摘要
水体富营养化是藻类爆发性生长的主要因素,为了对其进行实时监测预报,提出一种改进遗传神经网络(QGANN),以实现智能预测.该网络从遗传算法(GA)和神经网络(NN)两方面及其相互关系着手,构造了一个基于量子力学原理的量子平衡交叉算子,设计了一种NN混合优化策略,将两者合并共生获得了一类快速、高效的神经网络预测模型.水库和湖泊蓝绿藻爆发预测实验表明:该改进遗传算法(QGA)性能优良;QGANN的泛化能力明显提高,比未经改进的方法(GAsNN)及简单改进的方法(DCGANN)取得了更加满意的效果.
In order to inspect and forecast rich nourishment and growth of blue-green algae, an improved genetic neural network(QGANN) was proposed, in which a quantum balance crossover operator for genetic algorithm (QGA) is developed and a hybrid optimization method based on neural network(NN) is introduced. The experimental results of forecasting algae in lakes and reservoirs indicate that the proposed QGA has a good performance, the QGANN can enhance modeling speed and boost the generalization performance of NN, and has more satisfactorily effect than basic genetic neural network(GAsNN) and improved genetic neural network (DCGANN).
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2008年第2期262-265,共4页
Journal of Shanghai Jiaotong University
基金
国家高技术研究发展计划(863)项目(2003AA601040-02)
上海市科委国际合作处资助项目(052307055)
关键词
水体富营养化
遗传算法
神经网络
蓝绿藻
量子平衡交叉算子
rich nourishment of water
genetic algorithm(GA)
neural network(NN)
blue-green algae
quantum balance crossover operator