RIASSUNTO
In order to solve the problems of high cost, low precision and poor measurement effect of ammonia nitrogen concentration in intensive mariculture production, a soft sensor modeling method based on Genetic Algorithm and Stochastic Configuration Network (GA-SCN) is proposed. Firstly, water temperature, dissolved oxygen, pH and conductivity are collected and used as auxiliary variables. Then, GA is used to optimize pre-selection weights W and thresholds B in SCN for the establishment of the network. Finally, the established SCN model is used to predict the ammonia nitrogen concentration in water during intensive mariculture production. The predicted results were compared with BP, GA-BP and SCN model. The experimental results show that the soft-sensing method based on GA-SCN has better prediction accuracy. It is a cross technology of mariculture and artificial intelligence, which provides effective operation guidance for the control and optimization of mariculture production.