基于数值降雨预报信息的梯级水库群发电优化调度

在水库群发电调度中,考虑预报信息的联合调度是提高水库群发电效益和稳定性的有效途径。然而降雨预报信息的不确定性将直接影响发电调度的效益。为此,本文首先将降雨量进行分级,给出各降雨量等级的实际降雨概率分布;然后将蓄水库容与径流相融合,并基于参数–模拟–优化模型(PSO)建立PSO-Hedging Rule Curves(PSO-HRCs)调度图。在此基础上,根据实时预报降雨信息,获得各频率条件下的降雨量、径流量和调度决策,同时评估各频率调度决策的弃水和蓄水风险和损失。本文以浑江梯级水库群为例,采用美国全球预报系统(GFS)发布的降雨信息开展研究,结果表明在实例水库采用70%~85%概率对应的调度决策具有较高的效益和稳定性。

Optimal operation of cascade reservoir hydropower generation based on precipitation forecasts

Efficiency and stability of cascade reservoirs in hydropower generation can be improved by using the data of quantitative precipitation forecasts (QPFs), but the improvements are affected by high uncertainties in QPFs. To enhance the usefulness of inflow forecasts resorting to QPFs, precipitation forecasts were divided into several levels and the probability distributions of the actual precipitation at different levels analyzed in this study. Then, a parameter-simulation-optimization (PSO) model and hedging rule curves (HRCs) were used to construct a PSO-HRCs model for hydropower operation, which fuses the reservoir storage and inflow as a single variable and thus makes the solution computationally tractable. Based on the probability distributions, the PSO-HRCs model can generate precipitation, inflow and operation policy for different probabilities, and hence the corresponding risks and losses in water releasing and storing schemes can be evaluated. This paper presents a case study of the Hun River cascade reservoir system that used the QPFs of ten days lead time, issued by the Global Forecast System (GFS), the U.S. National Centers for Environmental Prediction, to analyze and compare the efficiency and stability of hydropower generation with different probabilities.