Optimization On Parameters Of Rational Formula And Its Application

Storm flood forecasting plays an important role in the water conservancy construction,flood control and drought. Mishui River is a tributary of Xiang Jiang, and locates at the humidregion in the south of China. As this river is a kind of the mountain rivers, and storm floodsalways sudden rise and fall, so there are a lot of difficulties to forecast the storm floodsaccurately. Every year the economic losses caused by flood damage is very serious andsignificantly restricts the economic growth of local areas. With the development of society,the economic losses caused by flood damages will also increase year by year. Therefore, wemust strengthen the researchs on forecasting models of Mishui River, and then ultimatelyreduce the economic loss caused by storm flood.Wulipai station, one of stations of Mishui River, was selected as research object in thispaper. The problems on the computerized solution of flood peak discharge, sensitivity anlysisof parameters, model parameter calibration, parameter optimization, the time of the forecastedflood peak and so on are discussed in details. The major research conclusions are as follows:1. Develop the calculation software of medium and small basin flood peak flow. First,the data of the rainfall runoff are divided into twenty-four groups of floods. Second, thesoftware of flood calculation of medium and small basins is developed base on the rationalformula method of caculationg flood peek using VB developing environment. Practicalexamples prove that it is simple, convenient and reliable to calculate the flood peek with thehelp of developed software.2. First, the sensitivity analysis on the parameters in the rational formula is performed toidentify the importance of model parameters and select the design parameters which will beused in the optimization analysis. Then, the trail-and-error method is used to estimate theparameters of model with the help of developed software. On this basis, the model parametersare further optimized by genetic algorithm.18historical floods from2000to2005of Wulipaistation are applied to calibrate these parameters, and6historical floods in2006are used tovalidate the optimized parameters. Results show that using Genetic Algorithms to optimizemodel parameters not only improves the forecast accuracy but also improves the predictionefficiency. 3. Using the statistical model and rational formula, respectively, to forecast flood peakdischarge. And using multiple linear regression and MATLAB to forecast the time of floodpeak. Results show that: Comparison of two models of peak flow forecast result, it is notdifficult to find that the reasoning formula forecasting results is better than these of statisticalmodel. At last, inspections on the flood peak forecasting and peak time forecasting areperformed. Tests show that the simulate accuracy of the flood peak is100%while thesimulate accuracy of the peak time is91.7%. On the whole, the results of the simulateaccuracy are satisfactory, and meet the states flood forecast accuracy.