人工神经网络在赤潮预报中的应用仿真研究

赤潮是目前世界沿海国家面临的十分紧迫的海洋环境问题之一，也是一种全球性的海洋灾害。建立有效的赤潮灾害监测和预报系统已迫在眉睫。本文利用人工神经网络中的BP网络，建立赤潮生物密度与环境因子的人工神经网络的预报模型。以各种理化因子：水温．溶解氧．盐度、总氰．可溶性无机磷．浮游植物密度等为参数．试验人工神经网络的预报效果。针对BP网络训练易陷入局部最优的缺点．本文采用了遗传算法改进网络训练方法，保证网络达到全局最优。结果证明、采用人工神经网络进行赤潮预报是行之有效的。

Implement and simulation of artificial neural network in red tide forecast

At present, red tide is one of very seriously environmental problems along the coastal nations of the world; it is also a global hazard of sea. It is an urgent need to establish an effectively detective and predictive system. This article introduces a predictive model of Artificial Neural Network of red tide biology density and environment factors by use of the back propagation (BP) network. Using physical and chemical parameters, such as temperature of water, Do, salinity, nitrogen, dissolubility abio-phosphorus, density of plankton, test the predictive effect of artificial neural network. Aim at the defect of the back propagation (BP) network ' s training which is easy to get into part of optimization, the article adopts genetic algorithms to improve on the training means of network, and it assures the network to all of optimization. The results show that Artificial Neural Network is an effective way to predict red tide.