摘要
针对年内不同时期径流成因与变化规律的差异,本文建立了月径流的分期组合预报模型,对不同时期径流成因特点选用相应的组合预报模型。针对过渡期与丰水期径流预报不确定性较大的特点,采用贝叶斯平均模型进行组合预报,能够发布概率预报。为量化预报误差对水电站发电调度造成的影响,建立了水电站预报优化调度的模糊风险分析方法,从而为科学决策提供参考依据。实例应用表明,分期组合预报模型优于单一模型,且发布的概率预报结合模糊风险分析对制定水电站的月调度计划具有一定的参考价值。
A staged ensemble model is developed to predict monthly variation of runoff, adopting a different ensemble model for each period of different runoff formation and characteristics. Due to the large uncertainly of runoff prediction during the transitional period and wet period, the Bayesian model averaging is used for combination forecasting, with which a probabilistie predietion can be released. A new method, called fuzzy risk analysis, is established to quantify the power generation risk resulted from forecasting error and to assist decision making. The application shows that this model is superior to the individual member model, and a fuzzy risk analysis of power generation ean be references for making monthly power generation plan.
出处
《水力发电学报》
EI
CSCD
北大核心
2009年第6期140-145,213,共7页
Journal of Hydroelectric Engineering
基金
国家自然科学基金与二滩水电开发有限公司雅砻江水电联合研究基金项目(50579095)
关键词
水电工程
组合预报
贝叶斯平均模型
神经网络
模糊风险
hydroelectric engineering
ensemble prediction
Bayesian model averaging
artificial neural network
fuzzy risk
