电力系统稳定约束最优潮流：模型、算法与并行化

发布时间：2019-02-11 10:16

【摘要】：稳定约束最优潮流是电力系统运行与控制决策中的重要研究课题,它能够在最小化系统运行成本的同时,通过调整稳态运行点提升系统受扰后的动态性能,包括系统的暂态稳定性和短期电压稳定性。稳定约束最优潮流在数学上属于动态优化问题,即含有微分代数方程组约束条件的非线性规划问题,在涉及含复杂模型的大规模电力系统、长仿真时间窗口和多预想故障时,其求解过程计算时间长、耗用内存多,计算复杂性是该问题研究的主要理论和技术障碍。本文着重研究了基于数值优化理论和高性能计算技术高效求解稳定约束最优潮流问题的优化算法及其并行化实现,主要研究内容及其学术成果包括： 1)提出了统一考虑电力系统暂态稳定和短期电压稳定约束的稳定约束最优潮流模型,给出了其基于动态优化问题的数学模型。同时针对复杂电力设备元件的动态模型集成问题,基于面向对象设计和自动微分技术,提出了应用于稳态和暂态分析的系统化复杂模型集成方法,进而设计并实现了应用于稳定约束最优潮流的模块化框架,提升了其算法实现的灵活性,拓展了该优化模型的应用前景。 2)针对动态优化问题的两个算法阶段,即微分代数方程组的转化阶段和非线性规划问题的求解阶段,提出了基于直接多重打靶法和简约空间内点法的两阶段数值优化算法。与已有研究成果相比,该优化算法能够充分利用稳定约束最优潮流的问题特点和结构性质,从而显著提高优化算法的收敛性能和计算效率。通过一系列大规模电力系统算例的数值实验,验证了所提出两阶段优化算法的有效性。 3)对于稳定约束最优潮流问题的优化求解过程,在不同的算法层面提出了可组合使用的四种并行分解策略,即预想故障分解策略、矩阵分块分解策略、打靶区间分解策略和轨迹灵敏度参数分解策略。能够充分利用基于多核CPU的计算集群、对称多处理平台和图形处理器(GPU)等多种高性能计算平台的计算资源,实现了问题求解的多层并行化,有效提高算法执行的计算效率,拓展能够求解的计算规模。
[Abstract]:Stable constrained optimal power flow is an important research topic in power system operation and control decision. It can minimize the operating cost of the system and improve the dynamic performance of the system after disturbance by adjusting the steady operation point. It includes transient stability and short-term voltage stability. Stable constrained optimal power flow is a dynamic optimization problem in mathematics, that is, nonlinear programming problem with constraints of differential algebraic equations. When large scale power systems with complex models, long simulation time windows and many preconceived failures are involved. The computational complexity is the main theoretical and technical obstacle in the research of the problem. This paper focuses on the optimization algorithm based on numerical optimization theory and high performance computing technology for solving stable constrained optimal power flow problem and its parallel implementation. The main research contents and academic achievements are as follows: 1) an optimal power flow model with stability constraints considering power system transient stability and short-term voltage stability constraints is proposed and its mathematical model based on dynamic optimization problem is presented. At the same time, aiming at the dynamic model integration problem of complex power equipment components, based on object-oriented design and automatic differential technology, a systematic complex model integration method applied to steady state and transient analysis is proposed. Furthermore, the modularization framework applied to stable constrained optimal power flow is designed and implemented, which improves the flexibility of the algorithm and expands the application prospect of the optimization model. 2) for the two stages of dynamic optimization, namely, the transformation of differential algebraic equations and the solving of nonlinear programming problems, a two-stage numerical optimization algorithm based on direct multiple target shooting method and reduced space interior point method is proposed. Compared with the existing research results, the proposed optimization algorithm can make full use of the characteristics and structural properties of the stable constrained optimal power flow, thus significantly improving the convergence performance and computational efficiency of the optimization algorithm. The effectiveness of the proposed two-stage optimization algorithm is verified by a series of numerical experiments of large-scale power system examples. 3) for the optimization of stable constrained optimal power flow problem, four combinable parallel decomposition strategies are proposed at different algorithm levels, that is, preconceived fault decomposition strategy and matrix partitioning decomposition strategy. Shooting interval decomposition strategy and trajectory sensitivity parameter decomposition strategy. It can make full use of the computing resources of multi-core CPU computing cluster, symmetric multi-processing platform and graphics processor (GPU), and realize multi-layer parallelization of problem solving, which can effectively improve the efficiency of algorithm execution. Expand the computational scale that can be solved.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM744

【参考文献】