基于数据挖掘的梯级水电站群指令调度优化方法

Data Mining Based Optimization Method for Instruction Dispatching of Cascade Hydropower Station Group

水电规模急剧扩大和电网调度精细化要求不断提高给水电调度的时效性和结果可用性带来极大挑战。提出一种基于数据挖掘的梯级水电站群指令调度优化方法,采用聚类分析从电站海量日发电数据中提炼出若干关键特性指标并聚类形成调度决策库;以此为基础,采用大系统分解协调方法对不同流域不同电站进行分层求解,并耦合逐步优化算法组合优选水电站群调度出力曲线及其变化幅值,快速得到合理可行的调度决策。澜沧江中下游梯级水电站群实例研究表明,所述方法能够快速获得水电站群发电出力曲线,且符合实际调度要求,是一种切实高效的实用化方法。

In recent years, the incessant enhancement of expansion of hydropower size and its meticulous management is posing huge challenges to hydropower system dispatching optimization. Hence, based on the data mining technology, an optimization method of instruction dispatching is designed to ensure computational efficiency and the validity of results. In the method, some key indexes are first selected from the massive operation data of the hydropower system, and fuzzy clustering is used to build the decision-making database. Then, the large system decomposition coordination model and progressive optimization algorithm are employed to search for the optimal decision. The proposed method is applied to the cascaded hydropower station group located on the Lancang River. The results show that with the method the feasible output curve for all the plants can be obtained expeditiously. This work is supported by National Basic Research Program of China （973 Program） （No. 2013CB035906） and National Natural Science Foundation of China （No. 51210014, No. 51579029）.