日光温室内作物干物质积累预测新方法研究  

Research of Forecasting New Method for Dry Matter Accumulation of Crop in Solar Greenhouse

作　　者：张云鹤[1] 乔晓军[1] 王成[1] 

机构地区：[1]国家农业信息化工程技术研究中心,北京100089

出　　处：《林业机械与木工设备》2005年第6期23-26,共4页Forestry Machinery & Woodworking Equipment

基　　金：国家863计划(2001AA247022);北京市工厂化高效农业项目(H020720030530);北京市农业技术试验示范项目(20012014)

摘　　要：介绍了一种有效结合神经网络和作物生长模型预测日光温室内作物干物质积累的方法,该方法既解决了经验模型的解释性和广适性差的缺点,又解决了机理模型难以理解和使用、输入多而输出欠稳定的不足。以有效光照积累和有效积温作为输入参数,以作物的干物质积累作为输出参数建立神经网络,利用大量实测数据对神经网络结构进行训练,建立作物干物质积累与有效光照累积和有效积温之间的关系,从而对干物质积累进行预测并以日光温室内黄瓜为例,采用此方法对其进行干物质积累的预测,与实际测量结果相比较,误差率不超过10%。In this paper, a method combine neural network and crop growth model to forecast dry matter accumulation of crop in solar had been introduced. The method solves not only the disadvantages of experience model in interpretability and eurytopicity but also the shortage of mechanism model in being difficult to understand, excessive input parameters and unstable output. A neural network has been built, effective accumulated light intensity and effective accumulated temperatures as the input parameters, and dry matter accumulation as the output parameter of the neural network. Applying actual measuring data to train the structure of neural network, the relation of dry matter accumulation with effective accumulated light intensity and effective accumulated temperatures has been confirmed. Making the cucumber as the example, apply the method to forecast the dry matter accumulation. After comparing the forecast value with the measured value, we can see that the error rate less than 10%.

关 键 词：干物质积累 预测模型 神经网络 日光温室 预测方法 作物 

分 类 号：S316[农业科学—作物栽培与耕作技术] TP183[农业科学—农艺学]