【摘要】 营养盐是控制长江口流域富营养化的关键因子之一。分析亚硝酸盐与其影响因子之间关系,引入人工智能方法,基于弹性BP神经网络,建立亚硝酸盐非线性预测模型,目的是通过影响因子在线监测,间接实现亚硝酸盐在线监测。依据神经网络权值和阈值获取方法不同,形成基于弹性BP神经网络、基于遗传算法和弹性BP神经网络,和基于改进的遗传算法和弹性BP神经网络的亚硝酸盐预测模型3种。通过仿真实验,分析3种模型对亚硝酸盐预测的影响,发现基于改进的自适应遗传算法和弹性BP神经网络的亚硝酸盐模型预测效果最优,为选择合适模型提供依据。更多还原

【Abstract】 Nutritive salt is one of key factors that control the Yangtze River estuary eutrophication. This paper analyzes the relationships between nitrite and its influence factors, and introduces the method of artificial intelligence to establish nonlinear nitrite predictive models based on elastic BP neural networks, which provides a way to indirectly achieve online nitrite detection through monitoring the influence factors. According to different methods of acquiring the weights and thresholds of the elastic BP neural network, this paper develops three nitrite predictive models, including the model based on an elastic BP neural network, on an adaptive genetic algorithm(AGA) and an elastic BP neural network, and on an improved AGA and an elastic BP neural network. Through simulation experiments, this paper analyzes the effects of the three forecasting models on nitrite prediction. It is found that the model based on an improved the adaptive genetic algorithm and an elastic BP neural network has the best effect to predict nutrient values, which provides a basis for selecting appropriate models.更多还原