EFFICIENT METHOD FOR HIGH RESOLUTION IMAGING AND RECONNAISSANCE OF BURIED SUBSURFACE PIPELINE AND OTHER INFRASTRUCUTRE USING ABOVE SURFACE GEOPHYSICAL SENSORS AND ARTIFICIAL INTELLIGENCE

Abstract

This invention relates to a method and system for the three-dimensional reconstruction of material properties of a target using remotely located physical sensors using deep learning artificial intelligence. Utilizing the special technique disclosed here, the method provides an order of magnitude improvement in computational speed and memory requirements over current state-of-the-art artificial intelligence-based systems. The improvement in accuracy and resolution obtained enables high resolution imaging of buried infrastructure using above ground sensors mounted on drones and other devices. This makes it feasible to perform several first order inspection tasks related to pipeline health, corrosion, integrity, and others that are currently only possible using inline inspection tools such magnetic flux leakage (MFL) and ultrasonic (UT) sensors or via deployment of fiber optic sensors.

Description

FIELD OF THE INVENTION

This invention relates to a method and system for the three-dimensional reconstruction of material properties of a target using remotely located physical sensors. The sensors can illuminate the target using an active transmitter source such as one of acoustic, electromagnetic, or some other origin, while recording the response from the target in receivers placed in suitable locations. Alternatively, the receivers can record the target response in the presence of a passive source such as gravitational attraction and its gradients, the geomagnetic field, the magneto telluric field and others. Utilizing the special technique disclosed here, the method can provide an order of magnitude improvement in computational speed and memory requirements over current state-of-the-art artificial intelligence-based systems. When compared against state-of-the-art methods that do not rely on artificial intelligence, the current method provides improvement in accuracy and resolution that enables high resolution imaging of buried infrastructure using above ground sensors mounted on drones and other devices possible. This level of improvement makes it feasible to perform several first order inspection tasks related to pipeline health, corrosion, integrity, and others that are currently only possible using inline inspection tools such magnetic flux leakage (MFL) and ultrasonic (UT) sensors.

BACKGROUND OF THE INVENTION

Three-dimensional image reconstruction using remote sensing sensors is a ubiquitous practice that cuts across many applications and industries ranging from the medical, oil and gas, mining, military, civil and environmental engineering, among others. The method uses physics-based algorithms to simulate the response of the target and its surroundings in the presence of an inducing field of electromagnetic, gravitational, seismic, ultrasonic, or some other origin and uses data optimization algorithms to find the material properties that can simulate a response that matches the recorded response by the receivers most closely.

The number of receivers recording the response is usually far fewer than the number of elements required to successfully simulate the observed response, leading to an underdetermined system with a non-unique (more than one) material property distribution that could potentially simulate the response observed by the sensors. This requires the imposition of certain a priori constraints on the nature of distribution of the material properties that are used to “match” the observed sensor response. In many geologic situations of increasing commercial interest, such constraints often lead to poorly reconstructed images which may not represent the subsurface at reliable levels of accuracy and/or resolution.

A key benefit of introduction of machine learning approaches to such efforts is the removal of explicit mathematical constraints on the distribution of the target material properties. Machine learning methods aim to “train” the system to “learn” the response of various material property realizations of the subsurface and then determine the “best” distribution of material property given the input of the observed sensor response. It has been observed that where the deployment of machine learning algorithms is technically, logistically, and commercially feasible, there is a step change improvement in the resolution and accuracy of the reconstructed image/material properties.

The major bottlenecks to such methods are twofold: 1) the large volume of simulations that need to be generated to accurately represent a “universe” of potential candidates that may represent the subsurface material property distribution. 2) The large memory consumption of the simulated models when being called for “training” by the machine learning algorithm. This effectively prevents the usage of machine learning algorithms for many problems of practical interest.

SUMMARY OF THE INVENTION

Most state-of-the-art deep learning machine learning architectures used to address image reconstruction issues follow the blueprint of dividing the image domain into several small pixels which are mathematically represented as two- or three-dimensional matrices. The input data is also cast into a matrix whose format is similar to the target image domain. A series of machine learning layers are introduced between the input data and the target or output image. Each of these layers comprise a set of smaller matrices which are then mathematically combined with a set of weights that help transform the values of the input data matrix to the output image matrix.

The two- and three-dimensional nature of the input and output matrix combined with the similar dimensions of the smaller matrices in the intermediate layers make this process memory intensive and is a key barrier for the solution of very large-scale imaging problems in a commercially effective manner.

The conventional method of image reconstruction that does not deploy machine learning methods, frequently stores this matrix as a one-dimensional vector and can map the input data to the dimensions of the output image by utilizing an adjoint operator. Given this transformation occurs at an intermediate step of a process that does not utilize artificial intelligence, the concept of utilizing the data post adjoint transformation as the initial input for machine learning is novel and not practiced anywhere. By making this change, the computational footprint of the image reconstruction problem is dramatically reduced by one or two major dimensions which then translates into order of magnitude savings in computation time and cost without compromising the accuracy and resolution gains made with machine learning methods.

Additionally, the method provides an easy method for designing machine learning algorithms for unstructured meshes, where the description of the images into clear cut divisions of U-, V-, and W-pixel units along each of the coordinate axes, x-, y-, and z- are not possible.

When applied to above ground magnetic sensor data, the method was able to image the top of pipe confidently at 1-1.5 m below ground surface with variation in susceptibility along pipe axis suggestive of changes in thickness, corrosion, and other issues. The uncertainty in resolving the top of the structure is thus within 50 cm. In comparison, the conventional state-of-the-art least squares inversion algorithm was able to provide an uncertainty bound of location of a pipe-like body somewhere between 0-3 m and not much else. This implies an uncertainty in resolution of the top of the structure of 3 m. This observation suggests a conservative estimate of 6-fold and more improvement in resolution of the pipe-like structure using artificial intelligence compared to conventional state-of-the-art least squares inversion.

BRIEF DESCRIPTION OF THE DRAWINGS AND FIGURES

FIG. 1. A simplified sketch of a drone based geophysical data acquisition system for acquiring data over buried pipelines and other infrastructure in an oilfield setting.

FIG. 2. A schematic representation of a conventional deep machine learning architecture for reconstructing 2-D and 3-D images and/or material property inversion using remote sensors.

FIG. 3. A schematic representation of deep machine learning architecture for reconstructing 3-D images and/or material property inversion using remote sensors using 1-D vector basis functions only.

FIG. 4. a. Map showing survey location for the Washington-on-Brazos case study in Texas. b. Orientation and extent of surveyed area. c. Handheld magnetometers used in the survey.

FIG. 5. a. Line depicting the vertical cross section right across the heart of the anomalies observed in the absolute amplitude map. b. Conventional least squares inversion results displayed in vertical cross section right across the heart of the 4 circular anomalies observed in FIG. 5a. c. The depth, and susceptibility distribution of the pipe is delineated and much more clearly visualized relative to conventional least squares inversion in FIG. 5.b.

FIG. 6. Threshold (0.01-0.06) value of normalized absolute susceptibility values from deep learning AI inversion. Smoothing applied for visual clarity. The reconstructed 3D image of the pipe like structure is enhanced using Al based inversion.

DETAILED DESCRIPTION OF THE INVENTION

FIG. 1 shows the general scheme for acquiring above ground geophysical sensor data using an airborne device (1) like drone. While being shown for illustrative purposes, such data can be collected by multiple other means, including but not limited to helicopters, airplanes, ground borne vehicles, handheld devices, as also towed by boat, as a submarine device, for subsea pipelines amongst others. The collected sensor data is processed to remove the influence of above ground metallic infrastructure (4 in FIG. 1) and the influence of the deeper subsurface geology (depicted by 2). The residual field is trained various potential buried subsurface pipe location, geometry, and states of material property such as magnetic susceptibility, electrical conductivity, sonic/ultrasonic velocities, amongst others to determine optimal location, geometry, and effective material property distribution to infer pipeline health, integrity, and other issues.

Referring to FIG. 2, a simple generic training architecture for current state-of-the-art deep multi-layer machine learning algorithm is shown for illustrative purposes. The input data, fed in the form of an N×R matrix, where N>1 and R≥1, present in the first layer depicted as 5, is processed by a set of mathematical operators present in the first hidden layer, depicted as 6, and its output matrix whose shape is P×Q sent to the second hidden layer, depicted as 7, wherein the shape of the output matrix is transformed to J×K×L. Eventually, these transformations will result in an output matrix whose dimensions will be the same as the desired output image (U×V×W). Based on the differences between the pixel values of the output matrix and those images used as ground truth for training, the system will continue to iterate until the difference between the pixel values of the predicted image and the ground truth are below a certain predetermined threshold and/or subsequent iterations do not alter this difference much.

In FIG. 3, the modification to this approach is discussed. The adjoint operator can be used to reproject the input data 10 to the same dimensions as the output image matrix 15 and recast as a vector, 11. Now, all the processing steps (12-15) are simplified as vector operations instead of matrices which reduce the overall computational footprint by about an order of magnitude.

While either approach is suitable for handling buried subsurface infrastructure like pipelines, the embodiment discussed in FIG. 3 is more efficient.

A practical demonstration of the method using data acquired by Texas A & M university students under the guidance of Prof. Mark Everett using handheld magnetic sensors is disclosed. FIG. 4a shows the location of Washington-on-Brazos State historic site where the data was acquired. FIG. 4b shows the dimensions of the field survey area and FIG. 4c shows the illustration of the magnetic sensor used for the data acquisition.

After suitable processing of data as discussed above, the results of inversion using conventional least squares method is shown in FIG. 5b and the corresponding values of relative susceptibility distribution using deep learning artificial intelligence is shown in FIG. 5c. For ease of comparison, both FIGS. 5b. and 5c. are displayed using the same color scale range. While the conventional inversion hints at the presence of a buried pipe somewhere between 0-3 m below the ground surface, the relative changes in susceptibility distribution results from the Al based inversion demonstrates a confident top of pipe at 1-1.5 m below ground surface with variation in susceptibility along pipe axis suggestive of changes in thickness, corrosion, and other issues, implying a conservative estimated 6-fold improvement.

In FIG. 6, a threshold (0.01-0.06) 3D distribution of the absolute values of relative susceptibility distribution are shown. The full 3D geometry of the pipe, its susceptibility and potential thickness variations are now visible. Such information is not easily inferenced otherwise.

While the present invention has been described in terms of particular embodiments and applications, in both summarized and detailed forms, it is not intended that these descriptions in any way limit its scope to any such embodiments and applications, and it will be understood that many substitutions, changes and variations in the described embodiments, applications and details of the method and system illustrated herein and of their operation can be made by those skilled in the art without departing from the spirit of this invention.

Claims

1. A novel physics-based formulation of the input data from remote sensing imaging sensors that enable the deployment of one-dimensional or low valued two-dimensional vector based deep machine learning architectures for multidimensional (3D & 4D) image reconstruction tasks and solving of inverse problems.

2. Enable the solution of claim 1) for both structured and unstructured mesh.

3. The resulting image from claim 1) and claim 2) is suitable for performing several interpretation monitoring, and reconnaissance tasks on subsurface pipelines that are typically performed by inline inspection devices in both onshore and offshore under water settings.

4. Imaging from claim 1) and claim 2) is suitable for identification of metallic objects located in the subsurface that could potential impede or cause safety hazards for development and construction projects.

5. The resulting image from claim 1) and claim 2) is also suitable to identify pipeline intersections as well as unknown abandon pipelines prior to new pipeline construction or repair of existing pipelines.

6. Imaging from claim 1) and claim 2) can assist in locating subsurface pipeline depth as well as pipelines that may be located beneath an existing pipeline sometimes run in line with one above another.

7. The resulting image from claim 1) and claim 2) can be used prior to traditional inline inspection devices to target areas of concern by identification of areas of corrosion and wall weakness.

8. Imaging from claim 1) and claim 2) supports the detection and location of leaks and breaches in pipeline integrity.