基于线性回归的水位预测模型研究

发布时间:2018-11-06 07:37
【摘要】:河流的水位是掌握水文情况和推算其它水文要素的宝贵资料。对于水利部门来说精确地观测水位对工程建设是可以减少很多损失,同时准确地观测水位对合理的利用水资源具有重大的意义。迄今为止,在测量河流水位中存在很多的问题,比如用水尺测量河流水位,水尺常年浸泡在河水中很容易被侵蚀,仪器的损坏会对测量结果产生很大的误差,利用水尺测量水位工作量较大,并且人工测量也容易造成误差;通过用精密设备测量河流水位,设备后期需要维护和保养,成本较高等等。鉴于目前测量水位方法存在以上缺陷,同时在影响水位的众多因素中,降雨量和流量的大小又是影响水位升降的重要因素。所以找到一个既能精确测量水位和又能使投入成本降低的方法很重要。因此本文首先基于一元线性回归模型来研究降雨量与水位的之间的关系,以降雨量为自变量,水位为因变量,利用一元线性回归模型通过降雨量预测出水位。其次基于多元线性回归模型来研究降雨量、流量对水位的影响,以降雨量和流量作为自变量,水位作为因变量,利用二元线性回归模型来预测出水位。本工作首先是从专业网站上如黄河网、中央气象台和河南雨量简明查询系统等收集了2011年到2016年的历史降雨量、流量与水位数据。并对数据进行数据挖掘,逐年逐月逐日进行分类整理,因为本文只考虑正常降雨量和流量对水位的影响,所以在整理数据时将特殊状况下的数据排除,比如在汛期发生特大暴雨时,水坝对水位的调节排除;以及在干旱时期,降雨量偏少的情况排除。依据收集和整理的历史数据,对这些数据进行数据挖掘分析,从其中找到潜在有用的信息和知识。将2011年到2016的降雨量和水位数据作为训练集,再将2014年到2016年的降雨量和水位数据作为测试集。利用训练集中的数据计算出线性回归方程中的回归参数,并进行回归检验,对预测的结果进行分析,将预测结果和实际观测值进行对比分析。再利用测试集中的数据检验回归方程的精确性。最终得出用线性回归模型可以准确地通过降雨量预测出水位的结论,而且该方法不仅可以降低测量水位的成本,而且还可以节省人力资源,提高测量水位的精确度。
[Abstract]:The water level of a river is a valuable data for mastering the hydrological situation and calculating other hydrological elements. For the water conservancy department, the accurate observation of water level can reduce a lot of losses to the construction of the project, and it is of great significance for the rational utilization of water resources to accurately observe the water level at the same time. Up to now, there are many problems in measuring river water level, such as measuring river water level with water meter, soaking in river water all year round is very easy to be eroded, the damage of instrument will cause great error to the result of measurement. It is difficult to measure water level by using water gauge, and it is easy to cause errors by manual measurement. Through the use of precision equipment to measure river water level, equipment later maintenance and maintenance, high cost and so on. In view of the above defects in the current method of measuring water level, among the many factors that affect the water level, the magnitude of rainfall and discharge are the important factors that affect the fluctuation of water level. So it is important to find a method that can measure water level accurately and reduce input cost. In this paper, the relationship between rainfall and water level is studied based on the univariate linear regression model. With rainfall as independent variable and water level as dependent variable, the water level is predicted by single linear regression model. Secondly, based on the multivariate linear regression model, the influence of rainfall and discharge on the water level is studied. The rainfall and discharge are taken as independent variables and the water level is taken as dependent variable, and the binary linear regression model is used to predict the water level. This work is to collect the historical rainfall, discharge and water level data from 2011 to 2016 from professional websites such as Yellow River net, Central Meteorological Station and Henan rainfall Concise query system. The data are mined and classified every month and day by year. Because the influence of normal rainfall and discharge on water level is only considered in this paper, the data under special conditions are excluded when sorting out the data. For example, in the flood season when the heavy rain occurred, the dam to adjust the water level exclusion; And in times of drought, less rainfall is excluded. According to the historical data collected and collated, the data are mined and analyzed to find the potentially useful information and knowledge. Rainfall and water level data from 2011 to 2016 are used as training set, and rainfall and water level data from 2014 to 2016 are used as test data. The regression parameters in the linear regression equation are calculated by using the data of the training set, and the regression test is carried out. The predicted results are analyzed, and the predicted results are compared with the observed values. The accuracy of the regression equation is verified by the data of the test set. Finally, it is concluded that the water level can be accurately predicted by rainfall with linear regression model, and this method can not only reduce the cost of measuring water level, but also save human resources and improve the accuracy of water level measurement.
【学位授予单位】:河南师范大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:P332

【参考文献】

相关期刊论文 前10条

1 范苏颖;;浅析计算机数据挖掘技术[J];信息系统工程;2017年01期

2 李欣;王超;赵虎川;;基于时空序列模型的RBF神经网络在河流水位预测中的应用[J];城市勘测;2016年05期

3 王苗苗;李博峰;沈云中;;顾及自变量与因变量误差及相关性的线性回归[J];同济大学学报(自然科学版);2016年03期

4 毕瑞锋;张发玲;;加权最小二乘法线性回归模型参数的理论推导与计算实例[J];计量与测试技术;2016年02期

5 林天水;陈佩树;;一元线性回归中异方差的处理[J];统计与决策;2015年21期

6 汪奇生;杨德宏;杨根新;;总体最小二乘求解线性回归模型的迭代算法[J];工程勘察;2014年09期

7 陈岚峰;杨静瑜;崔崧;潘庆超;李柳;;基于MATLAB的最小二乘曲线拟合仿真研究[J];沈阳师范大学学报(自然科学版);2014年01期

8 张春海;;Excel回归分析拟合河流水位-流量关系[J];吉林水利;2013年09期

9 伍立志;贾孝霞;沈其君;;模拟研究在线性模型自变量相对重要性估计中的应用[J];浙江预防医学;2013年08期

10 王善序;;水库动态汛限水位预报[J];水资源研究;2013年02期

相关硕士学位论文 前3条

1 苏千叶;线性回归模型参数估计及异常点检测方法的改进[D];山东大学;2015年

2 夏蕾;线性模型中参数估计相对效率的研究[D];青岛科技大学;2010年

3 施德路;基于MATLAB的图像处理系统的设计与实现[D];吉林大学;2007年



本文编号:2313585

资料下载
论文发表

本文链接:https://www.wllwen.com/shoufeilunwen/benkebiyelunwen/2313585.html


Copyright(c)文论论文网All Rights Reserved | 网站地图 |

版权申明:资料由用户e6d9d***提供,本站仅收录摘要或目录,作者需要删除请E-mail邮箱bigeng88@qq.com