interpret logging data and inevitably degrade the logging image quality.Therefore,it is important to carry out research on efficient ultrasonic imaging logging techniques.This paper uses the built ultrasonic imaging logging simulation platform as the basis to acquire the original data,from the perspectives of extraction the first-arrive time of the ultrasonic logging signal and enhancement of logging image processing,respectively,to achieve the goal of improving the quality of logging images.The problems of low image clarity and lack of contrast in the process of ultrasonic imaging logging are addressed.The following is a description of the paper’s primary research:1.Research of first-arrival time extraction method for ultrasonic logging signalIn order to show the benefits,drawbacks,and applicability of the various algorithms,this paper first investigates conventional first arrive-time extraction methods,such as the short and long time window averaging ratio method,the Akaike information theory criterion method,and the mutual correlation method.It then analyses the processing of analogue well signal data containing various levels of Gaussian noise.Furthermore,under the influence of various noise levels and with various scaled data sets,a PhaseNet deep learning network model was developed and applied to the first arrive-time extraction of ultrasonic imaging logging signals.The results were compared with conventional first arrive-time extraction methods.Finally,it is demonstrated that ultrasonic logging based on the PhaseNet network provides the advantages of extracting precision and stability.2.Research on the enhancement method of ultrasonic logging imagesThe combined filtering algorithm of median filtering and bilateral filtering is first compared with the effect of using these two filtering algorithms alone,and the effect of two common reference-free image quality evaluation index functions,Laplacian gradient function and entropy function,is analyzed to show the benefits of the combined filtering algorithm.This comparison is based on two logging images produced by the ultrasonic imaging simulation logging experimental platform.Meanwhile,the processed logging images were compared and analyzed using the two image enhancement techniques,histogram equalization and gamma transform,with the same reference-free image evaluation function being used as an indicator to gauge the impact of the image processing.The results revealed that the histogram equalization method was more successful in processing the logging images.Finally,a comparison of the generated logging images with the real simulated well profile photographs demonstrated the efficacy of the experimental methodology.The PhaseNet network model developed in this research offers higher accuracy and more stable performance than conventional methods,and may achieve good accuracy with minimal data samples,which can have a reference for future ultrasonic imaging logging of the first arrival time extraction.In addition,this work demonstrates the advantages of the combining filtering algorithms and the histogram equalization method for imaging logging processing and interpretation,with the potential to aid in the exploration and production of actual oil resources. |
