Application of a hybrid mechanistic/machine learning model for prediction of nitrous oxide (N2O) production in a nitrifying sequencing batch reactor

Development of local and global wastewater biochemical oxygen demand real-time prediction models using supervised machine learning algorithms

“…The random forest (RF) algorithm is based on assembling a group of trained decision trees (McMillan et al, 2022). The decision trees were trained on slightly different portions of the dataset to predict the numerical values using the bootstrapping technique (Mehrani et al, 2022) (Torregrossa et al, 2018). The modeling results and discussion are presented in section 4.1.2.…”

“…, where 𝐹 * (𝑥) is the incremental final prediction and 𝐹 𝑡 (𝑥) is the incrementing function for a number of 𝑡 trees (Mehrani et al, 2022). The modeling results and discussion are presented in section 4.1.4.…”

“…Several feature combinations were selected for the modeling stage. The feature selection shown in Table 4 was based on features that greatly influence the BOD5 in practice and their Pearson correlation coefficient (Mehrani et al, 2022;Ching et al, 2022). Machinelearning models were then developed, and the hyperparameters were tuned in an iterated way.…”

“…Mehrani et al (2022) developed a machine-learning model to predict nitrous oxide (N2O) during the nitrification process in sequencing batch reactor WWTP. The authors compared the prediction performance of ANN, gradient boosting algorithm (GB), and SVM.…”

Mainstream short-cut N removal modelling: current status and perspectives

“…For that reason, monitoring the N 2 O emissions is essential in pilot and full-scale applications to improve the applicability of N 2 O models in mainstream short-cut processes. Hybrid models can also be adopted for shortterm laboratory or pilot-scale studies with data scarcity where mechanistic models can be used to simulate the biological process and the data generated can be used in a data-driven model that acts as input to a N 2 O prediction algorithm (Mehrani et al 2022). N 2 O modelling in mainstream short-cut N removal systems shares similar challenges as in conventional nitrification and 2step denitrification systems.…”