Power system is a classical example of complex systems. In this paper it is shown that the power industry in China is facing a tremendous challenge. Then the complexity in power systems is investigated. First, the cascade failure in power systems is analyzed, and compared with sand-pile model. Then we show that the agent-based modelling is a proper way for power network. Mathematically, the geometric dynamics and differential inclusion are useful tool for the stability analysis of large scale power systems. As for power market, the game theory and generalized control system model are proposed. For a complex power system, an evolutive model may be more accurate in description and analysis. Finally, certain newly developed numerical method in power system computation is introduced. Overall, we are convinced that the theorem of complexity, combined with modern control theory, may be the right way to answer the challenge faced by the power industry in China.

对后验仿真准确度的研究是电力系统仿真可信度研究的重要组成部分，但当仿真与实测有差异时，如何进行仿真验证工作成为一个难题。根据广域测量系统(WAMS)能够提供独立于模型和参数的动态过程同步数据的特点，提出了基于WAMS的电力系统分层分块的动态仿真验证策略。分层策略利用电网由不同电压等级所组成的特点将电网分层，在不同层次上进行仿真验证;分块策略则是利用电网结构按照元件和区域的思想将系统进行解耦来实现仿真验证。通过上述策略，将WAMS量测值作为电网数值仿真的动态等值输入，实现了仿真的解耦，缩小了定位引起仿真误差元件的区间。结合电力系统仿真软件PSS/E给出了基于WAMS的分层分块仿真验证策略的工程实现途径。

Congestion management is one of the key functions of any independent system operator (ISO) in the restructured power industry. Improper implementation of congestion management will sabotage the security and reliability of the power system, and the free trading in the markets. Based on a hybrid market structure, this paper presents a new framework for dynamic congestion management when system stability is seriously infringed due to unexpected contingencies. An integrated dynamic congestion management market is proposed, in which all market participants are encouraged to participate by submitting their offers and bids for their owned control means. By optimally utilising the available resources in this market, a system operator can stabilise the postcontingency system in an economical way. A modified Quasi-Newton Method is employed to solve the formulated dynamic optimisation problem. The IEEE 30-bus test system is studied to illustrate the proposed method.

A load model is one of the most important elements in power system simulation and control. Recently, the constant impedance, constant current, and constant power load in combination with the induction motor load have been widely used as the composite load model, whose parameters are all identified from the field measurements in measurement-based load modeling practices so far. However, there is virtually no research conducted on whether all these parameters could really be identified. This paper investigates the possibility on reducing the number of composite load model parameters to be identified from field measurements. This paper first shows that direct application of the IEEE load motor parameters in the composite load model may be inadequate on describing the load dynamics over different operating status. Then the perturbation method is used to derive the trajectory sensitivities of the equivalent motor parameters, based on which the reduction on the identified parameters of the composite load model is presented. Two cases of measurement-based load modeling in North China and Northeast China are studied to illustrate the validity of the reduction method. It is shown that the reduction does not lose the model’s capability on describing the load dynamics. The reduction on the number of identified parameters not only provides a possible way to solve the multi-valued load model problem based on the current practices on measurement-based load modeling, but it also facilitates building of the load model with more components included in it. Meanwhile, reducing the identified parameters reduces the identification time; thus, the proposed strategy significantly enhances the efficiency of the load modeling work.

电力系统稳定域边界的逼近一直是应用能量函数方法分析电力系统暂态稳定的难点。基于半张量理论，文中给出了逼近电力系统稳定域边界的矩阵方程。通过矩阵运算，可以获得电力系统稳定域边界的高阶逼近表达式。所提出的算法没有对系统进行任何形式的非线性变换，不需要求解系统所有的特征根与特征向量，也无需进行时域积分；算法充分保留了系统的非线性结构，形式简洁。由于该算法完全基于矩阵运算，因此非常适于计算机实现。