论文部分内容阅读
摘 要：在傳统的水资源配置过程中，水库水量平衡计算时水库弃水量巨大，造成大量可利用的水资源被浪费。系统梳理水资源配置模型，分析得到水库弃水量大的主要原因是水资源配置模型存在“水源分割”现象。为了解决这个问题，对传统的模型进行改进，增强模型全局统筹能力，提出水资源动态配置模型，建立区域水资源的全局理念，切实有效地做到区域水资源的统筹管理、统筹调度、统筹配置，实现水资源配置动态平衡的目标。使用机器学习的思想，通过训练大数据使模型自主学习，通过交叉验证最终求解出模型的最佳参数组合。改进湖北漳河水库灌区的水资源合理配置模型，使24座大中型水库年均弃水量减少1 080.54万m3，研究区年均缺水量减少126.58万m3。水资源动态配置模型统筹考虑全部供水源的供水能力，减少了水库的弃水量和区域的缺水量，提高了水资源的利用效率，对水资源高效合理利用具有重要意义。
关键词：水资源配置;水资源动态配置模型;动态平衡;机器学习;训练大数据;交叉验证
中图分类号：TV214 文献标志码：A
doi：10.3969/j.issn.1000-1379.2021.08.010
引用格式：刘鑫，韩宇平.基于训练大数据的水资源动态配置模型研究[J].人民黄河，2021，43（8）：52-57.
Abstract： In the traditional allocation process of water resources， the amount of abandoned water in the reservoir is huge when calculating the water balance of the reservoir and it causes a large amount of available water resources to be wasted. Through systematic research of water resources allocation model， we found that the main reason for the large amount of abandoned water in the reservoir was the phenomenon of “water source division” in the water resources allocation model. In order to solve this problem， the traditional model was improved to increase the overall planning ability of the model， we proposed a Dynamic Allocation Model of Water Resources that established a global concept of regional water resources and effectively achieved the overall management， overall scheduling and overall allocation of regional water resources. It realized dynamic allocation and achieved a dynamic balance. We used the idea of machine learning. The model could learn autonomously through training big data， we could finally solve the best parameter combination of the model through cross validation. We improved the rational allocation model of water resources in Zhanghe Reservoir irrigation area in Hubei Province， the model reduced the annually average abandoned water volume of 24 large and medium reservoirs by 10.805 4 million cubic meters and reduced the annually average water shortage in the research area by 1.265 8 million cubic meters. The dynamic allocation model of water resources considered the supply water capacity of all water sources as a whole， reduced the amount of abandoned water in the reservoir and the water shortage in the region， and improved the use efficiency of water resources. It was of great significance to the efficient and reasonable use of water resources.
Key words： allocation of water resources; dynamic allocation model of water resources; dynamic balance; machine learning; training big data; cross validation
关键词：水资源配置;水资源动态配置模型;动态平衡;机器学习;训练大数据;交叉验证
中图分类号：TV214 文献标志码：A
doi：10.3969/j.issn.1000-1379.2021.08.010
引用格式：刘鑫，韩宇平.基于训练大数据的水资源动态配置模型研究[J].人民黄河，2021，43（8）：52-57.
Abstract： In the traditional allocation process of water resources， the amount of abandoned water in the reservoir is huge when calculating the water balance of the reservoir and it causes a large amount of available water resources to be wasted. Through systematic research of water resources allocation model， we found that the main reason for the large amount of abandoned water in the reservoir was the phenomenon of “water source division” in the water resources allocation model. In order to solve this problem， the traditional model was improved to increase the overall planning ability of the model， we proposed a Dynamic Allocation Model of Water Resources that established a global concept of regional water resources and effectively achieved the overall management， overall scheduling and overall allocation of regional water resources. It realized dynamic allocation and achieved a dynamic balance. We used the idea of machine learning. The model could learn autonomously through training big data， we could finally solve the best parameter combination of the model through cross validation. We improved the rational allocation model of water resources in Zhanghe Reservoir irrigation area in Hubei Province， the model reduced the annually average abandoned water volume of 24 large and medium reservoirs by 10.805 4 million cubic meters and reduced the annually average water shortage in the research area by 1.265 8 million cubic meters. The dynamic allocation model of water resources considered the supply water capacity of all water sources as a whole， reduced the amount of abandoned water in the reservoir and the water shortage in the region， and improved the use efficiency of water resources. It was of great significance to the efficient and reasonable use of water resources.
Key words： allocation of water resources; dynamic allocation model of water resources; dynamic balance; machine learning; training big data; cross validation