Page 112 - 2025年第56卷第5期

.
［ 21］ 姚蒙，何鹏程 . K-means 算法的初始值选取问题的研究［J］ 福建电脑，2023，39（7）：57-61.
［ 22］ GAN T，JIANG Y Z，ZHAO H L，et al. Research on low energy consumption automatic regulation of cascade gates
.
and pump in open-canal based on reinforcement learning［J］ Journal of Hydroinformatics， 2024， 26（7）： 1673-
1691.
.
［ 23］ 邓志鸿，唐世渭，张铭，等 . Ontology 研究综述［J］ 北京大学学报（自然科学版），2002（5）：730-738.
［ 24］ 张文秀，朱庆华 . 领域本体的构建方法研究［J］ 图书与情报，2011（1）：16-19，40．
.
［ 25］ 倪婷 . 多维时间序列周期图聚类分析及其在城市空气污染分类管理中的应用［D］ 上海：上海师范大学，
.
2022.
［ 26］ 杨才杰，王贺龙，温进化，等 . 基于 IA-PSO 的库坝梯级系统水资源优化配置与调度［J］ 水利水电技术（中
.
英文），2023，54（4）：60-68.
［ 27］ 方国华，郭玉雪，闻昕，等 . 改进的多目标量子遗传算法在南水北调东线工程江苏段水资源优化调度中的应

用［J］ 水资源保护，2018，34（2）：34-41.
.
［ 28］ 董陈超 . 基于遗传算法-生成对抗神经网络模型的宁夏自流灌区水资源优化调度研究［J］ 湖北农业科学，
.
2021，60（15）：174-180，187．

Intelligent identification of real-time regulation pattern in open channels based on
spatiotemporal dynamic knowledge graph

1，3
1，3
1，2

GAN Tian ，WANG Chao ，JIANG Yunzhong ，HAN Kun 1

（1. China Institute of Water Resources and Hydropower Research，Beijing 100038，China；

2. Department of Hydraulics，Yangtze River Scientific Research Institute，Wuhan 430010，China；

3. Key Laboratory of River Basin Digital Twinning of Ministry of Water Resources，Beijing 100038，China）

Abstract：Based on the hydraulic characteristics of open channels and the actual demand of engineering projects，
real-time pattern recognition for intelligent regulation can provide decision-making support for engineering manage⁃

ment，thus enhancing the intelligent level of water transfer projects. This study，aiming at the problem of low effi⁃

ciency in regulation pattern recognition due to the numerous related elements，complex relationships，and strong

time-varying nature of real-time open channel regulation，constructs a knowledge graph that describes the spatiotem⁃
poral dynamic changes of the hydraulic characteristics of open channels，using the Wangnou-Ruwugou river section

of the Jiaodong Water Transfer Project as an example. Firstly，HEC-RAS software was utilized for historical scenario

simulation analysis to obtain prior knowledge，such as optimal initial gate opening values and gate opening adjust⁃

ment trends under different regulation patterns that meet their respective regulation objectives. Then，based on the

Density-Based Spatial Clustering of Applications with Noise-Random Forest （DBSCAN-RF） classification algo⁃
rithm，historical scenario classification analysis was conducted to obtain a labeled classification sample set. Finally，

the Neo4j graph database was used to structurally store historical scenario characteristics and simulation and classifi⁃

cation knowledge，forming a spatiotemporal dynamic atlas，ultimately achieving intelligent recognition of real-time

scheduling patterns for open channels.

Keywords：knowledge graph；spatiotemporal dynamics；real time regulation of open channels；regulation pattern
recognition
（责任编辑：韩 昆）

— 658 —