[DOC] Discussion for Background Research #60

Open
opened 2025-05-13 15:20:42 +00:00 by nuluh · 8 comments
nuluh commented 2025-05-13 15:20:42 +00:00 (Migrated from github.com)

Documentation Type

Thesis Chapter/Section

Description

The history for every changes should be documented here with make sense reasons, either from author's thoughts or advisors.

Current State

No response

Proposed Changes

Possible structure section and subsection:

  1. Background

  2. Identification Probem

    1. Most existing SHM frameworks require dense sensor arrays, which are costly and impractical for many field applications.

    2. Many high-performing models rely on deep learning techniques, which demand high computational resources and may not be interpretable or easy to deploy.

    3. There is a lack of simple, generalizable approaches that balance minimal sensor usage with reliable damage localization.

  3. Key Issue

    1. To guide the direction of the study, the following key issues are addressed:

    2. Can vibration signals captured only at the top and bottom sensors of a column group retain sufficient features to accurately classify structural damage?

    3. Does pooling data from multiple column paths enhance model generalization, even with limited sensors per path?

    4. Can a simple classical machine learning algorithm compete with more complex models when applied to this reduced-input scenario?

  4. Problem Limitation

    1. This study is limited to the QUGS dataset, which is based on a specific laboratory steel-frame structure under controlled conditions.

    2. The trained model is not intended for transfer to different simulators or physical structures without retraining.

    3. The machine learning scope is limited to classical algorithms (e.g., SVM), excluding neural networks and deep learning models.

    4. The research does not explore varying types of damage or multi-class classification beyond what the dataset provides.

  5. Purpose

    1. To develop a simplified SHM pipeline that uses only a pair of sensors (top and bottom) from grouped columns of sensors (e.g., columns: (1,6,11,16,21,26)).

    2. To treat each sensor group as a simplified one-dimensional beam and assess whether damage characteristics are preserved in the vibration energy between endpoints.

    3. To append each grouped column as a singular dataset and conduct five separate tests, with each group serving as a validation set in turn.

    4. To include the column group’s own data in the training set, thereby generating one inclusive model for all column groups.

    5. To explore the feasibility of generalizing a single model across multiple column paths using endpoint sensor data.

  6. Novelty

    1. Introduces a new pipeline that reduces sensor dependency by reorganizing the layout into vertical groups and using only the first and last sensors.

    2. Demonstrates that vibration energy across a simplified vertical path retains enough information for effective classification.

    3. Uses cross-group augmentation: each group column contributes to training while also serving as an independent validation path.

    4. Proposes a cost-effective, scalable SHM method suitable for implementation in constrained environments.

    5. Offers an interpretable, classical machine learning-based solution as an alternative to black-box deep learning models.

Documentation Location

latex/chapters/01_introduction.tex

Priority

Critical (required for thesis)

Target Audience

Thesis Committee/Reviewers

References

No response

Additional Notes

No response

### Documentation Type Thesis Chapter/Section ### Description The history for every changes should be documented here with make sense reasons, either from author's thoughts or advisors. ### Current State _No response_ ### Proposed Changes Possible structure section and subsection: 1. Background 2. Identification Probem 1. Most existing SHM frameworks require dense sensor arrays, which are costly and impractical for many field applications. 2. Many high-performing models rely on deep learning techniques, which demand high computational resources and may not be interpretable or easy to deploy. 3. There is a lack of simple, generalizable approaches that balance minimal sensor usage with reliable damage localization. 3. Key Issue 1. To guide the direction of the study, the following key issues are addressed: 2. Can vibration signals captured only at the top and bottom sensors of a column group retain sufficient features to accurately classify structural damage? 3. Does pooling data from multiple column paths enhance model generalization, even with limited sensors per path? 4. Can a simple classical machine learning algorithm compete with more complex models when applied to this reduced-input scenario? 4. Problem Limitation 1. This study is limited to the QUGS dataset, which is based on a specific laboratory steel-frame structure under controlled conditions. 2. The trained model is not intended for transfer to different simulators or physical structures without retraining. 3. The machine learning scope is limited to classical algorithms (e.g., SVM), excluding neural networks and deep learning models. 4. The research does not explore varying types of damage or multi-class classification beyond what the dataset provides. 5. Purpose 1. To develop a simplified SHM pipeline that uses only a pair of sensors (top and bottom) from grouped columns of sensors (e.g., columns: (1,6,11,16,21,26)). 2. To treat each sensor group as a simplified one-dimensional beam and assess whether damage characteristics are preserved in the vibration energy between endpoints. 3. To append each grouped column as a singular dataset and conduct five separate tests, with each group serving as a validation set in turn. 4. To include the column group’s own data in the training set, thereby generating one inclusive model for all column groups. 5. To explore the feasibility of generalizing a single model across multiple column paths using endpoint sensor data. 6. Novelty 1. Introduces a new pipeline that reduces sensor dependency by reorganizing the layout into vertical groups and using only the first and last sensors. 2. Demonstrates that vibration energy across a simplified vertical path retains enough information for effective classification. 3. Uses cross-group augmentation: each group column contributes to training while also serving as an independent validation path. 4. Proposes a cost-effective, scalable SHM method suitable for implementation in constrained environments. 5. Offers an interpretable, classical machine learning-based solution as an alternative to black-box deep learning models. ### Documentation Location latex/chapters/01_introduction.tex ### Priority Critical (required for thesis) ### Target Audience Thesis Committee/Reviewers ### References _No response_ ### Additional Notes _No response_
nuluh commented 2025-05-14 07:41:38 +00:00 (Migrated from github.com)

Identifikasi masalah dihapus, disesuaikan dengan template tanpa identifikasi masalah.

> Identifikasi masalah dihapus, disesuaikan dengan template tanpa identifikasi masalah.
meowndor commented 2025-05-22 02:04:02 +00:00 (Migrated from github.com)
  1. Background
    • Pentingnya SHM
    • "Stage" SHM -> Salah satu stage: untuk mendapatkan lokasi secara akurat
    • Abdeljaber2017:
      • Menggunakan full array sensor
      • Output 30 CNN model
1. Background - Pentingnya SHM - "Stage" SHM -> Salah satu stage: untuk mendapatkan lokasi secara akurat - Abdeljaber2017: - Menggunakan full array sensor - Output 30 CNN model - - -
meowndor commented 2025-05-22 02:04:44 +00:00 (Migrated from github.com)
  • Mencari proof bahwasannya cost adalah salah satu pertimbangan dalam penentuan metode SHM.

Untuk mengefisiensikan proses dalam pengambilan data maupun training tanpa mengurangi akurasi ... -> dilakukan preprocessing, seperti yang dilakukan pada penelitian sebelumnya dengan jumlah sensor yang terbatas. <cite: 10.1007/s13349-023-00715-3> Menununjukkan bahwa tranformasi hilbert huang pada metode VMD-HT-CNN meningkatkan efisiensi waktu training.

  • Address gap research antara Abdeljaber2017 dan 10.1007/s13349-023-00715-3 tanpa "menjatuhkan"
- Mencari proof bahwasannya cost adalah salah satu pertimbangan dalam penentuan metode SHM. > Untuk mengefisiensikan proses dalam pengambilan data maupun training tanpa mengurangi akurasi ... -> dilakukan preprocessing, seperti yang dilakukan pada penelitian sebelumnya dengan jumlah sensor yang terbatas. <cite: 10.1007/s13349-023-00715-3> Menununjukkan bahwa tranformasi hilbert huang pada metode VMD-HT-CNN meningkatkan efisiensi waktu training. - Address gap research antara Abdeljaber2017 dan 10.1007/s13349-023-00715-3 tanpa "menjatuhkan"
meowndor commented 2025-05-22 02:26:09 +00:00 (Migrated from github.com)

Tujuan -> dengan sepasang sensor (yang dilimitasi)
Tujuan -> Pipeline -> Sebaiknya tidak disebut secara ekspilisit karena akan diperlukan pembahasan time complxity Abdeljaber2017 dan mencoba melakukan eksperimen metodologi yang disampaikan oleh Abdeljaber tetapi hanya dengan sensor yang dilimitasi (2 sensor)

Tujuan -> dengan sepasang sensor (yang dilimitasi) Tujuan -> **Pipeline** -> Sebaiknya tidak disebut secara ekspilisit karena akan diperlukan pembahasan time complxity Abdeljaber2017 dan mencoba melakukan eksperimen metodologi yang disampaikan oleh Abdeljaber tetapi hanya dengan sensor yang dilimitasi (2 sensor)
meowndor commented 2025-05-22 02:31:04 +00:00 (Migrated from github.com)

Jumlah point pada rumusan masalah seharusnya sama dengan point tujuan dan kesimpulan (rata-rata dibuat dalam point: tugas akhir atau paragraf: paper.

Jumlah point pada rumusan masalah seharusnya sama dengan point tujuan dan kesimpulan (rata-rata dibuat dalam point: tugas akhir atau paragraf: paper.
meowndor commented 2025-05-22 02:35:58 +00:00 (Migrated from github.com)
  • Lingkup penelitian disampaikan bahwasannya posisi sensor hanya diambil pada ujung-ujung luar (tidak dibagian dalam grid).
  • disampaikan juga limitasi mengenai program/library yang digunakan.
  • Benda uji yang ditinjau untuk penelitian ini diambil dari penelitian yang dilakukan oleh Abdeljaber2017
- Lingkup penelitian disampaikan bahwasannya posisi sensor hanya diambil pada ujung-ujung luar (tidak dibagian dalam grid). - disampaikan juga limitasi mengenai program/library yang digunakan. - Benda uji yang ditinjau untuk penelitian ini diambil dari penelitian yang dilakukan oleh Abdeljaber2017
meowndor commented 2025-05-22 02:42:21 +00:00 (Migrated from github.com)

Manfaat:

  • Menghasilkan metode baru dalam memprediksi lokasi kerusakan dengan sensor yang dilimitasi pada kedua ujung.
  • Metode hanya menghasilkan satu model lokalisasi kerusakan yang digeneralisasi yang dapat memprediksi seluruh lokasi kerusakan (bukan 30 model)
Manfaat: - Menghasilkan metode baru dalam memprediksi lokasi kerusakan dengan sensor yang dilimitasi pada kedua ujung. - Metode hanya menghasilkan satu model lokalisasi kerusakan yang digeneralisasi yang dapat memprediksi seluruh lokasi kerusakan (bukan 30 model)
meowndor commented 2025-05-22 02:48:50 +00:00 (Migrated from github.com)

Dasar Teori:

  • Fungsi windowing dan Hann Window dijadikan satu paragraf. Paragraf kedua dari Hann Window dimulai dari "Pemilihan jenis jendela ... waktu-frekuensi" karena hanya ada satu fungsi windowing yang digunakan.
  • Notasi jangan lupa dimasukkan pada daftar simbol dan harus sinkron/konsisten
  • Memungkinan adana kutipan untuk kelebihan dari dasar teori yang diangkat
  • Fungsi windowing (incl Hann window) dijadikan satu section pada STFT.
  • Notasi/Fungsi matematika (general) untuk setiap algoritma klasifikasi.
Dasar Teori: - Fungsi windowing dan Hann Window dijadikan satu paragraf. Paragraf kedua dari Hann Window dimulai dari "Pemilihan jenis jendela ... waktu-frekuensi" karena hanya ada satu fungsi windowing yang digunakan. - Notasi jangan lupa dimasukkan pada daftar simbol dan harus sinkron/konsisten - Memungkinan adana kutipan untuk kelebihan dari dasar teori yang diangkat - Fungsi windowing (incl Hann window) dijadikan satu section pada STFT. - Notasi/Fungsi matematika (general) untuk setiap algoritma klasifikasi.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: nuluh/thesis#60