The Soft Measure Model of Dissolved Oxygen Based on RBF Network in Ponds

Abstract

The paper establishes the prediction model of dissolved oxygen by using nonlinear approximation ability of RBF neural network, which is based on the analysis of infection factors of dissolved oxygen in aquaculture ponds, and introduces adaptive genetic algorithm to optimize the RBF neural network and make it faster convergence, because the conventional RBF neural network model often leads to longer training time and falls into local minimum easily. This paper applies the external environment factors controlled of aquaculture pond as a model input, which includes water temperature (T), water flux (Q), acidity (PH) and the oxygen machine speed (V). Experiment results have shown that the prediction accuracy of the proposed method of dissolved oxygen is higher than the conventional recursive RBF algorithm, prediction accuracy is significantly improved. The method furnishes the foundation for the monitoring system development of the intelligent aquaculture environment and factory aquaculture, and has actual production guidance.