新能源的大规模并网造成了电能供给的波动性，迫切要求电网在一定的安全裕度下运行，以应对这种波动性。本文以新能源发电波动性对电网安全运行的影响为切入点，以“静态安全距离”为工具刻画电网安全裕度，提出了新型的电网安全评估、决策模型与方法。为刻画新能源波动对电网潮流的影响，反映电网运行状态的安全裕度，本文提出了基于“距离”的静态安全评估方法，以发电机出力运行点到静态安全域各个边界的距离刻画运行点的安全裕度。静态安全距离能客观地反映电网中各支路对新能源发电波动的承载能力。本文通过辨识安全域的有效边界，减小了静态安全距离模型的求解次数；通过辨识起作用约束，降低了模型的规模；全面提高了静态安全距离的计算效率。为揭示电网在N-1故障、新能源发电等不确定因素综合作用下的安全性，本文提出了基于N-1校核的静态安全距离评估方法。针对N-1故障后运行点在静态安全域内、外的不同情形，建立了相应的静态安全距离模型，以分别刻画N-1故障对电网安全裕度的削减量，以及电网过载时调度员消除过载的难度；基于对N-1关键信息的提取和N-1故障后部分静态安全距离与基态下的近似性，提出了N-1下静态安全距离的快速计算方法。为适应实时运行评估的精度要求，本文建立了交流静态安全距离的模型，以精准、全景地展现电力系统实时运行状态的安全裕度，准确辨识电网的隐患和薄弱环节；利用直流、交流静态安全距离的近似性，设计了高效计算方法，以满足实时运行对安全评估计算效率的要求。为确保发电计划满足期望的安全裕度，本文建立了静态安全距离与发电计划一体优化模型。在传统经济调度模型中，以静态安全距离约束代替传统的安全约束，实现了发电计划决策者对发电计划安全裕度的精准控制，以最经济的方式提升了电网对新能源波动的承载能力。为求解考虑静态安全距离约束经济调度的两层优化问题，本文推导了静态安全距离与发电出力运行点之间的解析关系；通过辨识下层模型中的起作用不等式约束，降低了模型的规模和非线性；随着寻优过程中解的移动，及时更新起作用不等式约束集合，适时修正静态安全距离的解析式；成功地将两层优化的原模型转化为单层优化问题，避免了两层之间低效的迭代，获得极高的计算效率，解决了本文成果走向应用的关键难题。本文工作是在现有安全域理论基础上的发展。期待本文的工作能对提高我国电力系统在新能源大规模并网形势下的安全运行水平有重要的理论与应用价值。
The integration of renewable energy is greatly challenging power system secure operation since it induces electricity supply fluctuations. In order to withstand such fluctuations, the operation point should have certain security margins. Based on the impact mechanism study of renewable energy fluctuations on power grid security and employment of “steady-state security distance” to reflect power grid security margins, this thesis proposes a novel model and method for power grid security assessment, operation and decision-making.In order to describe the impact on power flows of renewable energy fluctuations and reflect the security margin of an operation point, this thesis proposes a “distance”-based steady-state security assessment approach. The models of steady-state security region (SSR) and steady-state security distance (SSD) are established, where the distance from a generation operation point to each SSR boundary is employed to describe the security margin of the operation point. SSD reveals the ability of each branch to accommodate renewable energy fluctuations. In order to enhance the computational efficiency of the SSD model, we identify the active boundaries of the SSR and substitute the original model with a partial constrained model, thus cutting the scale of the optimization model and the number of calculations.This thesis proposes a SSD-based N-1 assessment approach. Two SSD models are established, respectively coping with the situations where the operation point lies inside/outside of a post-contingency SSR. The two SSD respectively depict the security margin curtailment resulted from a contingency, as well as the difficulty for system operators to eliminate power network overload. A novel method is developed to accelerate post-contingency SSD calculation, by abstracting key SSD information under N-1 contingencies, and approximating some post-contingency SSD with pre-contingency SSD.In order to meet the accuracy requirement of real-time operation assessment, this thesis establishes an AC-based SSD model, to fully present the security margin of a real-time operation state, and to precisely uncover real-time security risks as well as vulnerable branches. An efficient algorithm for AC-based SSD calculation is proposed using the similarity of AC and DC power flows, to meet the efficiency requirements of real-time assessment.In order to provide a generation schedule with expected security margins, this thesis establishes an optimization model that incorporates SSD and generation schedules. The model substitutes the traditional security constraints with SSD constraints in the conventional economic dispatch problem. The novel SSD-constrained economic dispatch problem enables the system operators to precisely control the security margin of a generation schedule, and improves the ability of the power network to withstand renewable energy fluctuations in the most economic manner. In order to solve the proposed bi-level problem, we firstly establish the analytical relationship between the SSD and the operation point, secondly reduce the model’s scale and nonlinearity by identifying the active inequality constraints in the model’s lower level, thirdly update the lower level’s active inequality constraints and modify the analytic function of SSD as the operation point moves during the optimization process. The above algorithm successively converts the bi-level problem into a single-level optimization problem, avoids the inefficient iterations between two levels, maximizes the computational efficiency, and makes the SSD-constrained generation scheduling approach accessible to practical applications.The proposed work develops security region theories. Hopefully this thesis is of theoretical and practical value on improving the security operation of power systems with the integration of large-scale renewable energy.
