Department of Biosystems Engineering, Faculty of Agriculture, Ferdowsi University of Mashhad, Mashhad, Iran.
Department of Biosystems Engineering, Faculty of Agriculture, Ferdowsi University of Mashhad, Mashhad, Iran.
Bioresour Technol. 2022 Mar;347:126661. doi: 10.1016/j.biortech.2021.126661. Epub 2022 Jan 7.
Bioplastic biodegradation showed varying behavior during the process of biodegradation. The First-order and Gompertz models are the most prevalent models for monitoring biodegradation in an anaerobic digestion (AD) process, which do not suit adequately bioplastics fermentation modeling. This research aimed at studying the kinetics of methane production during AD of starch-based bioplastic by using a large library of non-linear regressions (NLRs) and an artificial neural network (ANN). Although 26 NLR models (25 were outlined in the AD literature + 1 modified by authors) have been analyzed, 9 of them were proper predictors for the whole AD process for methane production. In the end M9, which has been proposed by authors, was selected owing to the simplicity of regression as well as good statistical criteria. Moreover, MLP-ANN could outperform the NLR model and has been selected as the superior model that can define the kinetics of bioplastic AD.
生物塑料的生物降解在生物降解过程中表现出不同的行为。一阶和 Gompertz 模型是监测厌氧消化 (AD) 过程中生物降解的最常用模型，但它们不能充分适用于生物塑料发酵建模。本研究旨在通过使用大量非线性回归 (NLR) 和人工神经网络 (ANN) 来研究淀粉基生物塑料在 AD 过程中甲烷生成的动力学。尽管已经分析了 26 个 NLR 模型（AD 文献中概述了 25 个，作者修改了 1 个），但其中 9 个模型适用于甲烷生产的整个 AD 过程。最后，选择了由作者提出的 M9，因为其回归简单且具有良好的统计标准。此外，MLP-ANN 可以优于 NLR 模型，并被选为可以定义生物塑料 AD 动力学的优越模型。
