Khan Asif Irshad, Almalaise Alghamdi Abdullah S, Abushark Yoosef B, Alsolami Fawaz, Almalawi Abdulmohsen, Marish Ali Abdullah
Computer Science Department, Faculty of Computing and Information Technology, King Abdulaziz University, Jeddah, 21589, Saudi Arabia.
Department of Information Systems, Faculty of Computing and Information Technology, King Abdulaziz University, Jeddah, 21589, Saudi Arabia; Information Systems Department, HECI School, Dar Alhekma University, Jeddah, Saudi Arabia.
Chemosphere. 2022 Nov;307(Pt 3):136044. doi: 10.1016/j.chemosphere.2022.136044. Epub 2022 Aug 14.
The growth and implementation of biofuels and bioenergy conversion technologies play an important part in the production of sustainable and renewable energy resources in the upcoming years. Recycling sources from waste could efficiently ease the risk of world source strain. The waste classification was a good resolution for separating the waste from the recycled objects. It is inefficient and expensive to rely solely on manual classification of garbage and recycling sources. Convolutional neural networks (CNNs) have lately been used to classify recyclable waste, and this is the primary way for recycling the waste. This study presents a recycling waste classification using emperor penguin optimizer with deep learning (RWC-EPODL) model for bioenergy production. RWC-EPODL model focuses on recycling waste materials recognition and classification. When it comes to detecting and classifying trash, the RWC-EPODL model uses two stages. At the initial stage, the RWC-EPODL model uses AX-RetinaNet model for the recognition of waste objects. In addition, Bayesian optimization (BO) algorithm is applied as hyperparameter optimizer of the AX-RetinaNet model. Following the EPO algorithm with a stacked auto-encoder (SAE) model, the EPO algorithm is used to fine-tune the parameters of the SAE technique for trash classification. The RWC-EPODL model's experimental validation is examined through a number of studies. The RWC-EPODL approach has a 98.96 percent success rate. The comparative result analysis reported the better performance of the RWC-EPODL model over recent approaches.
生物燃料和生物能源转换技术的发展与应用在未来几年可持续和可再生能源资源的生产中发挥着重要作用。从废物中回收资源可以有效缓解世界资源紧张的风险。垃圾分类是将废物与可回收物分离的有效方法。仅依靠人工对垃圾和回收资源进行分类效率低下且成本高昂。卷积神经网络（CNN）最近已被用于对可回收废物进行分类，这是废物回收的主要方式。本研究提出了一种用于生物能源生产的基于帝企鹅优化器与深度学习的回收废物分类（RWC-EPODL）模型。RWC-EPODL模型专注于回收废物材料的识别和分类。在检测和分类垃圾时，RWC-EPODL模型采用两个阶段。在初始阶段，RWC-EPODL模型使用AX-RetinaNet模型来识别废物对象。此外，贝叶斯优化（BO）算法被用作AX-RetinaNet模型的超参数优化器。在使用堆叠自动编码器（SAE）模型的EPO算法之后，EPO算法用于微调SAE技术的参数以进行垃圾分类。通过多项研究对RWC-EPODL模型进行了实验验证。RWC-EPODL方法的成功率为98.96%。比较结果分析表明，RWC-EPODL模型的性能优于最近的方法。
