Elman网络在养殖水体氨氮预测中的应用研究

Application of Elman Neural Network in Aquaculture Water NH_3-N Prediction

利用2014年6~10月养殖塘口记录的饲料投喂量、水体溶解氧量、水温、气温、浊度、降雨量作为模型输入,检测的氨氮作为模型输出,建立了用于养殖水体氨氮模拟的Elman网络。利用2014年11月的观测数据,对模型的模拟能力进行了检验。结果表明:建立的养殖水体氨氮预测模型,可以较好地模拟水体中氨氮浓度的变化趋势,模拟的绝对误差平均值为0.016 mg/L,决定系数R^2为0.74。说明Elman网络建立的预测模型在养殖水体氨氮含量变化预测中具有强非线性动态描述能力,对养殖水体中氨氮的预测有较好的适用性和预测精度。

Elman artificial neural network model was developed to predict the change of water NH3-N in aquaculture pond. The indexes inclu- ding feed ration, dissolved oxygen in water, water temperature, air temperature, water turbidity, rainfall were recorded and chosen as the input variables, while the NH3-N content in the corresponding pond was chosen as output variable. The above data were collected everyday from June to October in 2014. They were used to develop model in this test, and the data collected in November of 2014 were chosen to evaluate the devel- oped model. The results showed that the changing trend of water NH3 -N in aquaculture pond could be simulated well by the model, the predictive absolute error mean was 0. 016 mg/L, and R2 was 0. 74. The prediction model based on Elman neural network had a strong ability to describe the nonlinear dynamic changes of NH3-N content in aquaculture water, and it showed the good adaptability and accuracy in practical application.