Multilayer feed-forward neural network approach for optimal dispatch of UPFC embedded pool-bilateral electricity market.

SUMMARY The significance and influence of bilateral component on optimal dispatch of mixed pool-bilateral electricity market is the major issue to be examined for trade-off between pool and pre-specified bilateral transactions. This paper proposes single part optimal dispatch model that dispatches pool in coordination with privately negotiated bilateral transactions while minimizing cost of generation, accounting power balance equality constraints and inequality constraints. The critically loaded transmission line is identified, and when Unified Power Flow Controller (UPFC) is placed in it, optimal dispatch solution gets modified because of incorporation of UPFC parameters in the developed formulation. The methodology of multilayer perceptron network is adopted in this work with inputs to the neural network as percentage of bilateral component and status of UPFC for online evaluation of generation cost and incremental cost. The network is trained with back propagation training algorithm, and for IEEE 30 and IEEE 118 bus systems, it is established that Levenberg-Marquardt algorithm outperforms other algorithms in terms of accuracy. Copyright © 2014 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]