| DC 欄位 |
值 |
語言 |
| DC.contributor | 化學學系 | zh_TW |
| DC.creator | 張詔傑 | zh_TW |
| DC.creator | Zhao-Jie zHANG | en_US |
| dc.date.accessioned | 2023-8-17T07:39:07Z | |
| dc.date.available | 2023-8-17T07:39:07Z | |
| dc.date.issued | 2023 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:444/thesis/view_etd.asp?URN=110223073 | |
| dc.contributor.department | 化學學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 染料敏化太陽能電池（DSSC）由於其成本效益、靈活性和理想的穩定性而作為
一種有前景的光伏技術引起了廣泛關注。然而，由於染料分子結構和元件性能之
間的複雜關係，設計高效、穩定的 DSSC 染料敏化劑仍然具有挑戰性。在這項
研究中，我們提出了一種專門為鋅基的紫質染料敏化劑量身定製的可靠且可解釋
的定量結構-性質關係（QSPR）模型。通過將機器學習技術(ML)與密度泛函理論
(DFT) 計算相結合，我們構建了一個包含 140 個分子的資料集。 ML 模型經過
訓練來預測功率轉換效率 (PCE)，並使用 Shapley Additive Explanations Theory 進
一步解釋其預測。開發的模型表現出卓越的準確性，通過 10 折交叉驗證並利用
bagging 技術，均方根誤差 (RMSE) 達到 1.09%。利用這個模型，我們對來自眾
所周知且容易獲得的供體和受體的大量分子進行了計算機虛擬篩選。結果，我們
使用這種方法成功鑑定了九種有前景的高 PCE 鋅基的紫質染料。此外，使用
Shapley Additive Explanations Theory 對預測模型的解釋使我們能夠推導出有意義
的化學規則，這有助於製定 DSSC 實際應用的鋅基紫質染料分子設計原則。 | zh_TW |
| dc.description.abstract | Dye-sensitized solar cells (DSSCs) have attracted significant attention as a
promising photovoltaic technology due to their cost-effectiveness, flexibility, and
desirable stability. However, designing efficient and stable DSSC dye sensitizers
remains a challenge due to the sophisticated relationship between molecular structure
and device performance. In this study, we propose a reliable and interpretable
quantitative structure-property relationship (QSPR) model specifically tailored for
zinc-based porphyrin sensitizers. By combining machine learning technique with
density functional theory (DFT) calculations, we constructed a dataset comprising 140
data. The ML model was trained to predict the power conversion efficiency (PCE) and
its predictions were further interpreted using the Shapley Additive Explanations Theory.
The developed model demonstrated remarkable accuracy with a root mean square error
(RMSE) of 1.09% achieved through 10-fold cross-validation and utilizing bagging
technique. Leveraging this model, we performed in silico virtual screening of a large
number of molecules derived from well-known and readily available donors and
acceptors. As a result, we successfully identified nine promising zinc-based porphyrin
dyes with high PCE using this approach. Additionally, the interpretation of the
prediction model using the Shapley Additive Explanations Theory allowed us to deduce
III
meaningful chemical rules, which can contribute to the formulation of design principles
for practical applications of DSSCs utilizing zinc-based porphyrin dyes. | en_US |
| DC.subject | 紫質 | zh_TW |
| DC.subject | 機器學習 | zh_TW |
| DC.subject | 染料敏化太陽能電池 | zh_TW |
| DC.subject | 理論計算 | zh_TW |
| DC.subject | 密度泛函理論 | zh_TW |
| DC.subject | Porphyrin | en_US |
| DC.subject | Machine Learning | en_US |
| DC.subject | Dye-Sensitized Solar Cells | en_US |
| DC.subject | theoretical calculation | en_US |
| DC.subject | DFT | en_US |
| DC.title | 透過機器學習加速探索有前景的紫質染料用於染料敏化太陽能電池 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Expedited Exploration of Prospective Porphyrin Dye for Dye-Sensitized Solar Cells by Machine Learning | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |