Research On The Optimization And Uncertainty Of P-Ⅲ Type Curve Parameter

Hydrological frequency analysis is based on probability and statistics,as the method of model construction with their own characteristics,its parameter estimation results conclude inevitable uncertainty.This study is on the basis of hydrological frequency linear such as Pearson Ⅲ type curve,which starts from the establishment of the objective function,respectively from the fuzzy weighted algorithm,numerical integral frequency calculation and optimal line standards to study the uncertainty of hydrological frequency parameter estimation.This paper studied it from the following five aspects:(1)This article will expound the basic principle of the social spiders optimization algorithm systematically,and put forward that the social spider optimization algorithm should be applied to hydrological parameters of frequency curve,and at the same time be compared with the traditional parameter estimation method(moment method,the weight function method,probability weighted moment method,genetic algorithm).It is showed that in this method the search efficiency is high,the result is stable and the global optimal solution of parameters can be can find out.(2)For fuzzy weighted curve fitting method problems exist in the process of wiring are nomograph length limited and membership function is not on the premise of large sample.The nomograph by the optimize optimum line method is employed to recount and it is extended to the whole frequency range.The membership function based on central limit theorem is proved that the lower half of normal distribution can increase the weight calculation.In turn,with the method of statistical test,the improved method has been tested,by scoring method and improving the percentage method show that the effect of improved method is good and at the same time been applied to the actual data showing that the method is effective.(3)In the numerical integration of gamma function,the absolute error formula is used to determine the basic step function,deduced and established a relatively perfect step change function,at the same time,with this function the frequency value of the corresponding hydrological sequence can quickly calculated.The trapezoidal algorithm,Simpson algorithm,dragon berg algorithm,adaptive error integral method for numerical integration of gamma function,can be applied to hydrological frequency parameters in optimization calculation with the intelligent optimization algorithms.The ideal data and the measured sequences show that the launch of a new adaptive error algorithm has the advantage of high accuracy of the results,can be well applied to the engineering practice.(4)In the optimum line method when calculating the initial values using the moment method,the weight function method,probability weighted moment method and the methods of linear moment method,the calculation of different initial values of the precision of the method is different,but it is difficult to find a unified criterion that can show its superiority.This research from four different optimal line standards,using Topsis and fuzzy comprehensive evaluation criterion to work out the traditional calculation method of the initial value to judge the uncertainty of initial value calculation method.(5)From uncertainty of the weighted function and numerical integral,with Tao Linkou inflow runoff series as an example,is based on the metropolis Hastings sampling algorithm of Bayesian MCMC estimation to do the uncertainty analysis of hydrological frequency parameter estimation.In this paper,we introduce the social spiders optimization algorithm which has higher searching excellent stability and searching efficiency and the weighted fuzzy optimization suitable line method in membership degree function determination and nomogram B value for the corresponding improvement and statistical test results showed that the improved method is unbiased and effective and it has been greatly improved.The new derived adaptive numerical integration error algorithm can effectively solve the numerical integration error which is difficult to effectively control,integral operation of contradictions.And derived adaptive numerical integration error algorithm is compared to the Trapezoidal,Simpson,Romberg algorithm to show that the calculation speed is fast and fitting accuracy is high.TOPSIS-fuzzy comprehensive evaluation method provides a new evaluation method and evaluation standard for the fitting effect of hydrological frequency parameters.Based on M-H sampling Bayesian parameter uncertainty analysis method the improved fuzzy weighted algorithm,numerical integral algorithm parameters estimation of uncertainty analysis is worked out effectivly.The results show that suitable line method of uncertainty has been better improved by the fuzzy weighted algorithm,and the results of the numerical integration algorithm for the parameter estimation of uncertainty are consistent.