首页> 外文期刊>Journal of Infrastructure Systems >Discussion of 'Prediction of Water Pipe Asset Life Using Neural Networks' by D. Achim, F. Ghotb, and K. J. McManus
【24h】

Discussion of 'Prediction of Water Pipe Asset Life Using Neural Networks' by D. Achim, F. Ghotb, and K. J. McManus

机译:D. Achim,F.Ghotb和K.J. McManus讨论的“使用神经网络预测水管资产寿命”

获取原文
获取原文并翻译 | 示例
           

摘要

The authors have addressed an important topic needed for the development and implementation of efficient asset management systems for municipal infrastructures. They developed a model for predicting the number of breaks/kilometer/year of water mains using neural networks (NNs) and compared the accuracy of predicted results against those of two previously developed regression models. The data used in their model development cover a period of 3 years (1997-1999) in a section of the water distribution network in Melbourne, Australia. The authors used pipe diameter, year of construction, age, and length and two geographic coordinates for pipe location as input parameters to their NN model. They found that predictions improved if age and year of construction were included as input. Despite that apparent redundancy, they did not provide an explanation for pipe breakage. It will be helpful if the authors report on the significance or contribution of each input parameter to the output of their model. In view of the relatively low coefficient of determination of their model, they suggested the inclusion of additional input parameters beyond the six stated above.
机译:作者已经解决了开发和实施市政基础设施的有效资产管理系统所需的重要主题。他们开发了一个模型,用于使用神经网络(NN)预测水管的中断数/公里/年,并将预测结果的准确性与两个先前开发的回归模型的准确性进行比较。他们的模型开发中使用的数据涵盖了澳大利亚墨尔本一部分供水网络中为期3年(1997-1999年)的数据。作者使用管道直径,建造年份,年龄和长度以及管道位置的两个地理坐标作为其NN模型的输入参数。他们发现,如果将建设的年龄和年份作为输入,则预测会改善。尽管有明显的冗余,但它们没有提供管道破裂的解释。如果作者报告每个输入参数对其模型输出的重要性或贡献,将很有帮助。考虑到他们模型的确定系数相对较低,他们建议在上述六个参数之外再添加其他输入参数。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号