温室环境进化RBF网络控制模型的研究

Research on Greenhouse Environmental RBFNN Control Model Based on ES

温室环境模型是制定温室控制策略、优化控制方法的重要基础。采用径向基网络构建北方地区温室环境控制模型,以温室内的温度和相对湿度为输出,采用进化策略优选网络中心,简化网络结构,并与OLS算法进行性能对比。试验结果表明:进化RBF网络具有更小的网络结构,在逼近能力和泛化能力上优于OLS算法。该模型为设计北方温室环境控制方案和结构奠定了良好基础。

Greenhouse environmental models were the important basis of instituting control tactics and optimizing control methods. The RBFNN was used to construct environmental control model of north greenhouse in China. NN outputs were temperature and humidity inside greenhouse. Centers of NN hidden unit were optimized, and NN configuration was simplified by means of ES. RBFNN based on ES and OLS algorithms were compared. Experimental results were showed： approximate and generalization ability of RBFNN based on ES were better to OLS algorithm. The model is beneficial to design control scheme and configuration of the north greenhouse.