Analysis of ammonia nitrogen concentration using stochastic configuration networks

In the process of intensive aquaculture production, it is essential and important to measure ammonia nitrogen concentration in water on-line. It affects the health of the aquaculture objects and it is the basis of closed-loop control and optimization. At present, it is measured by Ness's reagent method which is offline and time delay. To solve the online problem, aquaculture experiments were conducted and stochastic configuration networks (SCN) method was applied to establish the model. The experimental result demonstrated the effectiveness of the proposed modelling techniques, and it also showed its better accuracy and generalization ability than regress model, BP neural networks, RBF neural networks and random vector functional link (RVFL) model.