Study On Curve Fitting Methodology Of Frequeney Analysis Of Rainstorms

The study of Hydrological frequency curve is one of the important content in hydrology, and the study of curve fitting methodology which is the core content. Through fitting the frequency curve and the measured data points, obtain the best hydrological statistical parameters of the curve, so then to use some data series to study the statistical laws of hydrological phenomena, statistical parameters instead of the overall sample, provide the basis for the design of annual runoff, storms and floods.Based on the research in existence of this area, focus on the issue for the design storm and flood of interest in the design, the author research methodology of frequency curve fitting, and aimed at improving the proper line optimization criteria, so then proposed by storm size setting different weights curve segments optimization criteria. The specific optimization goal setting is:the curve segment which the frequency is less than 10% is calculated the 4th power of the sum of the absolute value of the difference between the measured data and the fitted curve, the curve segment which the frequency is between 10%～25% is calculated the 3th power of the sum of the difference, the curve segment which the frequency is 25%～90% is calculated the quadratic sum of the difference, the curve segment which the frequency is 90%～95% is calculated the 0.5 party of the difference, and computing power of the rest of the fitted curve segment by both the absolute value of the difference, the calculated result of the total minimum fitting error for each curve segment is optimal. Utilizing the principle of local search approach, and combining the principles of simulated annealing algorithm optimization process to achieve these goals. The main contents of this thesis and conclusions obtained are:1, sort out each lasted statistical study of the site area (10min,60min,6h,24/2 and 3d) storm value, and establish a database, renew the data of the site up to 2013, the series increase 15 years longer than the previous series of statistical data;2, On the basis of statistical data, use the method proposed in this paper to calculate The mean value Ex, coefficient of variation Cv and the coefficient of skew Cs values of the various stations (122 stations)and statistical analysis of the distribution of different duration rain storm, the results of the study area shows that the storm mean value of the southern bigger than the northern;3, Through a large number of sites in the study area of statistical data, using the least squares method and the optimization criteria of setting segmentation proposed in this paper, fitting the sites with which has measured data by different duration curve in the study area. By comparison with the two methods, segmentation settings the resulting optimization criterion curve fitting methods in the frequency region of≤10% is better, in line with the appropriate design flood line key consideration dilute and can be used to design storm and flood design.