In supplementary irrigation areas, it is very important to formulate an irrigation plan based on the previous precipitation. However, because of the difficulty of accurately predicting the weather, most studies on irrigation decision making have recommended irrigation schemes under different historical weather years. In this study, a water-saving irrigation system for winter wheat based on the DSSAT model and a genetic algorithm was optimized for different historical years (1970–2017). Accordingly, a decision-making method for determining whether to irrigate in the growth stage of winter wheat was developed using a support vector machine algorithm based on the amount of precipitation in the early stage of winter wheat and the amount of irrigation. The results indicated that the decision-making accuracy in the wintering period, regreening period, and jointing period were 89.4%, 95.7%, and 93.6%, respectively. Thus, the proposed method can effectively determine the irrigation schemes corresponding to different growth stages of winter wheat and facilitate optimal decision making for water-saving irrigation in the winter wheat season in supplementary irrigation areas.

在补充灌溉区，根据先前的降水量制定灌溉计划非常重要。但是，由于难以准确预测天气，大多数有关灌溉决策的研究都建议采用不同历史天气年份的灌溉方案。在本研究中，针对不同的历史年份（1970-2017年）优化了基于DSSAT模型和遗传算法的冬小麦节水灌溉系统。因此，基于冬小麦早期的降水量和灌溉量，使用支持向量机算法开发了一种用于决定是否在冬小麦生育期进行灌溉的决策方法。结果表明，越冬期，复绿期的决策准确度，拔节期分别为89.4％，95.7％和93.6％。因此，所提出的方法可以有效地确定与冬小麦不同生育期相对应的灌溉方案，有利于补充灌区冬小麦节水灌溉的最优决策。