【作者】 张爱国；

【导师】 张建华；

【作者基本信息】 华北电力大学（北京） ， 电力系统及其自动化， 2011， 博士

【摘要】 静止同步串联补偿器(Static Synchronous Series Compensator, SSSC)作为柔性交流输电系统(Flexible AC Transmission Systems, FACTS)的重要装置之一,具有响应速度快、控制精度高、动态性能好等优越性。快速潮流控制是SSSC的主要功能之一,然而选择好的控制方法是充分发挥其潮流控制作用的关键。论文首先在两相同步旋转d-q坐标系下建立了SSSC的数学模型,在分析此模型的基础上提出了基于输入变换的有功和无功解耦策略,给出了能使输电线路上有功和无功实现动态解耦和实现静态解耦的输入变换矩阵。结合有功和无功动态解耦策略,利用神经网络自整定PI算法设计了SSSC的潮流控制器。根据SSSC控制系统的特点,把SSSC的控制分为两部分,即内环控制和外环控制,每个部分都有其明确的被控制对象,并可以分别独立的设计其控制器。在内环控制中,考虑到由于电感的存在,线路电流在电容电压稳定前不会有大的变化,选择SSSC装置本身为被控对象,以SSSC的电容电压为控制目标,选用传统的PID控制器去维持电容电压的稳定。考虑到传统PID控制器的参数很难整定、不具有适应性和响应时间长等方面的缺陷,利用人工神经网络和传统PID控制器相结合的方法设计了神经网络自整定PID控制器,并用此控制器去控制直流侧电容电压。因为利用神经网络的非线性拟合能力在线整定PID控制器的3个参数,使其适应系统的变化,所以解决了传统PID控制器由于参数固定而产生的缺陷。在外环控制中,选择装设SSSC的输电线路为被控对象,以输电线路的有效阻抗为控制目标,选择传统的PID控制器去控制线路的有效阻抗。但是,装设SSSC的输电线路是一个复杂的非线性系统,传统的PID控制器的控制效果不理想。本文采用启发式近似动态规划(Heuristic Dynamic Programming, HDP)算法设计了SSSC外环控制器,控制器总共包含3个神经网络,即模型网络、评价网络和动作网络,动作网络的作用是产生控制量,而评价网络的作用是评价这个控制量的好坏,它们互相协作,从而得到最佳的控制序列。由于将一个复杂的大系统分成两个设计相对容易的子系统,并且每个子系统都有各自的控制对象,所以降低了控制器的复杂性和设计的难度。在MATLAB动态仿真环境中对所建立的神经网络模型、所设计的内环控制器和外环控制器进行了仿真,并与传统的PID控制器进行了对比,仿真结果验证了该控制器的有效性和适用性。更多还原

【Abstract】 Static synchronous series compensator (SSSC) with advantages of faster response, higher control precision and perfect dynamic performance is one of the important FACTS (Flexible AC Transmission Systems) devices. One of the main functions of SSSC is fast power flow control, which will be fully played by the choice of a good control method.A two- machine system mathematic model is built in a synchronously rotating d-q frame by using Park’s transformation, in which the power systems with SSSC is controlled object. In the deeply research of the mathematic model, a d-q axes decoupling method based on the input transformation is proposed, using which the transmission line active and reactive power flow can be decoupled. Meanwhile, the decoupling model which can be used for dynamic decoupling and the transformation matrix which can be used for static decoupling are present. Afterwards the outer-loop controller of SSSC is designed with the neural network self-tuning PI algorithm in the basic of dynamic decoupling.According to the characteristics of SSSC control system, the control of SSSC is able to divide into two parts which have the controlled objects and can be independently designed respectively, namely inner-loop control and outer-loop control. In inner control strategy, because of the existence of inductance, the transmission line current will not be changes greatly before the capacitance voltage stability, of which is taken into account the traditional PID controller is used to maintain the stability of capacitor voltage, in which the SSSC device itself is selected as control object, and taken the capacitor voltage as control objective. However, considering the defects of tuning difficult, inadaptability and long response time of traditional PID controller, a neural network self-tuning PID controller which is emplied to control DC side capacitor voltage is designed using artificial neural networks and traditional PID controller. The three parameters of traditional PID controller can be tunning on line using neural network which has nonlinear fitting ability, so as to adapt the changes of controlled system. Therefore the traditional PID controller’s defects arised due to fixed parameters are able to overcome.In outer control strategy, the traditional PID controller is used to control transmission line effective impedance, in which the transmission line installed with SSSC is selected as control object and taken the effective impedance as control objective. However, the traditional PID controller is not satisfied to morden control since the power system installed with SSSC is a complex nonlinear system. Therefore, the outer controller using Heuristic Dynamic Programming (HDP) is designed for SSSC. The controller based HDP contains a total of three neural networks, in which the model network is to estimate the system’s output, the action network to give the control variable, and the critic network to value this control variable.In this paper, the complicacy and design difficulty of SSSC controller are able to debase because a large complex system is divided into two small systems whose controllers are easily designed correspondingly and control objects are unambiguous.A studying example is carried out to estimate good robustness and adaptability of the proposed controller in the MATLAB dynamic simulation platform. The results verified the availability and feasibility of the proposed control strategy in power flow control of power systems.更多还原