Particle Swarm Intelligence Based Reduction Method Applied to Power System Descriptor Models

To model and analyze a large complex dynamic system such as power systems is a very challenging task. Modelling of large real- time systems results in a large number of differential equations that lead to transfer function models that represents a higher order system. Higher order systems impose heavy computational burden, along with additional memory requirements. Therefore, it is necessary to reduce the power system model for simplifying the simulation and controller design. Pragmatic methods are preferred in power system model reduction for good performance, as they are simple to use along with their ability to maintain the physical structure of the model. A good algorithm to model the order reduction of power system applications should preserve the important characteristic and performability of the original system. The dimensions and density of typical accurate power system models give arise to difficulties which have been handled by several techniques. In this paper Particle Swarm Optimization (PSO) algorithm, evolutionary technique is employed to a two power system models. The first model considered is a single input single output (SISO) single machine connected to an infinite bus (SMIB) and the second model considered is a multiple input multiple output (MIMO) single machine connected to an infinite bus (SMIB). The PSO algorithm is based on the minimisation of the integral squared error between the transient responses of original higher order and reduced order model pertaining to a unit step input. The reduced models show the preservation of stability and other characteristic parameters of original system. The reduced order model so obtained shows minimum integral square error comparable with other reduction techniques.