Application Of Intelligent Optimization Algorithm In Estimation Of Annual Precipitation Frequency Distribution Parameters

Precipitation frequency analysis is of great significance in providing design basis for urban planning and water conservancy project construction.It is necessary to select appropriate hydrologic distribution lines and statistical parameters for frequency analysis,at present,but there is no uniform and applicable distribution lines and parameter estimation method.In addition,under different distribution lines,traditional parameter estimation methods need to derive complicated formulas,so intelligent optimization algorithm is introduced.In general,the intelligent optimization algorithm only needs to know its numerical relationship,which is not restricted by the continuity and smoothness of functions and avoids a lot of parameter calculation,so it is applied to the calculation of hydrological frequency.In this paper,the parameter estimation methods and distribution lines of hydrological frequency calculation are summarized,and the intelligent optimization algorithm is introduced to solve the problems of parameter estimation in traditional methods.The application of intelligent optimization algorithm in hydrologic frequency analysis and calculation is discussed by using Matlab 2018 b platform.Taking annual precipitation data series of 37 stations in Guanzhong plain of Shanxi as the research object,and in this paper,four traditional parameter estimation methods and four intelligent optimization algorithms are applied,namely the method of moment(MOM),maximum likelihood method(MLM),the method of maximum entropy(POME),linear moment method(L-M),self-adaptive differential evolution algorithm based on opposition-based learning(OL-ADE),dragonflies algorithm(DA)and hybrid genetic and particle swarm algorithm(HGAPSO).According to the six kinds of optimization criterion,the optimization of distribution parameters,design value calculation and goodness of fit evaluation were carried out.The study made the following conclusions:(1)According to the principle of 4 kinds of traditional parameter estimation,four kinds of commonly used distribution lines(G2,P-?,GEV and GLD),the parameters of the formula was deduced.The application principles of four intelligent optimization algorithms and the specific steps of solving the distribution parameters of annual precipitation series are introduced in detail.(2)The four traditional parameter estimation methods and four intelligent optimization algorithms have the following rules: for G2 distribution and GEV distribution,except for the intelligent optimization algorithm under PPCC criterion,the parameter values calculated by the eight parameter estimation methods are roughly the same.For the GLD distribution,the parameter values calculated by the three traditional parameter estimation methods are quite different,while the parameter values calculated by the four intelligent optimization algorithms are roughly the same.For P-? distribution,the parameter values calculated by 4 parameter estimation methods are quite different,when the position parameter is positive,the four traditional parameter estimation method to calculate the values of the parameters is roughly same;when negative position parameters,estimated parameter values are different.(3)The optimal parameter estimation method and distribution line are optimized through 6 indexes of goodness of fit evaluation.In parameter estimation,the fitting effect of intelligent optimization algorithm is generally better than that of traditional parameter estimation method.HGAPSO algorithm in intelligent optimization algorithm is the best,L-M method in traditional algorithm is the best,and MLM method is the worst.In the distribution line,the example verifies that the GEV distribution line and GLD distribution line have a good fitting effect in most sites.In a few sites,P-? distribution also get a better optimal line results.(4)The fitting results of hydrological frequency curve of annual precipitation series and the evaluation results of the goodness of fit show that the intelligent optimization algorithm has better applicability in the calculation of hydrological frequency and the results are more accurate.