文献知网节
  • 记笔记
摘要:This study focused on the capability of ANN model as an alternative tool to forecast kiwifruit production with respect to the various energy inputs in Mazandaran province which is the most important center of kiwifruit production in Iran.For developing the ANN model the quantity of various energy inputs used for kiwifruit production and the fruit yield was considered.The data was partitioned into 70% for training and 30% for testing data sets.Several MLP network architectures such as three,four and five layers have been developed,generalized and their performance was compared using the quality parameters aiming at finding the one that could result in the best overall performance.The quality parameters of coefficient of determination(R2),the mean absolute error(MAE),mean bias error(MBE) and the root mean square error(RMSE) were considered.The results revealed that an ANN model consisting of one input layer with six input variables,single hidden layer with four neurons and one output layer with one output variable,had the highest coefficient of determination(0.98) and the lowest values of RMSE(1050.36 kg ha-1) and MAE(818.21 kg ha-1).Also the MBE for the selected ANN model was found to be-221.82 kg ha-1,indicating a negligible underestimation of kiwifruit yield by the selected ANN model.Finally,the results showed that the ANN predicted yield was tended to follow the corresponding actual ones quite closely.
会议名称:

2010国际农业工程大会

会议时间:

2010-09-18

会议地点:

中国上海

  • 专辑:

    农业科技

  • 专题:

    园艺

  • 分类号:

    S663.4

  • 手机阅读
    即刻使用手机阅读
    第一步

    扫描二维码下载

    "移动知网-全球学术快报"客户端

    第二步

    打开“全球学术快报”

    点击首页左上角的扫描图标

    第三步

    扫描二维码

    手机同步阅读本篇文献

  • CAJ下载
  • PDF下载

下载手机APP用APP扫此码同步阅读该篇文章

下载:3 页码:399-405 页数:7 大小:183k

相关推荐
  • 相似文献
  • 读者推荐
  • 相关基金文献
  • 关联作者
  • 相关视频