You are on page 1of 13

Check Sheet for M. Sc.

Thesis Proposal and Topic Approval

1 Title Page
2 Table of contents
3 Problem Statement
4 Aims & Objectives
5 Literature Review
6 Methodology of Research
a) Theoretical Studies
b) Experimental Set Up
c) Method of Analysis
d) Result Expected
7 Utilization of Research Results
8 Work Schedule Plan
9 Budget Description
10 References
11 Curriculum Vitae
12 Undertaking
13 Board of Post Graduate Studies Remarks

Certified that the Synopsis/Topic approval report is according to the items listed in
Check Sheet.

Research Scholar: Director Postgraduate Studies,


(CPSD)

Name_________________ Name___________________

Signature______________ Signature________________

Date __________________ Date____________________


M.Sc. Thesis Proposal

Deep Learning Based Seismic Impedance Inversion


Optimization to Recover the High-Frequency from a Low-
Frequency Seismic Trace

Submitted By

Irshad Ali
21-MS-DS-02

Supervisor

Dr. Rao Wakeel Ahmad


Lecturer

DEPARTMENT OF COMPUTER SCIENCES


UNIVERSITY OF ENGINEERING AND TECHNOLOGY
TAXILA

November 2022
Table of Contents

Problem Statement 1

Aims & Objectives 1

Literature Review 3

Methodology of Research 5

Work Schedule Plan 6

References 7

Curriculum Vitae 8

Undertaking 9

Supervisor’s Comments 10

Board of Post Graduate Studies Remarks 10


1. Problem Statement

The subsurface attenuates seismic wave propagation. A narrow frequency spectrum


results from this loss, which is frequency dependant. Higher frequencies are
absorbed more quickly than lower frequencies. It is frequently challenging to
pinpoint the precise borders of the reflections since the seismic trace is a
superposition of numerous overlapping reflections. Therefore, a technique to
increase the frequency content of seismic data is required to make delicate geologic
features more discernible. Since the event can look more clearly defined and less
impacted by low frequency noise, enhanced seismic data exhibits a significant
increase in resolution, providing comprehensive reservoir stratigraphy, crisper
pinchouts, and more discernible structures. The majority of the time, however,
frequency augmentation also results in greater noise levels. But in most
circumstances, increasing frequency also results in a rise in noise levels in addition
to increased resolution. The subsurface attenuates seismic wave propagation. A
narrow frequency spectrum results from this loss, which is frequency dependant.
Higher frequencies are absorbed more quickly than lower frequencies. It is
frequently challenging to pinpoint the precise borders of the reflections since the
seismic trace is a superposition of numerous overlapping reflections. Therefore, a
technique to increase the frequency content of seismic data is required to make
delicate geologic features more discernible. The resolution of the enhanced seismic
data has significantly increased, revealing precise reservoir stratigraphy.

2. Aims & Objectives

Seismic impedance inversion, which recovers the high-frequency impedance from a


low-frequency seismic trace, can be thought of as a signal reconstruction challenge
comparable to the image super resolution problem. According to the perceptual
loss, the reconstructed image ought to be comparable to the original in terms of
both pixels and features. The typical approach is to extract the features from various
layers using a pre-trained network, such as VGG-16 (Simonyan and Zisserman, 2014)
[1], and then quantify the content difference by calculating the Euclidean distance
between the true and predicted features. Experimental evidence of the perceptual
loss supports the value of high-frequency information reconstruction.

We need to relate well data and seismic data. We want to be able to relate features
from well data (tops,fault,cored intervals) to the peaks and toughs on the seismic
line that pass through the well location.
1
Similarly, we may want to relate some features on the seismic to the well data. To do
this, we have to develop a depth-to-time relation.

We also want to model how the individual wavelets caused by the various
impedance boundaries around our zone of interest (i.e., reflection coefficients)
combine to give us the resultant seismic trace.

2
3. Literature Review

A comparison of deep learning methods for seismic impedance Inversion[1] i.e.,


Deep learning is widely used for seismic impedance inversion, but few work provides
in-depth research and analysis on designing the architectures of deep neural
networks and choosing the network hyperparameters. It is dedicated to
comprehensively studying on the significant aspects of deep neural networks that
affect the inversion results. We experimentally reveal how network hyperparameters
and architectures affect the inversion performance, and develop a series of methods
which are proven to be effective in reconstructing high-frequency information in the
estimated impedance model. Experiments demonstrate that the proposed multi-
scale architecture is helpful to reconstruct more high-frequency details than a
conventional network. Besides, the reconstruction of high-frequency information
can be further promoted by introducing a perceptual loss and a generative
adversarial network from the computer vision perspective. More importantly, the
experimental results provide
valuable references for designing proper network architectures in the seismic
inversion problem.

Predicting seismic-based risk of lost circulation using machine learning[2]


i.e., Lost circulation during well drilling and completion wastes productive time,
and even kills the well in severe cases. Timely identifying lost circulation events
and taking countermeasures has been the focus of related study. However, a real
prediction of lost circulation risk before drilling would be an active response to
the challenge. In this paper, a technical solution is proposed to evaluate
geological lost-circulation risk in the field using 3D seismic data attributes and
machine learning technique. First, four seismic attributes (variance, attenuation,
3
sweetness, RMS amplitude) that are the most correlated with lost circulation
incidents are recommended. Then a prediction model is built by conducting
supervised learning that involves a majority voting algorithm. The performance
of the model is illustrated by six unseen drilled wells and shows the ability and
potential to forecast lost circulation probability both along well trajectory and in
the region far away from the drilled wells. The prediction resolution in the lateral
and vertical direction is about 25m and 6m (2 ms), respectively, which are distinct
advantages over the traditional description of geological structures using seismic
data. It shows that the lost circulation risk can be hardly recognized by
interpreting one specific seismic attribute, which is a common practice. Finally,
the challenges in predicting lost circulation risk using seismic data are
summarized. Overall, the study suggests that machine learning would be a
practical solution to predict various construction risks that are related to seismic-
based geological issues. Knowing in advance the risks, people could avoid or at
least minimize the losses by optimizing well deployment in the field and taking
preventive measures.

Multi-Task Deep Learning Seismic Impedance Inversion Optimization Based


on Homoscedastic Uncertainty[3] i.e., Seismic inversion is a process to obtain
the spatial structure and physical properties of underground rock formations
using surface acquired seismic data, constrained by known geological laws and
drilling and logging data. The principle of seismic inversion based on deep
learning is to learn the mapping between seismic data and rock properties by
training a neural network using logging data as labels. However, due to high cost,
the number of logging curves is often limited, leading to a trained model with
poor generalization. Multi-task learning (MTL) provides an effective way to
mitigate this problem. Learning multiple related tasks at the same time can
improve the generalization ability of the model, thereby improving the
performance of the main task on the same amount of labeled data. However, the
performance of multi-task learning is highly dependent on the relative weights
for the loss of each task, and manual tuning of the weights is often time-
consuming and laborious. In this paper, a Fully Convolutional Residual Network
(FCRN) is proposed to achieve seismic impedance inversion and seismic data
reconstruction simultaneously, and a method based on the homoscedastic
uncertainty of the Bayesian model is used to balance the weights of the loss
function for the two tasks. The test results on the synthetic datasets of
Marmousi2, Overthrust, and Volve field data show that the proposed method can

4
automatically determine the optimal weight of the two tasks, and predicts
impedance with higher accuracy than single-task FCRN model.

4. Methodology of Research

4.1 Theoretical Studies

Following is the learning curve which may be attained by carrying out this
research:
 Seismic and Well data mapping
 UNET and Deep Learning Model for Image frequency enhancement
 Pre-processing techniques
 Fault tolerance techniques
 All of the previous proposed algorithms
 Applied task scheduling algorithms
 Implementation strategies
 Tools and techniques
 Data acquisition and implementations
 Results omissions
.

4.2 Method of Analysis

 Accuracy
 Precision
5
 F1 Score
 Recall
 Jaccard Index
 AUC

5. Work Schedule Plan


.

Collection of literature Two Weeks


Study of Literature Two Weeks
Analysis of Proposed Scheme One Month
Preparation of Scheme/Model One Month
Implementation of Scheme/Model One Month
Analysis and Simulation One Month
Result Formulation Two Weeks
Final Write-up & Thesis Submission Two Weeks

5.1 Proposed Time Schedule

Jan Mar
NOV Dec Feb
Activity 202 202 Apr 2023
2022 2022 2023
3 3
Collection of Literature

Study of Literature
Analysis of Proposed
Scheme
Preparation of Schemes
/ Model
Implementation of
Schemes/Model
Analysis & Simulation

Result Formulation
Final Write-up & Thesis
Submission
6
6. References

1. Zhang, S. B., Si, H. J., Wu, X. M., & Yan, S. S. (2022). A comparison of deep learning
methods for seismic impedance inversion. Petroleum Science.
2. Zheng, X., Wu, B., Zhu, X., & Zhu, X. (2022). Multi-task deep learning seismic impedance
inversion optimization based on homoscedastic uncertainty. Applied Sciences, 12(3), 1200.
3. Geng, Z., Wang, H., Fan, M., Lu, Y., Nie, Z., Ding, Y., & Chen, M. (2019). Predicting seismic-
based risk of lost circulation using machine learning. Journal of Petroleum Science and
Engineering, 176, 679-688.
4. Wu, J., Shi, Y., & Wang, W. (2022). Fault Imaging of Seismic Data Based on a Modified U-Net
with Dilated Convolution. Applied Sciences, 12(5), 2451.
5. Zhu, W., Mousavi, S. M., & Beroza, G. C. (2019). Seismic signal denoising and decomposition
using deep neural networks. IEEE Transactions on Geoscience and Remote Sensing, 57(11),
9476-9488.
6. Jarvis, K. (2006). Integrating well and seismic data for reservoir characterization: risks and
rewards. ASEG Extended Abstracts, 2006(1), 1-4.
7. Sun, M., Li, Z., Liu, Y., Wang, J., & Su, Y. (2021). Low-Frequency Expansion Approach for
Seismic Data Based on Compressed Sensing in Low SNR. Applied Sciences, 11(11), 5028.

7
7. Curriculum Vitae

Name: IRSHAD ALI


Father’s Name: MUHAMMAD DAWOOD
Registration No: 21-MS-DS-02
Contact: 0322-5005895
Address: Benazir Colony Wah Cantt
E-mail (s): Irshad.Ali@students.uettaxila.edu.pk
Courses Passed:
Subject Exam Held Grade GPA
in
1 Big Data Analytics DS-5104 A- 3.66

2 Data Visulization DS-5115 A- 3.66

3 Machine Leaning DS-5103 Fall-2021 A- 3.66

4 Advanced Computer Vision DS-5108 B+ 3.33

5 Tools and Techniques in Data Science DS-5102 A 4.00

6 Natural Language Processing DS-5106 A- 3.66


Spring-2022
7 Optimization Methods for Data Science and A 4.00
Machine Learning DS-5123

8
8. Undertaking

I certify that research work titled “Deep Learning Based Seismic Impedance Inversion
optimization to Recover the High-Frequency from a Low-Frequency Seismic Trace” is
my own work. The work has not, in whole or in part, been presented elsewhere for
assessment. Where material has been used from other sources it has been properly
acknowledged/referred.

Signature of Student
IRSHAD ALI
21-MS-DS-02

9
9 Supervisor’s Comments

Signature of Supervisor
Dr. Rao Wakeel Ahmad
Lecturer

10. Board of Post Graduate Studies Remarks

10

You might also like