利用Free Search算法推求水位流量关系曲线

Estimating Water Level-discharge Curve by Free Search

建立适宜的水位流量关系是水文资料整编工作的核心内容,但基于最小二乘法与残差平方和最小原则的传统水位流量关系曲线推求方法存在精度不高、难以兼顾绝对误差和相对误差等问题。为此,引入源于动物群体迁移行为的自由搜索(Free Search,FS)算法,以残差平方和、绝对残差绝对值和、相对残差绝对值和作为拟合准则,将FS单个动物每步探查行走时的位置向量作为一组参数值,利用FS算法的动物群体迁移行为直接拟合水位流量关系,从而推求最优的水位流量关系曲线。实例研究结果表明,利用FS率定水位流量关系曲线参数十分简便,寻优一次即可得到满意结果,且与目标规划法和遗传算法相比,拟合精度总体有所提升。

To establish the relationship between water level and discharge is an important job of hydrological data reorganization,but it is a low precision method of parameters calibration in water level-discharge curve using least square method and minimum residual sum of square,and it is difficult to give consideration to the problems such as absolute error and relative error.In this paper,Free Search（FS）,which derived from the animal group migration behavior,was used as a method for estimating water level-discharge curves.The position vector of each exploratory step for the individual animal was ascertained as an initial water level-discharge curve by FS.The simulation results showed that the satisfactory results could be achieved by FS for optimizing the water level-discharge curves,and runoff simulation exhibits relatively high precision water level-discharge compared with the goal programming method and genetic algorithm.