Parallel Reactive Power Optimization Based on Differential Evolution and PC-cluster

An approach to on-line optimization of reactive power flow is proposed. It is based on a new branch of evolution algorithm named differential evolution (DE) and parallel computing techniques implemented on a PC-cluster. Case study on the IEEE 118-bus system shows that DE, simple as it is, converges fast to the approximate optimal solution. When DE is parallelized and executed on a PC-cluster of moderate size, it is possible to shorten the computing time of reactive power optimization within an acceptable scope, which is needed for on-line application.