Fine sediment transportation has played a vital part in the ecosystem. However, fine sediments can also be pollutants that may threaten the ecosystems and environments due to human activities. The fine sediment study is still a challenging issue as it involves a lot of uncertainties and limitations. Due to the complex nature of fine sediment, with the existence of modern technology, modelling of artificial intelligence with different optimisation techniques appear as a suitable tool to estimate the settling velocities of fine sediment in water bodies.
Different models, such as general equations, perceptron, General Regression Neural Network (GRNN), Feedforward Backpropagation Neural Network have been developed and evaluated. Data pre-processing was then performed by applying wavelet transform and firefly optimisation algorithm Firefly Algorithm was also integrated with the developed model to further improve the models in terms of accuracy.
An optimum Artificial Neural Network model was identified and proposed as the suitable model to estimate the settling velocities of fine sediments. It is a Firefly-FBNN model with Bubble and Quick Sort function, which has achieved the highest R-Value of 0.94675 and the lowest RMSE of 0.0003166.
This study is innovative in terms of the introduction of artificial intelligence technique in the model development for settling velocity estimation of fine sediment and seek the possibility of performance enhancement by integrating the developed model with the optimisation algorithm. The research findings may be beneficial to the relevant stakeholders, especially while developing the strategy to deal the sedimentation issues.
