A1 A. Rathinam

AD SRM Univ., Chennai

A1 S. Padmini

A1 V. Ravikumar

PB iet

T1 Application of supervised learning artificial neural networks [CPNN, BPNN] for solving power flow problem

JN IET Conference Proceedings

SP 156

OP 160

AB Power flow study is performed to determine the power system static states at each bus to find the steady state operating condition of a system. Power flow study is the most frequently carried out study performed by power utilities and it is required to be performed at almost all the stages of power system planning, operation and control. In this paper, two supervised learning networks namely counter propagation neural networks (CPNN) and multilayer feedforward network with back propagation algorithms are proposed to solve power flow problem under different loading/contingency conditions for computing bus voltage magnitudes and angles of the power system. The artiflcail neural networks implement a pattern mapping task. The CPNN is a network which can obtain a mapping from inputs to outputs by competitive learning and supervised learning. Extensive studies have been made by varying the network parameters of both the networks. The hidden neurons are also varied to fix the optimum architecture for the problem to be solved. Due to its fast training, the proposed CPNN will be particularly useful for power system planning studies, as a number of combinations can be tried within a small time frame. The mathematical model of power flow comprises a set of non-linear algebraic equations conventionally solved with the Newton-Raphson method. The effectiveness of the proposed CPNN based approach for solving power flow is demonstrated by computation of bus voltage magnitudes and voltage angles for different loading conditions and single line-outage contingencies in IEEE 14-bus system. Results of the both the ANN closely agrees with that obtained by fast decoupled load flow. The computation time of both the ANN is much smaller than that by fast decoupled load-flow. However, comparing the training time and suitability for online application in power systems, CPNN is best suited due to its fast learning based on Euclidean distance calculations.

K1 power system planning

K1 decoupled load-flow

K1 power system angles

K1 multilayer feedforward network

K1 bus voltage magnitudes

K1 loading-contingency conditions

K1 supervised learning artificial neural networks

K1 power flow problem

K1 nonlinear algebraic equations

K1 power system control

K1 back propagation algorithms

K1 mathematical model

K1 counter propagation neural networks

K1 Euclidean distance calculations

K1 power systems online application

K1 power system static states

K1 Newton-Raphson method

K1 steady state operating condition

DO https://doi.org/10.1049/ic:20070603

UL https://digital-library.theiet.org/;jsessionid=30qhba0dqbehe.x-iet-live-01content/conferences/10.1049/ic_20070603

LA English

SN

YR 2007

OL EN