Predictive models for estimationof labyrinth weir aeration efficiency

Purpose: The purpose of the study is to estimate the aeration efficiency (E20) of Labyrinth weir using artificial intelligent (AI)-based models. Design/methodology/approach: The aeration efficiency (E20) was collected by using the nine models of Labyrinth weir with different shapes and dimensions. A total of 180 observations were used out of which 126 used to train the AI-based models and the remaining used to test the model. This observation consists of input variables such as Fraud number (Fr), Reynolds number (Re), numbers of keys (N), the ratio of head to the width of the channel (H/W), the ratio of crest length to width of the channel (L/W), the ratio of drop height to width of the channel (D/W) and shape factor (SF) and E20 as the output variables. The AI-based models used were Fuzzy Logic, multi-linear regression (MLR), adaptive neurofuzzy interface system (ANFIS), and artificial neural network (ANN). Findings: The main findings of this investigation are that ANN is the best AI-based model that can estimate the E20 accurately than MLR, ANFIS, and Fuzzy Logic. Sensitivity analysis depicts that drop height at labyrinth weir is the essential factors for the estimation of E20; further, parametric studies have also been performed. Research limitations/implications: The proposed AI-based models can be used in the estimation of E20 with different shapes of labyrinth weir but still it needs improvement for the different dimensions. Practical implications: The best AI-based model can be used to calculate the E20 with the different values of input variables. Originality/value: There are no such AI-based models such as ANN, ANFIS, and Fuzzy Logic, available in the literature which can estimate the values of E20 accurately.