Abstract:

The Muskingum method is an important method for the flood prediction. The calibration of parameter and coefficient is a key and difficult point affecting the forecast accuracy. In this paper, we used the minimum error of modelled result matching the measured values as an evolutionary objective on the basis of introducing the basic idea and performance analysis of extension field search genetic algorithm in order to direct search parameter of the Muskingum prognostic equation, so that we could obtain relation equation of upstream and downstream discharge. This paper studied the flood process from Jiahetan to Gaocun in the lower reaches of the Yellow River, and the results showed the mean absolute error of traditional method was 240 m³/s and the mean relative error was 0.13; the mean absolute error of genetic algorithm method was 95 m³/s and the mean relative error was 0.05. It can be seen from the results that the precision of genetic algorithm is significantly higher than that of traditional method. In practice, for the riverways whose flood wave propagating changes a lot, we should simulate the flood propagating according to different flood magnitudes, and then forecast the flood of corresponding magnitudes.

Key words: genetic algorithms, Muskingum method, parameter optimization, flood prediction