摘要:
The invention concerns a method for automatic detection of planar heterogeneities intersecting the stratification of an environment from images of a borehole wall or developed core samples of said environment, which consists in using an original image defined in a system of axes (X1, Y1, Z1) associated with the borehole whose axis is Z1, said image containing, for a zone, traversed by the bore hole planar heterogeneities consisting of stratification planes (2 to 14) and planar heterogeneities (15 to 23) intersecting the stratification planes. The invention is characterised in that it consists in: determining a dominant orientation of the stratification planes located in at least on part of said original image; filtering the original image to eliminate the planar heterogeneities of the stratification planes (2 to 14) located in the dominant direction; determining on said filtered image at least two contour segments (15' to 23') of the heterogeneities intersecting the stratification planar heterogeneities.
摘要:
The invention concerns a method for characterising the coherence of an environment characteristic measurements, consisting in: taking N types of reference measurements in each observation (Xi) of a reference set; constituting at least a training set with N dimensions containing all said observations (Xi); taking at least N application measurements in each observation (Qi) of an application set; comparing each application observation (x) with all the observations (Xi). The invention is characterised in that it further consists in: constructing for each observation (Xi) a neighbouring domain (Di), said neighbouring domains constituting said training set and an acceptation class (Ca), and defining a degree of appurtenance of the application observations (x) to said acceptation class (Ca). The invention is particularly applicable to the identification of a lithologic facies in an oil field.
摘要:
The invention concerns a method for characterising the coherence of an environment characteristic measurements, consisting in: taking N types of reference measurements in each observation (Xi) of a reference set; constituting at least a training set with N dimensions containing all said observations (Xi); taking at least N application measurements in each observation (Qi) of an application set; comparing each application observation (x) with all the observations (Xi). The invention is characterised in that it further consists in: constructing for each observation (Xi) a neighbouring domain (Di), said neighbouring domains constituting said training set and an acceptation class (Ca), and defining a degree of appurtenance of the application observations (x) to said acceptation class (Ca). The invention is particularly applicable to the identification of a lithologic facies in an oil field.
摘要:
Each of the N surface images, used to restore the borehole or the core sample surface image, is segmented into related components (C (12), C (11),...) which are grouped into related regions (C (8)), each touching both the right and left edges of the image. After smoothing out their contours, the related regions are represented in the form of an image, called bed image, representative of the stratification beds. The contours appearing on the N bed images are matched to construct the bed boundary sinusoïdal curves. The invention is particularly useful in determining stratification of a geological site.
摘要:
Each of the N surface images, used to restore the borehole or the core sample surface image, is segmented into related components (C (12), C (11),...) which are grouped into related regions (C (8)), each touching both the right and left edges of the image. After smoothing out their contours, the related regions are represented in the form of an image, called bed image, representative of the stratification beds. The contours appearing on the N bed images are matched to construct the bed boundary sinusoïdal curves. The invention is particularly useful in determining stratification of a geological site.
摘要:
The invention concerns a method for detecting fractures in logging signals made up of logs of different types (1 to 6) which consists in: selecting a portion of each of said logs (1 to 6) such that all the selected portions have one single common depth interval, one of the selected portions being taken as reference portion (log 1); selecting a parent wavelet function and constructing, from said parent function, a family of wavelet analysing functions according to spatial frequency and depth. For each selected log portion and for each depth, the absolute mean gradient value of the characteristic coefficient of the wavelet transform for the various spatial frequencies is calculated. The peaks of the mean gradient absolute value of the characteristic coefficient are then identified as fractures.