基于SVMs-GA模型的微灌系统管网优化设计

为了优化农业管道输水灌溉中管网的组合方式,提高灌溉效益,节约灌溉成本,在不限定灌溉面积的假设下,以支、毛管的标准段数为决策变量,以系统允许的最大压力差和构成管网的最低管道数目为约束条件,以单位面积的管材成本最低为目标函数建立模型,采用支持向量机耦合遗传算法对模型进行求解,得到满足约束条件和目标函数的不同管径支管的标准段数最优组合。结果表明:当毛管双向铺设时,单位面积的管材成本达到最低,最低值为8 755.7 元/hm²,灌溉面积为24.15 hm²,管网水头差达到系统允许的97.6%~99.7%;与毛管单向铺设相比,每公顷的管材成本降低了6.5%,灌溉面积增加了103%。与已有文献相比,每公顷成本降低了5%,灌溉面积增加了23%。因此支持向量机耦合遗传算法可以较好地优化管网设计,为农业节水灌溉提供了可借鉴的依据。

Optimizing Design of Pipe Network in Micro-irrigation System Using SVMs-GA

The aim of this research is to optimize the combination mode of pipe network in agricultural pipeline irrigation, improve irrigation efficiency and cut irrigation costs. Support vector machine (SVM) coupled with genetic algorithm (GA) was employed to optimize the model, and the optimal combination of the number of standard sections of branch pipes with different diameters that met the constraints and the objective function was obtained. The objective function was to minimize the cost per unit area of pipeline under the assumption of unlimited irrigation area; the decision variables were the number of standard sections of branch and capillary; the constraint conditions were the maximum allowable pressure difference of the system and the minimum number of pipes constituting the pipe network. Results showed that when capillary pipes were laid in both directions, the cost of pipes per unit area was the lowest, 8 755.7 yuan/hm², the irrigation area was 24.15 hm², and the water head difference of the pipe network reached 97.6%-99.7% allowed by the system. Compared with one-way laying, the cost of pipe per hectare of two-way laying was reduced by about 6.5%, and the irrigation area was increased by about 103%. Compared with the results in existing literature, the cost per hectare of pipeline decreased by 5%, and the irrigation area increased by 23%. In conclusion, support vector machine coupled with genetic algorithm can better optimize the design of pipe network, and provides a reference for agricultural water-saving irrigation.