Biogas maximization using data-driven modelling with uncertainty analysis and genetic algorithm for municipal wastewater anaerobic digestion

Innovative approach for predicting biogas production from large-scale anaerobic digester using long-short term memory (LSTM) coupled with genetic algorithm (GA)

An artificial neural network (ANN) model called long-short term memory (LSTM), coupled with a genetic algorithm (GA) for feature selection, was used to predict biogas production of large-scale anaerobic digesters (ADs) of Tehran South Wastewater Treatment Plant (Iran), with a biogas production of approximately 30,000 Nm3/d. In order to employ the real conditions, the hydraulic retention time (HRT) of the ADs (21 days) was considered as the LSTM look-back window. To evaluate the model predictions, three different scenarios were defined. In the first scenario, the model predicted the produced biogas by using raw wastewater characteristics and reached the coefficient of determination of R2 = 0.84. The GA selected four out of eleven parameters of raw wastewater, including loads of BOD5, COD, TSS, and TN (kg/d), as the most informative data for the model. In the second scenario, the model predicted the produced biogas by employing the data of the thickened sludge streams entering the ADs and yielded a higher accuracy (R2 = 0.89). In this scenario, GA selected two out of six parameters of the sludge streams, including total flow rate (m3/d) and average solids content (w/w%). Finally, in the third scenario, by putting the parameters of the two previous scenarios together, the model's prediction accuracy increased slightly (R2 = 0.90). The results demonstrated that the GA-LSTM modeling technique could achieve reliable performance in predicting biogas production of large-scale ADs by including HRT in modeling procedure. It was also found that the raw wastewater characteristics severely affect AD behavior and can be successfully used as the input data of the AD models.

Machine learning methods for the modelling and optimisation of biogas production from anaerobic digestion: a review

Biogas plant operators often face huge challenges in the monitoring, controlling and optimisation of the anaerobic digestion (AD) process, as it is very sensitive to surrounding changes, which often leads to process failure and adversely affects biogas production. Conventional implemented methods and mechanistic models are impractical and find it difficult to model the nonlinear and intricate interactions of the AD process. Thus, the development of machine learning (ML) algorithms has attracted considerable interest in the areas of process optimization, real-time monitoring, perturbation detection and parameter prediction. This paper provides a comprehensive and up-to-date overview of different machine learning algorithms, including artificial neural network (ANN), fuzzy logic (FL), adaptive network-based fuzzy inference system (ANFIS), support vector machine (SVM), genetic algorithm (GA) and particle swarm optimization (PSO) in terms of working mechanism, structure, advantages and disadvantages, as well as their prediction performances in modelling the biogas production. A few recent case studies of their applications and limitations are also critically reviewed and compared, providing useful information and recommendation in the selection and application of different ML algorithms. This review shows that the prediction efficiency of different ML algorithms is greatly impacted by variations in the reactor configurations, operating conditions, influent characteristics, selection of input parameters and network architectures. It is recommended to incorporate mixed liquor volatile suspended solids (MLVSS) concentration of the anaerobic digester (ranging from 16,500 to 46,700 mg/L) as one of the input parameters to improve the prediction efficiency of ML modelling. This review also shows that the combination of different ML algorithms (i.e. hybrid GA-ANN model) could yield better accuracy with higher R2 (0.9986) than conventional algorithms and could improve the optimization model of AD. Besides, future works could be focused on the incorporation of an integrated digital twin system coupled with ML techniques into the existing Supervisory Control and Data Acquisition (SCADA) system of any biogas plant to detect any operational abnormalities and prevent digester upsets.

Review of explainable machine learning for anaerobic digestion

Anaerobic digestion (AD) is a promising technology for recovering value-added resources from organic waste, thus achieving sustainable waste management. The performance of AD is dictated by a variety of factors including system design and operating conditions. This necessitates developing suitable modelling and optimization tools to quantify its off-design performance, where the application of machine learning (ML) and soft computing approaches have received increasing attention. Here, we succinctly reviewed the latest progress in black-box ML approaches for AD modelling with a thrust on global and local model interpretability metrics (e.g., Shapley values, partial dependence analysis, permutation feature importance). Categorical applications of the ML and soft computing approaches such as what-if scenario analysis, fault detection in AD systems, long-term operation prediction, and integration of ML with life cycle assessment are discussed. Finally, the research gaps and scopes for future work are summarized.

Multilinear Regression Model for Biogas Production Prediction from Dry Anaerobic Digestion of OFMSW

The aim of this study was to develop a multiple linear regression (MLR) model to predict the specific methane production (SMP) from dry anaerobic digestion (AD) of the organic fraction of municipal solid waste (OFMSW). A data set from an experimental test on a pilot-scale plug-flow reactor (PFR) including 332 observations was used to build the model. Pearson′s correlation matrix and principal component analysis (PCA) examined the relationships between variables. Six parameters, namely total volatile solid (TVSin), organic loading rate (OLR), hydraulic retention time (HRT), C/N ratio, lignin content and total volatile fatty acids (VFAs), had a significant correlation with SMP. Based on these outcomes, a simple and three multiple linear regression models (MLRs) were developed and validated. The simple linear regression model did not properly describe the data (R2 = 0.3). In turn, the MLR including all factors showed the optimal fitting ability (R2 = 0.91). Finally, the MLR including four uncorrelated explanatory variables of feedstock characteristics and operating parameters (e.g., TVSin, OLR, C/N ratio, and lignin content), resulted in the best compromise in terms of number of explanatory variables, model fitting and predictive ability (R2 = 0.87).

Optimal sensor placement method for wastewater treatment plants based on discrete multi-objective state transition algorithm

Parameters monitoring is essential to maintain the stability and efficiency of the wastewater treatment process, which has spurred ubiquitous installation of sensors in wastewater treatment plants (WWTPs). As the rich process data of WWTPs is not effectively transformed into actionable knowledge for system optimization due to improper sensor installation, the sensor placement scheme needs to be optimized. In this paper, a weighted sensor placement optimization model based on sensor cost, information richness and reliability is established to transform the sensor optimization problem to a nonlinear mathematical programming problem. Then a discrete multi-objective state transition algorithm is proposed to find the Pareto optimal solutions. Finally, an evaluation strategy is designed to select the most suitable solution for industrial application. The results of simulation experiments on three different WWTPs demonstrate the validity and superiority of the proposed method, increasing the degree of variable observability and measurement redundancy while keeping the sensor cost at a low level.

Application of machine learning in anaerobic digestion: Perspectives and challenges

Anaerobic digestion (AD) is widely adopted for remediating diverse organic wastes with simultaneous production of renewable energy and nutrient-rich digestate. AD process, however, suffers from instability, thereby adversely affecting biogas production. There have been significant efforts in developing strategies to control the AD process to maintain process stability and predict AD performance. Among these strategies, machine learning (ML) has gained significant interest in recent years in AD process optimization, prediction of uncertain parameters, detection of perturbations, and real-time monitoring. ML uses inductive inference to generalize correlations between input and output data, subsequently used to make informed decisions in new circumstances. This review aims to critically examine ML as applied to the AD process and provides an in-depth assessment of important algorithms (ANN, ANFIS, SVM, RF, GA, and PSO) and their applications in AD modeling. The review also outlines some challenges and perspectives of ML, and highlights future research directions.

Greenhouse gas emission estimation from municipal wastewater using a hybrid approach of generative adversarial network and data-driven modelling

Greenhouse gas (GHG) emissions including carbon dioxide (CO₂), methane (CH₄), and nitrous oxide (N₂O) created via wastewater treatment processes are not easily modeled given the non-linearity and complexity of biological processes. These factors are also impacted by limited data availability making the development of artificial data generation algorithms, such as a generative adversarial network (GAN), useful for determination of GHG emission rate estimates (EREs). The main objective of this study was to develop a hybrid approach of using GAN and regression modelling to determine GHG EREs from a cold-region biological nutrient removal (BNR) municipal wastewater treatment plant (MWTP) in which the aerobic reactor has previously been established as the main GHG emission source. To our knowledge, this is the first application of GAN used for MWTP modelling purposes. The EREs were generated from laboratory-scale reactors used in conjunction with facility-monitored operating parameters to develop the GAN and regression models. Results showed that regression models provided reasonable EREs using parameters including hydraulic retention time (HRT), temperature, total organic carbon, and dissolved oxygen (DO) concentrations for CO₂ EREs; HRT, temperature, DO and phosphate (PO₄^3-) concentrations for CH₄ EREs; and temperature, DO, and nitrogen (nitrite, nitrate, and ammonium) concentrations for N₂O EREs. Additionally, the addition of 100 GAN-created virtual data points improved regression model metrics including increased correlation coefficient and index agreement values, and decreased root mean square error values. Clearly, virtual data augmentation using GAN is a valuable resource in supplementation of limited data for improved modelling outcomes. Genetic algorithm optimization was also used to determine operating parameter modifications resulting in potential for minimization (or maximization) of GHG emissions.

Prediction Model of Hot Metal Silicon Content Based on Improved GA-BPNN

The inconsistency of the detection period of blast furnace data and the large time delay of key parameters make the prediction of the hot metal silicon content face huge challenges. Aiming at the problem that the hot metal silicon content is not consistent with the detection period of time series of multiple control parameters, the cubic spline interpolation fitting model was used to realize the data integration of multiple detection periods. The large time delay of the blast furnace iron making process was analyzed. Moreover, Spearman analysis was combined with the weighted moving average method to optimize the data set of silicon content prediction. Aiming at the problem of low prediction accuracy of the ordinary neural network model, genetic algorithm was used to optimize parameters on the BP neural network model to improve the convergence speed of the model to achieve global optimization. Combined with the autocorrelation analysis of the hot metal silicon content, a modified model for the prediction of hot metal silicon content based on error analysis was proposed to further improve the accuracy of the prediction. The model comprehensively considers problems such as data detection inconsistency, large time delay, and inaccuracy of prediction results. Its average absolute error is 0.05009, which can be used in actual production.