在能源互联网开放互联使命的驱动下，热电耦合系统将原本相互独立的电力系统与供热系统联合运行，打破了两个系统间的物理和信息壁垒，从而提升了能效、降低了温室气体排放。由于电力系统动态时间短、响应迅速而供热系统动态时间长、热惯性大，热电耦合系统具有多时间尺度特性。然而，现有潮流分析与运行优化研究对多时间尺度特性考虑不充分：考虑多时间尺度的研究简化了系统的拓扑与控制方式；而充分考虑系统拓扑与控制方式的研究忽略了热电多时间尺度特性。此外，已有潮流和优化算法上存在一定的收敛性、可解释性的问题。这些问题限制了热电耦合系统的安全与效率的提升。本论文研究旨在解决热电耦合系统潮流分析与运行优化中多时间尺度、拓扑结构和控制方式带来的诸多问题，其主要贡献为：（1）在潮流分析方面，提出考虑热电不同时间尺度、环状热网结构、变流量调节的热电耦合系统的通用化潮流模型，并提出了一种基于前推回代的计算求解方法。与实测数据相比准确度高；与已有文献相比能够快速收敛，避免了计算结果的发散；与成熟软件相比，可克服软件无法准确计算热动态过程的缺陷。（2）在考虑热网变流量调节的同步优化调度方面，提出非凸优化模型，降低已有模型中整数变量带来的复杂度。改进广义Benders分解方法，在提升收敛性的同时实现加速计算。与已有方法对比，所提模型和方法可以降低运行成本，并解决IPOPT的发散问题。（3）在异步优化调度方面，针对电力系统和供热系统调度周期不同的挑战，提出混合时间尺度调度模型，解决了已有调度方法安全性差、结果不可执行的问题，并研究了调度周期对运行成本和计算效率的影响。（4）在频率控制方面，提出了全分布式最优频率控制方法，克服了已有研究未充分考虑热电耦合特性而导致机组出力越限的问题，同时该方法在阻尼系数不准确时仍具有鲁棒性。综上所述，本文研究面向热电耦合系统潮流分析和运行优化中来自多时间尺度的挑战，通过模型和算法的创新，解决了已有模型对拓扑结构和控制机制的过度简化，克服了已有算法在收敛性、准确性上存在的问题。研究成果可应用于能源互联网中广泛存在的热电耦合综合能源系统，为实现精准高效的潮流计算、经济调度与频率控制服务。
Driven by the objective of constructing an open and interconnected Energy Internet, combined heat and power systems break the physical and cyber barriers between the electric power system and the heating system and thus improves the system’s efficiency and reduces greenhouse gas emission. However, current studies on power flow analysis and optimization do not fully address the different time scale characteristics of the electric power system and the heating system: The research considering the different time scales simplifies system topology and control mechanisms; The research without compromise on system topology and control mechanisms fails to consider the different time scales of electricity and heat. Additionally, existing solution methods for power flow and optimization have convergence problems and sometimes are not interpretable. These problems threaten the efficiency and security of combined heat and power systems.In this dissertation, the challenges from different time scales are addressed in the power flow, economic dispatch, and frequency control for the combined heat and power system, where the topology and control mechanisms are not simplified. The main contributions are:1) In power flow analysis, the power flow model of combined heat and power systems is proposed considering the heat dynamic process, the meshed network, and variable mass flow. A decomposition solution method based on backward-forward iteration is proposed to solve the nonlinear power flow model iteratively and sequentially. The results of the proposed method have high accuracy compared with real-time measurement. Also, the proposed method has fast convergence speed and avoids the divergence problems of existing methods. Moreover, the proposed method overperforms commercial software in terms of the heat dynamic process. 2) In the synchronous economic dispatch with variable mass flow, the proposed optimization model reduces the complexity from integers in the existing optimization model without compromising on accuracy. The resulting non-convex model is solved by the proposed modified Generalized Benders Decomposition method with improved convergence and acceleration. Compared with existing methods, the proposed method has lower overall costs and overcomes the divergence problem of solver IPOPT. 3) In the asynchronous economic dispatch, the different adjustment time scales of the electric power system and the heating system are addressed by the asynchronous dispatch models. The comparison with the traditional synchronous methods demonstrates that the asynchronous method can overcome the security problems and infeasible results. Also, the influence of the dispatch interval on the overall costs and computational efficiency is studied. 4) In the frequency control, a fully-distributed frequency control method is developed with the system-wide optimality. The proposed control method does not violate the constraints of generator’s feasible regions compared with existing control methods and is robust to the inaccurate damping coefficient.In summary, this dissertation studies the challenges from different time scales in power flow analysis and optimization of combined heat and power systems. By the innovation of models and solution methods, the simplifications on topology and control mechanisms are overcome, while the convergence and accuracy are improved compared with existing methods. The research in this dissertation can be applied for the accurate and efficient power flow calculation, economic dispatch, and frequency control of the combined heat and power systems in the Energy Internet.
