Study On Nonlinear Control Strategy For Differential-Algebraic Model Based Power System With FACTS Considerations

Modern power systems, characteristized by large capacity units, EHV (Extra High Voltage), long distance transmission, heavy load demand, inter-area connection and AC-DC power system parallel, are typical nonlinear, high dimensional, large-scale and dynamic systems. With the interconnections among different power systems, and the increasing numbers of newly introduced devices, a lot of issues on system security and stability raising up while the system size and complexity growing and making the generation and transmission more economic and efficient. Although a lot of efforts and achievements for how to maintain and improve security and stability of power systems have been made, it is still an important issue of most interest to be studied further. This thesis presents intensive and systematic study on transient stability control of structure preserving multi-machine power system, control strategy design of Flexible AC Transmission System (FACTS) and optimal configuration, et al. The main achievements are as follows:Consider that most power system models are Nonlinear Differential-Algebraic Systems (NDAS). This thesis applies Hamilton system theory to the NDAS including the definition, the realization approache and the design of control strategy. Furthermore, the asymptotic stability region is investigated, and it is also improved that the proposed control strategy can enlarge the stability region. The result of a simple case study expressed by system phase portrait and Hamilton function level value diagram illustrates that the proposed theory and methods are correct and effective.This thesis discusses the generalized Hamilton realization of the structure preserving multi-machine power system models applying the NDAS based Hamilton theory as previously mentioned, and presents the corresponding nonlinear excitation control strategy, which could stabilize the disturbed system. The case study of a 2-machine system using the proposed technique validates its correctness and effectiveness. Consider the uncertainty in modeling and the influence of external disturbance on the system, the robust control theory and the passive system theory are applied to solve the L2-gain disturbance attenuation in structure preserving multi-machine power system models. Therefore, a nonlinear, decentralized, disturbance attenuation control strategy is deduced to improve system robustness to dynamic uncertainties and to attenuate bounded exogenous disturbances on the system in the sense of L2-gain. Numerical results of a 4-machine system demonstrate that the proposed nonlinear excitation controller is more efficient in improving system transient stability and robustness against the external disturbance comparing with the conventional PSS. Furthermore, the calculation of Critical Clearing Time (CCT) shows that to some extent the nonlinear excitation controller can enlarge the stability region of the system (CCT is as the benchmark).This thesis presents the design of Unified Power Flow Controller (UPFC) based on the generalized Hamilton theory. The power injection based UPFC model is constructed for the Hamilton theory analysis. As the foundation of investigating the multi-machine power system models, this thesis designs and simulates the UPFC controller in a SMIB system first. Subsequently, the UPFC controller in multi-machine power system models is designed. The case study of a typical 3-machine system using the presented control law validates its correctness and effectiveness.This thesis presents a novel mathematical model for determining the optimal locations and parameters of UPFCs to maximize the system loadability subject to the transmission line capacity limits and specified voltage level. Since it is a nonlinear programming problem, an improved computational intelligence method - Self-Adaptive Evolutionary Programming (SAEP) is employed for the solution of this problem. It is worth noting that the SAEP represents strong global optimization capability, self-adaptability and versatility in solving such problem with nonlinear, discrete and even non-convex characteristics. The computation schemes are discussed in detail, such as the construction of chromosome, handling of equality and inequality constraints and the creation of fitness function, et al. Numerical results demonstrate that the proposed models and methods are correct and effective.