Water quality and resources scarcity are two problems facing the world today due to the geographical, anthropological and socioeconomic influences [1, 2]. Accordingly, sustainable recovery and reuse of the wastewater resources have long been considered as the promising approaches to overcome these problems. However, the presence of a broad range of hazardous compounds in wastewater, comprising pesticides, polycyclic aromatic hydrocarbons, polyphenolic compounds, oil, surfactants, and nitroaromatics, have made it more difficult to develop strategies with desirable efficiency to remove these contaminants [3, 4].
Among the afore-mentioned contaminants nitroaromatic compounds (NAC) are considered as one of the major sources of the contamination in wastewater resources that can induce serious problems if not removed properly [5, 6]. These compounds are a particular class of aromatic molecules with an at least a nitro group (-NO2) at their benzene ring. It has been demonstrated that NACs have widely been distributed in the environment [7, 8], and 1,3,5-trinitrobenzene (TNB), 2,4,6-trinitrotoluene (TNT), trinitrophenol, tetryl nitramine, hexanitrobenzene are among the well-known NACs [1].
TNT (2,4,6-trinitrotoluene) is a multifunctional aromatic substance used in wide variety of fields including medicines, insecticides, fungicides, herbicides, polyurethane foams, and dyes. Besides, it is among the most traditional explosives still utilized in mining industry [8, 9]. The TNT production process encompasses several steps generating waste materials that are eventually released into the surrounding environment and could potentially contaminate water resources and soil. Mono-nitro toluene (MNT), di-nitro toluene (DNT), sulfates, dinitro toluene sulfonate (DNTS), and various other nitrobenzenes (NB) are also the major derivatives of the TNT with potential health risks. It is well-known that varying quantities of these nitrogenous substances are present in TNT-contaminated wastewater and responsible for the discoloration (e.g., orange water, red water, yellow water, etc.) [5]. TNT-polluted wastewater has also been reported with high chemical oxygen demand (COD) levels, varying from 600 to 6000 mg/L [10, 11]. In accordance, the presence of TNT and other related aromatic chemicals in wastewater were shown to massively affect the regional ecology. Based on the Environmental Protection Agency's (EPA) guidelines, the maximum concentration of TNT in water and soil should not exceed 2 µg/L and 17.2 mg/L, respectively [12]. Others have also shown that the presence of more nitrobenzene derivatives—MNT, DNT, sulfonates, and others- could increase the contamination levels [9, 13]. Considering these notes, TNT-contaminated wastewater should be treated properly before its discharge into waterways.
Up to now, several treatment methods including physical, chemical, and biological techniques have been employed to destroy or adsorb this contaminant presented in wastewater [14]. Although some of them, have advantages but they have also their drawbacks and sometime are not cost-effective or reliable for large-scale use [15, 16]. However, biological treatment processes that use microbial consortia have widely been considered in this context that were shown to capable of mineralizing TNT and producing innocuous byproducts (microbial biomass, H2, and CO2). In in this regard, due to their affordability, great efficacy, and environmental friendliness, biological processes have recently attracted a lot of interest in the treatment of such wastewater. Furthermore, both aerobic and anaerobic conditions were applicable to effective biodegradation of these pollutants [17–19].
Based on the literature, numerous modifications have been made to improve the activated sludge process efficiency. Among these, extended aeration-activated sludge (EAAS) is the most commonly modified version of the activated sludge (AS) system, however, major restrictions such as high retention time (HRT), low active biomass, and low organic loading rate are faced with the employment of this technique in wastewater treatment. On the contrast, substantial aeration capacity of this system and full blending of the container contents contribute to the process reliability [20, 21]. Moreover, low BOD outflow, low residual activated sludge, and low ammonia discharge are among the advantages of the EAAS. Therefore, since the advantages of the EAAS system are often exceeded its disadvantages, thus it seems to be essential to adjust the condition under which the best removal efficiency could be attained [22, 23].
Numerous mathematical modeling techniques have been developed during the past ten years to enhance the treatment procedures. Artificial intelligence (AI) has been widely used in this context for accurate biological process optimization as well as data modeling purposes. Adaptive neuro-fuzzy inference systems (ANFIS), fuzzy logic, support vector machines, genetic algorithms (GA), response surface methodologies (RSM), and artificial neural networks (ANN) are some of the most widely utilized AI techniques [24]. An innovative and successful method for simulating input-output interactions in complex systems is the adaptive neuro-fuzzy inference system (ANFIS).. [25]. With this approach, the problem may first be solved on a fuzzy inference system using an ANN trained on training data (FIS). Finally, a FIS in the ANFIS network precisely identifies the hidden layers. By using this technique, the difficult task of discovering and forecasting the ANN model's hidden layers is removed. While the ANFIS technique does not have a sophisticated mathematical model and is a quick and flexible way to create predictive models of biochemical therapy processes, this might be a compelling argument for employing it [26]. Another study technique for finding precise or approximative answers to optimization issues is the genetic algorithm (GA) [26]. GAs are optimization methods that are guided by the concepts of natural selection and evolution [27]. This method's capacity to provide understandable models for complicated and challenging systems is one of its benefits. Because discontinuous, unpredictable, and nonlinear functions are not suited for typical optimization patterns, this approach can be used to optimize systems instead [26].
A popular machine learning strategy that is a part of AI is called an ANN. Neural networks fall under the category of "black box" models since it is not essential to understand the physical properties of the process. It creates a connection between the factors affecting the input and output. With the passage of time and increased studies, due to the many advantages of ANN including modeling complex nonlinear functions with high accuracy, supporting multiple input and multiple output (MIMO) modeling, working with messy and incomplete data, requiring less processing effort, and The possibility of updating or training the model using new data, it was found that there is a great desire to use these models. But along with its advantages, ANN modeling also has disadvantages. These drawbacks include the fact that the model parameter (number of nodes, hidden layers) has no physical meaning and that there is no accepted method for determining the network design. Additionally, tests and faults can lead to overfitting or underfitting and do not offer a singular solution. Additionally, a poorly trained network may congregate to a local minimum. Therefore, in this study, data were simulated, modeled and optimized using ANFIS and GA approaches. The purpose of this study was to investigate the effectiveness of the extended aeration biological system in eliminating TNT from wastewater, as well as to model the system's optimum performance and identify its kinetic coefficients.