Results of evolutional random search procedure for load flow optimization in electric networks

A derivation-free random search procedure following the principles of biological evolution was chosen to solve the constrained reactive optimal power flow in order to overcome convergence difficulties originating from the imposed constraints of different physical dimension (scaling problems!) and from the ‘fissured’ solution space caused by them. Starting from an arbitrary system state the solution process is decomposed into two steps, determining first a feasible and then optimal power flow. If the pregiven range of variation of normal control parameters is unsufficient to obtain a feasible state a further facility was included which allows additional load adjustments in some prespecified nodes. The performance of higher developed evolutionary strategies was investigated with some example networks and two real systems. Results and experiencies about the proper choice of strategy parameters which provide best performance are reported.