城市供水管网水质检验监测模型研究-市政工程专业论文 word格式.docxVIP

优秀毕业论文精品参考文献资料摘 要经自来水厂处理后的出厂水，已经满足生活饮用水的卫生标准。但经过复杂 的城市输水管网输送到用户时，水质可能发生明显的变化。为了改善管网的水质， 加强对管网水质的运行管理，有必要研究水质在管网中的变化情况。通过建立供 水管网的水质模型，可以对管网水质进行预测，为水质的运行管理提供可靠的理 论基础，对保障饮用水的水质安全具有重要的意义。本文以苏州市某管网末梢水质指标监测中的 9 个监测点为研究对象，结合实 际情况，对监测得的实验数据采用 4D 检验模型、Dixon 检验模型、Q 检验模型、 Grubbs 检验模型这 4 种统计学检验模型进行离群数据数据的检验，并对检验结 果进行分析，做出科学的取舍判断，为水质模型的建立和水质的预测奠定基础。以浑浊度为模型的控制指标，并选用多元统计回归和人工神经网络方法来建 立管网水质统计模型。其中，运用 SPSS 软件建立多元回归模型并进行了求解， 运用 MATLAB 来进行人工神经网络的训练检验。对研究范围内的对 1 号和 9 号 监测点建立了水质模型，然后对这两个监测点管网水的浊度进行了预测。结果表 明神经网络的模拟效果要比多元统计回归理想，模型的准确性较高，能应用于实 际，而且进行离异数据筛选后的数据应用于模型后预测效果更佳。最后，结合研究范围内的实际情况，提出了水质统计模型的建立必须以大量 的数据为基础，这是保证模型准确性和精度的前提；另外还提出了在建立水质模 型之前，有必要进行离群数据的检验。在分析引起预测误差原因的同时，对水质 统计模型的研究发展做出了一定的展望。 关键词:输水管网，水质模型，预测，浑浊度，多元统计回归，人工神经网络MasMaster Dissertation of Suzhou University of Science and TechnologyAbstractIIIIAbstractWater quality has met the drinking water health standards after it left the treatment plant. However, through complex urban drinking water distribution networks,it might deteriorate seariously when reached the users.In order to improve water quality drinking of water distribution networks, to strengthen the running management, it was necessary to study its change in the networks.Establishing water quality model of distribution networks, not only could forecast the water quality, but also provided reliable the oretical basis for the running management of water quality, all of these had vital significance for the water quality security.This paper makes a study of nine indicators of water quality monitoring spots within a network endings in Suzhou, and combining with actual situationtests the ex perimental data with four kinds of the statistical test model, involved 4D,Dixon,Q and Grubbs test model. After analyzing the test results, it could make a scientific trade-off judgment for the prediction of water quality model and water quality,This paper presented the application of two empirical models for simulating and forecasting turbidity within drinking water distribution systems