Hybrid Machine Learning for Predicting Particle Froude Number in Auto-washout Drainage Systems with Sedimented Beds

Abstract

For proper utilization of the channel of the sewage system, an auto-washout method should be used to keep the deposited bed of the channel clean. Particle Froude number (PFr) is important in auto-washout methods. The prediction of PFr values becomes a very complex problem due to its multiple dependencies. In this study, five heterogeneous datasets collected from existing literature, covering a wide range of hydraulic and sediment conditions, were used to develop and validate the models. We used the Multilayer Perceptron Regressor (MLPR) as the base model and the Random Committee (RC-MLPR) as a hybrid machine learning (ML) model to predict the PFr value. Several performance measures were employed to assess the suggested models, including the agreement index, and other commonly used error criteria available in the published literature. The RC-MLPR model outperformed other proposed ML models, state-of-the-art ML models, and existing empirical equations. We also perform sensitive analysis that found the volumetric sediment concentration (Csed) is the most sensitive variable to predict PFr value by the hybrid RC-MLPR model.

Received: 18 November 2025 | Revised: 24 June 2026 | Accepted: 25 August 2026

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available at https://doi.org/10.5281/zenodo.20591098.

Author Contribution Statement

Sanjit Kumar: Methodology, Software, Formal analysis, Data curation, Writing – original draft, Writing – review & editing, Visualization. Mayank Agarwal: Conceptualization, Validation, Investigation, Resources, Writing – original draft, Writing – review & editing, Supervision. Upaka Rathnayake: Conceptualization, Validation, Writing – review & editing, Supervision, Project administration.