nuluh/thesis
1
0
Fork 0
Code Issues 47 Pull Requests Actions Packages Projects Releases 1 Wiki Activity

[DOC] Enhance background research coherence and connectivity #75

New Issue
Open
opened 2025-05-17 14:38:02 +00:00 by meowndor · 8 comments
meowndor commented 2025-05-17 14:38:02 +00:00 (Migrated from github.com)
Copy Link

Documentation Type

Thesis Chapter/Section

Description

Restructure and enhance the background research section to establish clearer connections between problems, research questions, and prior work based on thesis committee feedback. The current background sections exist as somewhat independent segments rather than forming a cohesive narrative that demonstrates how each research problem connects to and builds upon others.

Current State

The thesis currently contains separate background research sections covering:

  1. Time-domain signal processing for accelerometer data
  2. Frequency-domain transformations (STFT)
  3. Machine learning approaches for sensor data
  4. Prior work in similar domains

However, these sections are presented as largely independent topics without sufficient integration or explanation of how they inform each other. My defense committee specifically noted that the background problem statements should "connect each other" and "answer each other" to form a more cohesive research foundation.

Proposed Changes

  1. Create a new structure for the background chapter with explicit connecting sections:

    • Add transition paragraphs between major sections that highlight relationships
    • Develop a "research gap analysis" section that synthesizes findings across all background research
    • Create a visual research map showing how each area relates to others

  2. Rewrite the problem statement section to show progression:

    • How limitations in time-domain analysis necessitate frequency-domain approaches
    • Why traditional signal processing is insufficient without machine learning
    • How prior research questions have evolved and led to my specific research focus

  3. Add explicit cross-references between sections:

    • Reference findings from Section X when discussing approaches in Section Y
    • Show how conclusions from one research area inform methodology in another

  4. Create a synthesis table/matrix showing:

    • How each research problem connects to others
    • Where gaps exist in the literature across different domains
    • How my research questions specifically address these intersectional gaps

  5. Develop a "theoretical framework" section that:

    • Integrates concepts across signal processing and machine learning domains
    • Establishes the connections between seemingly disparate research areas
    • Provides justification for the integrated approach used in this thesis

Documentation Location

latex/chapters/01_background/*

Priority

High (important for understanding)

Target Audience

Thesis Committee/Reviewers

References

  • Defense committee feedback notes from [May 17, 2025]

Additional Notes

This revision is critical for addressing a specific concern raised during my defense. The committee emphasized that a stronger narrative thread must connect the various research areas to demonstrate comprehensive understanding of the problem space.

This will require significant restructuring rather than simple additions. I'll need to:

  1. Review all background sections holistically
  2. Identify natural connection points
  3. Possibly conduct additional literature review to fill gaps between research areas
  4. Develop new diagrams/visualizations to illustrate these connections

Estimated time: 2-3 weeks for comprehensive revision
Will require feedback from advisor after initial restructuring draft

### Documentation Type Thesis Chapter/Section ### Description Restructure and enhance the background research section to establish clearer connections between problems, research questions, and prior work based on thesis committee feedback. The current background sections exist as somewhat independent segments rather than forming a cohesive narrative that demonstrates how each research problem connects to and builds upon others. ### Current State The thesis currently contains separate background research sections covering: 1. Time-domain signal processing for accelerometer data 2. Frequency-domain transformations (STFT) 3. Machine learning approaches for sensor data 4. Prior work in similar domains However, these sections are presented as largely independent topics without sufficient integration or explanation of how they inform each other. My defense committee specifically noted that the background problem statements should "connect each other" and "answer each other" to form a more cohesive research foundation. ### Proposed Changes 1. Create a new structure for the background chapter with explicit connecting sections: - Add transition paragraphs between major sections that highlight relationships - Develop a "research gap analysis" section that synthesizes findings across all background research - Create a visual research map showing how each area relates to others 2. Rewrite the problem statement section to show progression: - How limitations in time-domain analysis necessitate frequency-domain approaches - Why traditional signal processing is insufficient without machine learning - How prior research questions have evolved and led to my specific research focus 3. Add explicit cross-references between sections: - Reference findings from Section X when discussing approaches in Section Y - Show how conclusions from one research area inform methodology in another 4. Create a synthesis table/matrix showing: - How each research problem connects to others - Where gaps exist in the literature across different domains - How my research questions specifically address these intersectional gaps 5. Develop a "theoretical framework" section that: - Integrates concepts across signal processing and machine learning domains - Establishes the connections between seemingly disparate research areas - Provides justification for the integrated approach used in this thesis ### Documentation Location latex/chapters/01_background/* ### Priority High (important for understanding) ### Target Audience Thesis Committee/Reviewers ### References - Defense committee feedback notes from [May 17, 2025] ### Additional Notes This revision is critical for addressing a specific concern raised during my defense. The committee emphasized that a stronger narrative thread must connect the various research areas to demonstrate comprehensive understanding of the problem space. This will require significant restructuring rather than simple additions. I'll need to: 1. Review all background sections holistically 2. Identify natural connection points 3. Possibly conduct additional literature review to fill gaps between research areas 4. Develop new diagrams/visualizations to illustrate these connections Estimated time: 2-3 weeks for comprehensive revision Will require feedback from advisor after initial restructuring draft
nuluh commented 2025-05-26 05:53:26 +00:00 (Migrated from github.com)
Copy Link

Reorganize and Add Key Topics to Background

  • Structural Health Monitoring (SHM):
    • Introduce the importance of SHM and its relevance to this research.
    • Add the "5 stages of SHM," with emphasis on the stage requiring accurate damage location determination.
## Structural Health Monitoring (SHM)

Structural Health Monitoring (SHM) is a critical area of research across various disciplines, including Aerospace, Civil, and Mechanical Engineering. Its primary objective is to ensure the safety and reliability of structures by detecting damage at the earliest possible stage. By combining advanced sensing technologies with real-time data analysis, SHM systems empower engineers to monitor structural integrity, optimize maintenance strategies, and predict a structure's remaining service life.

### The Five Stages of SHM
A robust SHM system follows a hierarchical framework to identify and assess structural damage. According to [3], the five stages of SHM are as follows:

1. **Level 1: Damage Presence**  
   Determines whether any damage exists in the structure. At this stage, sparse vibration measurements may suffice to ascertain the existence of damage.

2. **Level 2: Damage Location**  
   Identifies the geometric location of the damage, whether single or multiple, often requiring the adoption of a structural model for accuracy.

3. **Level 3: Damage Type**  
   Characterizes the nature of the damage, which could include cracks, altered boundary conditions, or changes in structural connectivity.

4. **Level 4: Damage Severity**  
   Quantifies the extent or size of the damage, often through calibrated experiments or models that describe damage effects, such as stiffness reduction or crack length.

5. **Level 5: Damage Prognosis**  
   Predicts the remaining useful life of the structure, relying on comprehensive real-time monitoring and high-fidelity models that describe damage progression over time.

### Challenges in SHM Systems
The higher the level of damage identification, the more demanding the requirements for sensors, algorithms, and model parameters. While **Level 1** might only require sparse data for damage detection, **Level 5** necessitates advanced real-time data acquisition and reliable predictive models. The challenge lies in developing SHM systems that can effectively address multiple stages of damage identification under both normal and catastrophic conditions, such as earthquakes.

By focusing on these five stages, this research seeks to build upon existing methodologies to address the challenges associated with damage diagnosis and location accuracy (Level 2), while optimizing sensor placement and algorithm design for enhanced cost-efficiency.
Reorganize and Add Key Topics to Background - **Structural Health Monitoring (SHM):** - Introduce the importance of SHM and its relevance to this research. - Add the "5 stages of SHM," with emphasis on the stage requiring accurate damage location determination. ```md ## Structural Health Monitoring (SHM) Structural Health Monitoring (SHM) is a critical area of research across various disciplines, including Aerospace, Civil, and Mechanical Engineering. Its primary objective is to ensure the safety and reliability of structures by detecting damage at the earliest possible stage. By combining advanced sensing technologies with real-time data analysis, SHM systems empower engineers to monitor structural integrity, optimize maintenance strategies, and predict a structure's remaining service life. ### The Five Stages of SHM A robust SHM system follows a hierarchical framework to identify and assess structural damage. According to [3], the five stages of SHM are as follows: 1. **Level 1: Damage Presence** Determines whether any damage exists in the structure. At this stage, sparse vibration measurements may suffice to ascertain the existence of damage. 2. **Level 2: Damage Location** Identifies the geometric location of the damage, whether single or multiple, often requiring the adoption of a structural model for accuracy. 3. **Level 3: Damage Type** Characterizes the nature of the damage, which could include cracks, altered boundary conditions, or changes in structural connectivity. 4. **Level 4: Damage Severity** Quantifies the extent or size of the damage, often through calibrated experiments or models that describe damage effects, such as stiffness reduction or crack length. 5. **Level 5: Damage Prognosis** Predicts the remaining useful life of the structure, relying on comprehensive real-time monitoring and high-fidelity models that describe damage progression over time. ### Challenges in SHM Systems The higher the level of damage identification, the more demanding the requirements for sensors, algorithms, and model parameters. While **Level 1** might only require sparse data for damage detection, **Level 5** necessitates advanced real-time data acquisition and reliable predictive models. The challenge lies in developing SHM systems that can effectively address multiple stages of damage identification under both normal and catastrophic conditions, such as earthquakes. By focusing on these five stages, this research seeks to build upon existing methodologies to address the challenges associated with damage diagnosis and location accuracy (Level 2), while optimizing sensor placement and algorithm design for enhanced cost-efficiency. ```
nuluh commented 2025-05-26 05:53:39 +00:00 (Migrated from github.com)
Copy Link

Integrate Key Studies and Gaps

  • Abdeljaber2017 Study:
    • Mention the use of a full array of 30 sensors on QUGS and its resulting 30 CNN models.
    • Highlight the key findings and contributions of Abdeljaber2017.
  • Gap Analysis:
    • Address the gap between Abdeljaber2017 and the newer study (10.1007/s13349-023-00715-3) without undermining the former.
    • Discuss how the newer methodology (VMD-HT-CNN) improves training efficiency and accuracy with fewer sensors.

The study by Abdeljaber et al. (2017) represents a significant milestone in SHM research, leveraging a full array of 30 sensors on QUGS and implementing 30 CNN models to achieve precise damage localization. Their approach highlighted the potential of deep learning in SHM, particularly in terms of accuracy and robustness. However, subsequent studies, such as the one outlined in <cite: 10.1007/s13349-023-00715-3>, have identified opportunities to enhance efficiency by reducing the computational overhead associated with sensor arrays. This gap highlights the need for a more resource-efficient SHM methodology that balances accuracy and cost-effectiveness without undermining the contributions of Abdeljaber et al.

Integrate Key Studies and Gaps - **Abdeljaber2017 Study:** - Mention the use of a full array of 30 sensors on QUGS and its resulting 30 CNN models. - Highlight the key findings and contributions of Abdeljaber2017. - **Gap Analysis:** - Address the gap between Abdeljaber2017 and the newer study (10.1007/s13349-023-00715-3) without undermining the former. - Discuss how the newer methodology (VMD-HT-CNN) improves training efficiency and accuracy with fewer sensors. > The study by Abdeljaber et al. (2017) represents a significant milestone in SHM research, leveraging a full array of 30 sensors on QUGS and implementing 30 CNN models to achieve precise damage localization. Their approach highlighted the potential of deep learning in SHM, particularly in terms of accuracy and robustness. However, subsequent studies, such as the one outlined in <cite: 10.1007/s13349-023-00715-3>, have identified opportunities to enhance efficiency by reducing the computational overhead associated with sensor arrays. This gap highlights the need for a more resource-efficient SHM methodology that balances accuracy and cost-effectiveness without undermining the contributions of Abdeljaber et al.
nuluh commented 2025-05-26 05:54:57 +00:00 (Migrated from github.com)
Copy Link

Cost Efficiency Argument

  • Add a "proof paragraph" emphasizing cost concerns in SHM:

To improve the efficiency of the Structural Health Monitoring (SHM) process, particularly in data collection and training phases, it is essential to consider methods that reduce computational and resource costs without compromising accuracy. For instance, preprocessing techniques can be employed, as demonstrated in prior studies that utilized a limited number of sensors. An example is the research outlined in <cite: 10.1007/s13349-023-00715-3>, which showed that applying Hilbert-Huang Transform (HHT) within the Variational Mode Decomposition-Hilbert Transform Convolutional Neural Network (VMD-HT-CNN) framework significantly reduced training time while maintaining high accuracy. This approach underscores the importance of preprocessing in optimizing SHM methodologies for practical applications.

Cost Efficiency Argument - Add a "proof paragraph" emphasizing cost concerns in SHM: > To improve the efficiency of the Structural Health Monitoring (SHM) process, particularly in data collection and training phases, it is essential to consider methods that reduce computational and resource costs without compromising accuracy. For instance, preprocessing techniques can be employed, as demonstrated in prior studies that utilized a limited number of sensors. An example is the research outlined in <cite: 10.1007/s13349-023-00715-3>, which showed that applying Hilbert-Huang Transform (HHT) within the Variational Mode Decomposition-Hilbert Transform Convolutional Neural Network (VMD-HT-CNN) framework significantly reduced training time while maintaining high accuracy. This approach underscores the importance of preprocessing in optimizing SHM methodologies for practical applications.
nuluh commented 2025-05-26 05:55:05 +00:00 (Migrated from github.com)
Copy Link

Research Purpose and Scope

  • Refine research purpose to focus on:
    • Experiments with a limited sensor approach.
    • Re-examining Abdeljaber2017 methodology but with fewer sensors.
    • Avoid explicitly framing it as "creating a pipeline."
  • Clearly define the scope:
    • Position of sensors limited to the top-end and bottom-end along the path (no mixed placement with middle sensors).
    • Libraries, programs, and dataset details (e.g., QUGS from Abdeljaber2017).

The purpose of this research is to explore the feasibility of implementing a limited-sensor approach to Structural Health Monitoring (SHM), revisiting the methodology proposed by Abdeljaber et al. (2017) while adapting it for fewer sensors. The scope of this study is confined to using sensor placements at the top-end and bottom-end of the structure, avoiding mixed placements in the middle. Additionally, this research will exclusively utilize the QUGS dataset and leverage well-established tools and libraries to ensure consistency with prior work.

Research Purpose and Scope - Refine research purpose to focus on: - Experiments with a limited sensor approach. - Re-examining Abdeljaber2017 methodology but with fewer sensors. - Avoid explicitly framing it as "creating a pipeline." - Clearly define the scope: - Position of sensors limited to the top-end and bottom-end along the path (no mixed placement with middle sensors). - Libraries, programs, and dataset details (e.g., QUGS from Abdeljaber2017). > The purpose of this research is to explore the feasibility of implementing a limited-sensor approach to Structural Health Monitoring (SHM), revisiting the methodology proposed by Abdeljaber et al. (2017) while adapting it for fewer sensors. The scope of this study is confined to using sensor placements at the top-end and bottom-end of the structure, avoiding mixed placements in the middle. Additionally, this research will exclusively utilize the QUGS dataset and leverage well-established tools and libraries to ensure consistency with prior work.
nuluh commented 2025-05-26 05:55:17 +00:00 (Migrated from github.com)
Copy Link

Address Novelty and Methodology

  • Highlight the novelty:
    • Propose a generalized model capable of detecting all 30 localized damages with a single methodology (not 30 separate models).
  • Develop a "theoretical framework" that integrates concepts across SHM, signal processing, and machine learning.

The novelty of this research lies in the development of a generalized SHM model capable of detecting all 30 localized damages with a single methodology, rather than requiring separate models for each sensor. This approach integrates concepts from signal processing and machine learning to propose a unified framework that is both efficient and scalable. By adapting existing methodologies and extending their applicability, this work aims to bridge the gap between resource constraints and high-accuracy SHM solutions.

Address Novelty and Methodology - Highlight the novelty: - Propose a generalized model capable of detecting all 30 localized damages with a single methodology (not 30 separate models). - Develop a "theoretical framework" that integrates concepts across SHM, signal processing, and machine learning. > The novelty of this research lies in the development of a generalized SHM model capable of detecting all 30 localized damages with a single methodology, rather than requiring separate models for each sensor. This approach integrates concepts from signal processing and machine learning to propose a unified framework that is both efficient and scalable. By adapting existing methodologies and extending their applicability, this work aims to bridge the gap between resource constraints and high-accuracy SHM solutions.
nuluh commented 2025-05-26 05:55:27 +00:00 (Migrated from github.com)
Copy Link

Restructure Problem Statement

  • Show progression between topics:
    • How limitations in time-domain analysis necessitate frequency-domain approaches.
    • Why traditional signal processing is insufficient without machine learning.
    • How prior research questions have evolved and led to my specific focus.

The limitations of time-domain analysis in capturing complex structural dynamics highlight the necessity of frequency-domain approaches. Traditional signal processing methods often fall short in handling the intricacies of SHM data, paving the way for the integration of machine learning techniques. This progression from conventional methods to advanced computational approaches reflects the evolution of research questions that underpin this study, ultimately focusing on efficient and accurate damage localization with minimal sensor utilization.

Restructure Problem Statement - Show progression between topics: - How limitations in time-domain analysis necessitate frequency-domain approaches. - Why traditional signal processing is insufficient without machine learning. - How prior research questions have evolved and led to my specific focus. > The limitations of time-domain analysis in capturing complex structural dynamics highlight the necessity of frequency-domain approaches. Traditional signal processing methods often fall short in handling the intricacies of SHM data, paving the way for the integration of machine learning techniques. This progression from conventional methods to advanced computational approaches reflects the evolution of research questions that underpin this study, ultimately focusing on efficient and accurate damage localization with minimal sensor utilization.
nuluh commented 2025-05-26 05:55:37 +00:00 (Migrated from github.com)
Copy Link

Add Visuals and Synthesis

  • Create diagrams/visualizations:
    • Research map showing relationships between areas.
    • A synthesis table/matrix illustrating:
      • Connections and gaps in literature.
      • How research questions address intersectional gaps.

To better illustrate the relationships and gaps within the SHM research domain, this study will include visual aids such as research maps and synthesis tables. These tools will clarify how various methodologies connect and where unresolved challenges persist. For instance, a synthesis matrix will outline the intersections of existing literature, highlighting how this research addresses critical gaps in resource-efficient SHM with minimal sensors.

Add Visuals and Synthesis - Create diagrams/visualizations: - Research map showing relationships between areas. - A synthesis table/matrix illustrating: - Connections and gaps in literature. - How research questions address intersectional gaps. > To better illustrate the relationships and gaps within the SHM research domain, this study will include visual aids such as research maps and synthesis tables. These tools will clarify how various methodologies connect and where unresolved challenges persist. For instance, a synthesis matrix will outline the intersections of existing literature, highlighting how this research addresses critical gaps in resource-efficient SHM with minimal sensors.
nuluh commented 2025-05-26 05:55:43 +00:00 (Migrated from github.com)
Copy Link

Align Key Issues with Research Purpose and Conclusion

  • Ensure key issues align with the research purpose and conclusion.
  • Use numbered bullet points to structure these sections clearly (as advised for bachelor theses).
Align Key Issues with Research Purpose and Conclusion - Ensure key issues align with the research purpose and conclusion. - Use numbered bullet points to structure these sections clearly (as advised for bachelor theses).
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
bug
Something isn't working
 critical
Block Progress
 documentation
Improvements or additions to documentation
 documentclass
Improvements or additions to LaTeX documentclass
 duplicate
This issue or pull request already exists
 enhancement
New feature or request
 good first issue
Good for newcomers
 help wanted
Extra attention is needed
 invalid
This doesn't seem right
 latex
Related to the writings
 question
Further information is requested
 wontfix
This will not be worked on
No Label documentation enhancement latex
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
nuluh
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: nuluh/thesis#75
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?