Load Flow Analysis Using Particle Swarm Optimization Trained Neural Network

A new method of load flow analysis of a power system using Particle Swarm Optimization (PSO) trained neural network is proposed in this paper. Load Flow Analysis is the first and foremost step in power system analysis. It is necessary to solve load flow problem for Power System Planning, Contingency analysis, and Stability analysis. The results obtained from the load flow analysis are Voltage magnitudes and phase angles at various buses and line losses. Newton-Raphson (NR) method is mostly used for solving the load flow problem. This conventional method involves the formation of Jacobian matrix, which increases the complexity of the solution algorithm, and also it fails to solve the problem for ill conditioned systems where line reactance to resistance ratio is small. The proposed method using PSO trained neural network reduces the mathematical computations. Trained neural network is used for solving the load flow problem. The generalized back propagation algorithm for training the neural network is replaced by an optimization technique. The weighing factors in the weighing nodes of the neural network constructed are tuned using PSO. With the advantage of global search abilities of PSO, the network constructed results in the better training. The above-proposed method is tested on standard 5 bus system. The results are compatible with the results obtained by conventional NR method. Nomenclature: nnumber of buses ViVoltage magnitude at bus i δi – Voltage angle at bus i in radians Yij – Magnitude of ij th element of bus admittance matrix θij – Angle of ij th element of bus admittance matrix Pgi – Real power generation at bus i Qgi Reactive power generation at bus i PdiReal power demand at bus i Qdi Reactive power demand at bus i