Modeling the Biokinetics of Autothermal Thermophilic Aerobic Digestion: Application of Integrated Multi-layer Perceptron Tuned with Nature-Inspired Technique

The Autothermal Thermophilic Aerobic Digestion (ATAD) process stands as a modern wastewater treatment method used for sludge digestion. This process uses thermophilic microbes, such as Thermotogaceae and Clostridiaceae, to break down organic waste naturally without needing external energy sources. This makes it a good way to stabilize the waste, by degrading it into simpler biochemicals through endogenous respiration. Despite its effectiveness, this process is complex, involving microbiological, thermodynamic, biochemical, and kinetic intricacies. This study delves into the kinetics aspect, acknowledging the potential operational setbacks arising from inadequate comprehension. The investigation emphasizes the intricate nature of the biokinetics, encompassing diverse bioreactions related to microorganism growth and sludge digestion. Recognizing the challenges of classical modeling approaches, the paper advocates for artificial neural networks (ANNs) as a promising alternative, citing their ability to handle complex and non-linear data structures. The study used kinetic data to construct an optimized ANN model predicting the kinetic rate constant (KATAD). The model was further tuned with the genetic algorithm (GA), which is a well-known nature-inspired optimizer, to demonstrate exceptional accuracy (more than ۹۹%). Model evaluation using causal index (CI) showed that temperature (TATAD) was the most influential parameter (CI = ۱.۲۳), followed by the primary to secondary sludge ratio (P/S) parameter (CI = -۰.۴۷), and secondary sludge concentration (Cs) with the least impact on KATAD (CI = ۰.۱۹). This research presents a novel study exploring the kinetics of the recently developed ATAD technology. Moreover, it used a cutting-edge approach to tackle the complexities of ATAD. This work advances our understanding of ATAD kinetics and lays the groundwork for improved wastewater treatment strategies. The successful application of the ANN-GA model paves the way for more accurate and effective treatment processes.

Keywords:

Artificial neural network , Autothermal Thermophilic Aerobic Digestion , genetic algorithm , Hybrid Modeling , Kinetic Rate Constant , Wastewater treatment