基于GIS与多子群遗传神经网络模型的水位方案优选

Normal water level scheme optimization based on GIS and multi-population genetic neural network model

水利水电工程决策具有多目标、多层次、多影响因素的特点.针对现有正常蓄水位方案形成过程的不足,建立了基于GIS辅助分析、从蓄水位限值确定到获取方案指标、从方案形成到各方案比选的一整套流程方法.为解决正常蓄水位方案优选的非线性问题,避免综合优选模型中依靠主观经验赋权的人为因素影响,提出将多子群遗传神经网络模型应用于正常蓄水位方案优选.该方法有效减少了人为主观因素的影响,提高了指标权重确定的客观程度,为正常蓄水位方案的优选提供了新思路和解决办法.

Water resources and hydropower engineering decisions have the characteristics of multi target, multi-level, multi-affect factors. Due to the shortages of the formation process existing in the normal water level, this thesis establishes a set of processes based on GIS（Geographic Information System）-assisted analysis, which covers from the water storage level limits determined to get program indicators, from the program formation to the scheme corn parison. To solve the nonlinear problem of the normal water level preferred and to avoid the impact of subjective fac- tors artificially empower value, this paper proposes multi-subgroup genetic neural network model is applied to the normal water level which is preferable. This method is effective in reducing the impact of subjective factors, impro- ving the objective determined index weights, providing new ideas and solutions for the normal water level plan selec tion.