【作者】 赵瑞锋；

【导师】 颜伟；

【作者基本信息】 重庆大学 ， 电气工程， 2013， 博士

【摘要】 自动发电控制（Automatic Generation Control，AGC）是实现电网有功频率控制、维持系统频率质量以及联络线交换计划的一种重要手段。常规的AGC控制策略主要针对如何校正区域控制偏差，其研究重点在于区域总调节功率的生成和AGC机组调节功率的分配。然而，随着新能源的快速发展，CPS新标准的推广应用，大区联网新格局的逐步形成，以及电力市场化进程的进一步深化，现代电力系统的AGC已经发展成为一个需要解决快慢机组协调、分级分区协调以及考虑区域控制性能、安全约束和市场因素协调的复杂问题，需要应用新的思路来发展和完善现有AGC的理论和方法。本文在国家自然科学基金项目“输电网中长期状态的精细化模拟与概率评估的基础理论研究”的支持下，基于优化理论和方法，对互联电网下省级与网省两级AGC的动态优化策略、算法及在线计算平台等问题进行深入的研究和探索。主要研究内容和成果如下：提出了一种有效的省级电网AGC的动态优化策略，与现有研究比较，模型中考虑了系统频率变化对AGC控制的影响，改进了现有模型关于AGC机组最小持续爬坡时间的约束以及AGC辅助服务费用的目标。模型以CPS1指标最优和AGC辅助服务费用最少为目标，考虑省级电网的功率平衡约束、CPS指标约束、联络线功率偏差约束，以及AGC机组的爬坡能力及机组最小持续爬坡时间等约束，对未来15分钟AGC机组的基点功率进行最优决策。对重庆电网AGC进行的仿真结果表明，所提省级电网AGC的动态优化策略能满足AGC控制的各种性能要求，尤其在考虑风电功率波动较大的情况下，该动态优化策略在CPS指标、辅助服务费用以及AGC各种控制目标等方面均明显优于常规AGC控制策略。提出了适合求解省级电网AGC动态优化策略的多目标免疫进化规划算法。省级电网AGC的动态优化策略是一个大规模、多时段、含高阶复杂约束混合整数非线性规划问题。通过仿真发现，通过互补松弛技术转换成一般非线性问题，利用内点法求解时存在收敛性问题。为了有效求解该优化模型，利用目标相对占优方法来处理多目标问题，针对进化规划算法中易出现的早熟问题，将免疫系统的浓度调节机制引入进化规划算法，保证了个体的多样性，并指导进化规划局部寻优的方向。算例结果表明，所提多目标免疫进化规划算法能有效求解省级电网AGC的动态优化策略，表现出稳定的收敛性，并具有较快的计算速度。随着我国区域电网特高压的逐步互联，网省两级AGC机组的协调控制困难以及网际联络线功率波动的问题日益突出。针对这一问题，在省级电网AGC的动态优化策略研究的基础上，提出了网省两级AGC的动态优化策略。模型中以网省两级AGC机组的总调节费用最低为优化目标，考虑大区电网和区内各省级电网的功率平衡约束，各省级电网的CPS1和CPS2指标约束，网际特高压联络线和网内各省际联络线的功率偏差约束，以及网省两级AGC机组的动态调节能力约束。针对模型的特点，提出了适合网省两级AGC动态优化策略求解的免疫进化规划算法。算例结果表明，所提网省两级AGC的动态优化策略在满足网省两级电网联络线功率要求和各省CPS考核指标的条件下，有效降低了网省两级AGC机组的调节次数和辅助费用，实现了网省两级电网AGC机组的动态协调优化控制。为实现在不同控制中心对省级电网以及网省两级电网AGC的动态优化策略进行测试与应用，研究并开发了互联电网AGC动态优化策略在线计算平台。研究中提出了一种基于XML电力对象可扩展框架的软件功能自动生成方法，采用了基于进程通信的系统架构设计思想，并改进了现有的“看门狗定时器”容错技术，通过基于语义的异构数据集成和基于连接池的内存数据库管理中间件封装等多项关键技术的综合应用，所研发平台能够有效实现省级及网省两级AGC动态优化策略在线计算和测试，能够满足在线计算对计算时间和容错能力的要求，并具有良好的可移植性、可扩展性和可维护性。更多还原

【Abstract】 Automatic generation control (AGC) is an important research domain in powersystems because it can accomplish the active frequency control of the power grid,maintain the quality of system frequency, and it can maintain the tie line exchangepower within the reasonable range. The traditional AGC control strategies are mainlyfocused on how to correct area control error (ACE), and its investigation centers on theformulation of the total adjustment power and how to distribute. However, with therapid developing of new energy, popularizing of CPS criterion, marketization of powersystem, the AGC in the modern power system is thus developing to a more complicatedproblem. So the new theory and method must be studied to satisfy the new requirementof the new circumstance.Based on optimization theories and methods, this dissertation is to research andexplore on model, algorithms and the computing platform for the dynamic optimizationstrategy of AGC, which is well-supported by the national natural science foundation“fundamental theory research on the elaborated simulation and probabilistic evaluationof transmission grid’s states for the mid-term and long-term”. Main research contentsand achievements are listed below:An effective dynamic optimization strategy of AGC for the province grid isproposed. With respect to the existing studies, the variation of system frequency isconsidered for the AGC, the constraint of minimum duration time of unit adjustmentand the objective of AGC ancillary service cost are improved. CPS1is optimal and theAGC adjustment ancillary service cost is minimum as objective functions, the CPS1andCPS2indexes, tie line power deviation, the upper and lower limits of unit output, andother constraints of the power system are considered as inequality constraints, to obtainthe optimal plan of base point of AGC unit for the next15min. The results prove thatthe proposed dynamic optimization strategy of AGC for the province grid not only meetthe control performance requirement of the AGC, but also under a power grid with alarge fluctuation by wind power, the CPS index, AGC adjustment ancillary service costare obviously better than other AGC control strategies.A multi-objective immune evolutionary algorithm which is more suitable for thedynamic optimization strategy of AGC to solve is proposed. This model is multi-period,large-scale, with high-level complex constraints and mixed-integer nonlinear programming problem. Based on the complementary constraint relaxation technique,this model is converted as a general nonlinear problem, but it has convergence problemif the Interior Point Method is employed. For solving this problem, objective relativedominant is adopted to tackle with the multi-objectives, the density mechanism isnecessarily adopted into the evolutionary programming algorithm to maintain theindividual’s diversity and guiding the optimal exploring direction in the evolution ofgenerations. The solution of study cases verifies that the multi-objective immuneevolutionary algorithm is able to find the optimal solution of the dynamic optimizationstrategy for province grid. While it has the better optimization performance than theconventional evolutionary algorithm’s, and including excellent convergence ability andfast computation speed.As ultra-high voltage line (UHVL) is recently interconnected, the AGC units ofprovince grid (PD) and regional grid (RD) is difficult to coordinate and the problem ofpower deviation fluctuation of UHVL is increasing serious. Aiming these problems, adynamic optimization strategy of RD and PD AGC is proposed. The dynamicoptimization strategy of RD and PD AGC take AGC adjustment ancillary service cost isminimum as objective, consider power balance of RD and PD, the CPS index of eachPD’s control area, tie line power deviation of interconnected PD and the UHVL of RD,and the adjustment capacity of PD and RD AGC units as constraint. According to thecharacteristics of this model, the immune evolutionary programming algorithm is usedto solve this model. The study cases show that the proposed strategy can meet the AGCcontrol performance requirement and CPS index, and effectively decrease theadjustment times of PD and RD AGC unit and the ancillary service cost, and eventuallythe optimal coordination control for PD and RD AGC unit is implemented.For the purpose of test and application of dynamic optimization strategy of AGCfor the province and the regional grid in the different control center, an onlinecomputing platform of the dynamic optimization strategy of AGC for interconnectedpower system is developed. A function generate method is proposed based on powersystem objects extensible framework which is constructed by XML, the design idea ofsystem architecture based on inter-process communication is adopted, the mechanism ofwatchdog timers is improved. By means of the heterogeneous data integration based onsemantics, memory database management middleware based on connection pool, andother technology, the online computing platform of the dynamic optimization strategyof AGC is developed which can satisfy the requirement of computing time and fault-tolerance, meanwhile, the computing platform has a better performance ofportability, expandability, and maintainability.更多还原