基于BP神经网络的管网维修管径预测分析

以MH水司2004-2012年供水管网维修数据作为研究对象，以BP神经网络模型为研究方法，构建了MH水司供水管网维修的预测模型，对供水管网中待维修管道和管件的管径分布作了短期趋势预测。预测结果表明，该模型的预测精度较高，平均偏差最大为0．0054，均方差最大为0．0077；并给出了DN≤50、50〈DN≤100、100〈DN≤150、150〈DN≤200、200〈DN≤300、300〈DN≤500、500〈DN≤800和800〈DN≤1600的管道维修数量在历年和年内管道维修记录统计分析结果中的变化规律。

Prediction and analysis of maintenance of water supply network by BP neural network

Taking maintenance data of water supply network in MH Waterworks from 2004 to 2012 as research object, the prediction model of maintenance of water supply network was established in MH Waterworks using BP neural network as research method. The pipe diameter distribution of non- maintenance pipeline and fittings was predicted in short term. The results indicated that the prediction precision of this model was relatively high, the maximum average deviation was 0. 005 4 and the maximum mean square deviation was 0. 007 7. The variation regularity of maintenance times was obtained in different pipe diameters, such as DN ≤50,50 〈 DN ≤ 100,100 〈 DN ≤ 150,150 〈 DN ≤200,200 〈 DN ≤300,300 〈 DN ≤500,500 〈 DN≤800 and 800 〈 DN ≤1 600 in statistic analysis results of pipeline maintenance records in past years and in the year.