基于Fork/Join多核并行框架的梯级水库群优化调度

Optimal operation of cascaded reservoirs based on Fork/Join multi-core parallel framework

为了满足大规模梯级水库群优化调度精细化管理需求,解决决策计算耗时长及求解效率低等困难,提出了基于Fork/Join多核并行框架的梯级水库群优化调度并行求解方法,并以离散微分动态规划方法并行化为例,给出了梯级水库群优化调度方法在Fork/Join框架下的并行化实现方式。红水河大规模梯级水库群长期发电优化调度测试结果表明,并行计算能够充分发挥多核处理器的加速性能,有效缩短计算耗时,提高求解效率;选择合理的Fork/Join框架规模控制阈值是充分发挥并行优势的关键因素。

In order to meet the refined management demand of optimal operation of large-scale cascaded reservoirs and solve the problems of long running time and low computational efficiency,a parallel method based on the Fork/Join multi-core parallel framework is proposed for optimal operation of large-scale cascaded reservoirs. A parallel discrete differential dynamic programming （PDDDP） was designed to describe the parallelization scheme for optimal operation of cascaded reservoirs based on the Fork/Join framework. The long-term power generation optimal operation of large-scale cascaded reservoirs on the Hongshui River was used as a case study. The testing results show that the parallel method can make full use of the acceleration performance of the multi-core processor, significantly reducing the computation time, and improving the computational efficiency. Moreover,the choice of the reasonable scale control threshold of the Fork/Join framework is critical to taking full advantage of parallel computation.