Integrating mechanistic and deep learning models for accurately predicting the enrichment of polyhydroxyalkanoates accumulating bacteria in mixed microbial cultures. (January 2022)

• Record Type: Journal Article
• Authors: Xu, Run-Ze; Cao, Jia-Shun; Luo, Jing-Yang; Feng, Qian; Ni, Bing-Jie; Fang, Fang
• Is Part Of: Bioresource technology. Volume 344:Part B(2022)
• Publication Date: 2022-01
• Subjects: Recurrent neural networks (RNN) -- Long short-term memory (LSTM) -- Activated sludge model (ASM) -- Polyhydroxyalkanoates (PHA) -- Microbial enrichment; Biomass -- Periodicals; Biomass energy -- Periodicals; Bioremediation -- Periodicals; Agricultural wastes -- Periodicals; Factory and trade waste -- Periodicals; Organic wastes -- Periodicals; Bioénergie -- Périodiques; Déchets agricoles -- Périodiques; Déchets industriels -- Périodiques; Déchets organiques -- Périodiques; Déchets (Combustible) -- Périodiques; 662.88
• DOI: 10.1016/j.biortech.2021.126276
• Languages: English
• ISSNs: 0960-8524

Abstract

Graphical abstract: Highlights: An integrated approach coupling mechanistic and deep learning models was established. Time-dependent data was generated by ASM3 models for training RNN models. Dynamic change of PAB was accurately predicted by the trained RNNs models. Equilibrium concentration of PAB and enrichment time can be predicted. Time interval of ASM3 data greatly affects the performance of RNNs models. Abstract: The enrichment of polyhydroxyalkanoates (PHA) accumulating bacteria (PAB) in mixed microbial cultures (MMC) is extremely difficult to be predicted and optimized. Here we demonstrate that mechanistic and deep learning models can be integrated innovatively to accurately predict the dynamic enrichment of PAB. Well-calibrated activated sludge models (ASM) of the PAB enrichment process provide time-dependent data under different operating conditions. Recurrent neural network (RNN) models are trained and tested based on the time-dependent dataset generated by ASM. The accurate prediction performance is achieved ( R 2 > 0.991) for three different PAB enrichment datasets by the optimized RNN model. The optimized RNN model can also predict the equilibrium concentration of PAB ( R 2 = 0.944) and corresponding time, which represents the end of the PAB enrichment process. This study demonstrates the strength of integrating mechanistic and deep learning models to predict long-term variations of specific microbes, helping to optimize their selection process for PHA production.