Utilization of Genetic Algorithms and Constrained Multivariable Function Minimization to Estimate Load Model Parameters from Disturbance Data

• Main Author: Mertz, Christopher George
• Other Authors: Electrical and Computer Engineering
• Format: Others
• Published: Virginia Tech 2013
• Subjects: load modeling; genetic algorithms; induction machines; PSS/E
• Online Access: http://hdl.handle.net/10919/23289

As the requirements to operate the electric power system become more stringent and operating costs must be kept to a minimum, operators and planners must ensure that power system models are accurate and capable of replicating system disturbances. Traditionally, load models were represented as static ZIP models; however, NERC has recently required that planners model the transient dynamics of motor loads to study their effect on the postdisturbance behavior of the power system. Primarily, these studies are to analyze the effects of fault-induced, delayed voltage recovery, which could lead to cascading voltage stability issues. Genetic algorithms and constrained multivariable function minimization are global and local optimization tools used to extract static and dynamic load model parameters from postdisturbance data. The genetic algorithm's fitness function minimizes the difference between measured and calculated real and reactive power by varying the model parameters. The fitness function of the genetic algorithm, a function of voltage and frequency, evaluates an individual\'s difference between measured and simulated real and reactive power. While real measured data was unavailable, simulations in PSS/E were used to create data, and then compared against estimated data to examine the algorithms' ability to estimate parameters. === Master of Science