Artificial Neural Network for Reliability Evaluation of Power System Network with Renewable Energy

摘要

This paper presents the modelling of power system network with Renewable Energy Sources (RES) using Artificial Neural Network (ANN). The wind speeds and solar irradiances vary with time and cannot be correctly predicted. Therefore efficient multi-state classifications using ANN are done to reduce these errors. The states formed using ANN are modelled using Discrete Markov Chains to determine the inter-state transitions and state-wise probabilities. The errors due to numerous sampling of data are reduced using Monte-Carlo Simulations. The available capacity and probability of the wind turbine and the solar panel are calculated to obtain the reliability indices. The index calculated at the load end is Loss Of Load Expectations (L.O.L.E.) and that at the supply end are Expected Energy Not Supplied (E.E.N.S.) and Expected Demand Not Supplied (E.D.N.S.). Further, to find out the effect of RES on the transmission network, load flow analysis is done. The 1979 IEEE Reliability Test System (R.T.S.) is simulated as the test case. For the modified system consisting of both conventional units and RES, the reliability reduces as compared to the base system. Also, the power flow of the modified system improved as the active line losses are less as compared to the base case. MATLAB and MATPOWER software are used for the simulations.