|Title:||Robust power system stabilizer design by evolutionary algorithms|
|Subject:||Hong Kong Polytechnic University -- Dissertations.|
Electric power system stability -- Mathematical models.
|Department:||Department of Electrical Engineering|
|Pages:||xxi, 170 p. : ill. ; 30 cm.|
|Abstract:||With the growth of interconnected power systems and particularly the deregulation of the industry, problems related to low frequency oscillation have been widely reported, including major incidents. As the most cost-effective damping controller, power system stabilizer (PSS) has been widely used to suppress the low frequency oscillation and enhance the system dynamic stability. Among many PSS design methods, probabilistic PSS design approach can address robust PSS design over a wide range of operating conditions. However, there are some limitations on previous studies on probabilistic PSS design and other related issues, for example, some gradient-based nonlinear optimization methods suffer from the problem of high demand on initialization and trapping in local optimum; system contingencies are not systematically considered in previous approaches; the optimal siting, i.e. minimum PSS location, is not considered by and large. To address the above mentioned problems, this thesis is devoted to the extended development of robust and coordinated PSS design in power systems. Evolutionary algorithms (EAs) have attracted a great deal of attention recently and have been found to be robust approaches for solving non-linear, non-differentiable and multi-modal optimization problems, which can overcome the weakness of gradient-based nonlinear optimization methods. Genetic algorithm (GA), particle swarm optimization (PSO) and differential evolution (DE) are three representative EAs that will be employed to solve several PSS design problems in the thesis. Differential evolution (DE) is a novel evolutionary algorithm characterized as simple to implement and little tuning on control parameters. Thus DE is often recommended to receive primary attention when facing new optimization problems. The PSS design model in the thesis differing from most previous models lies in that the former includes a probabilistic eigenvalue-based optimization model. The probabilistic PSS design model will therefore be primarily investigated by DE method. The performance of the proposed DE-based PSS is demonstrated based on two test systems by probabilistic eigenvalue analysis and nonlinear simulation. The results indicate that the probabilistic PSS design by DE is more robust than a gradient-based conventional method. Genetic algorithm (GA) has been one of the most popular EAs during the past decade because it is computationally simple and easy to implement. The BLX-a operator based GA has been previously reported to be able to achieve prominent performance in PSS design. However, the previous PSS design methods did not include a systematic way to handle the system contingencies. In the thesis, the PSS design by the BLX-a GA approach will be further extended to consider system contingencies. The number of contingencies is first significantly reduced by a three-stage critical contingency screening process. The PSS design problem is thus formulated as a multi-objective optimization model with contingencies taken into account. The BLX-a GA will be recursively used to tune PSS parameters so that the prescribed damping criteria subject to contingencies are satisfied under a wide range of operating conditions. An-eight machine system is utilized to demonstrate the effectiveness of the proposed approach and a comparison of the proposed method with a pre-contingency tuning scheme is reported. Particle swarm optimization (PSO) is a swarm intelligence algorithm that mimics the movement of individuals (fishes, birds, or insects) within a group (school, flock, and swarm). PSO is reported to potentially have smaller population-size requirement than other population-based EAs so that PSO might converge faster when they are applied to those highly complex problems that require time-consuming simulations to determine the value of objective function. The probabilistic PSS design with system contingencies taken into account proposed in the thesis is a very complicated problem. Hence, PSO is employed to solve this optimization problem. The effectiveness of the proposed approach is discussed on an eight-machine system and a comparison study of PSO with the DE and the BLX-a GA is primarily conducted. The results also show that PSO consumes less computing time than the DE and the BLX-a GA. To consider the optimal-siting scheme in the probabilistic PSS design, a combination optimization model with mixed discrete and continuous variables is proposed. In this case, GA is also very powerful and flexible in handling this combination optimization problem, which is a little hard for DE or PSO. Hence, a mixed integer-binary coded GA is developed to solve this problem. A partially matched crossover (PMX) operator is introduced to cope with the integer bit conflict. The influence of the probability of crossover and mutation operator on the GA convergence performance is primarily investigated. The effectiveness of the proposed PSS is demonstrated based on two test systems by probabilistic eigenvalue analysis and nonlinear simulation. Case studies show that the proposed optimal-siting probabilistic PSS design method can achieve adequate robust stability, while using a reduced number of PSSs.|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: