Perovskite Solar Cell Stability Analysis Using Entropy‐Based Support Vector Machines Learning

Rupam Bhaduri,S. Manasa
DOI: https://doi.org/10.1002/pip.3861
2024-11-22
Progress in Photovoltaics Research and Applications
Abstract:In this article, we propose the use of machine learning (ML) to enhance the thermal, chemical, and structural stability of PSCs. The process begins with standard pre‐processing steps, including data normalization and data augmentation. Next, feature selection is performed using the modified grasshopper optimization algorithm (MGO) to eliminate redundant or irrelevant features. Finally, a novel machine learning approach employing entropy‐based support vector machines (ESVM) is introduced to classify stability as stable or unstable. Lead halide perovskites have demonstrated significant potential for photovoltaic (PV) applications over the past 10 years. Perovskite solar cells (PSCs) stability, however, continues to limit their commercialization, and the inability to compare previous stability data to assess possible directions for increasing device stability is caused by a lack of effectively established unified stability testing and disseminating standards. In this article, we suggest applying machine learning (ML) to improve the thermal, chemical, and structural stability of PSCs. Data normalization and data augmentation are common preprocessing steps that are where the process starts. Then, using the Modified Grasshopper Optimisation Algorithm (MGO), feature selection techniques are used to remove unnecessary or irrelevant features. Finally, there is a novel machine learning technique that uses support vector machines (ESVM) that are based on entropy to forecast the stability classification of stable/unstable. The proposed reaches an accuracy of 0.99% far better than the proposed methods.
materials science, multidisciplinary,physics, applied,energy & fuels
What problem does this paper attempt to address?