基于随机动态规划的混合式抽水蓄能电站水库中长期优化调度研究

Mid-long term optimization of reservoir operation for hybrid pumped storage power station based on stochastic dynamic programming

针对水库天然径流的不确定性,在描述径流随机过程的基础上,建立了混合式抽水蓄能水电站水库发电量期望值最大的中长期随机优化调度的数学模型。以发电流量和抽水时间为决策变量,采用双决策变量随机动态规划对模型进行求解。以白山混合式抽水蓄能电站为例进行实例计算,发现抽水时发电量的期望值从不抽水时的2.165×109kW·h增加到2.463×109kW·h,保证出力比不抽水时增加了约12MW,调度周期内各时段的水位也比不抽水时有所提高。将随机模型与确定性长系列法建立的模型进行了比较分析,通过实例对比发现随机模型取得的结果更优且更能反映天然径流的随机性,更符合实际。

Because of the uncertainty of the natural runoff of reservoir, a mid-long term reservoir operation stochastic optimization model is established for the hybrid pumping storage power station based on describing the stochastic process of runoff, which takes the maximum energy generating expectation as its objective. The stochastic dynamic programming algorithm has two control variables： water flow and pumping time. Case study of Baishan mixed pumped storage units＇ shows the energy generating expectation of scheduling period increases from 2 165 million kW＇h to 2 463 million kW＇h, the guaranteed output increases by 12 MW, and the water level is slightly higher in all periods. At the same time, compared with the model established by the long series of law, the stochastic model obtains better results and the model can better reflect the randomness of the natural runoff, even more practical.