Economically optimised power dispatch in local systems using evolutionary algorithms and dynamic programming

The importance of local power systems is increasing all over the world. The small system size requires a high flexibility of the management system to follow the fluctuations in generation from renewable sources like wind and sun and in consumption caused by sudden switching of comparable large single loads. The energy management system described in this paper uses predictions of generated and consumed power realised by an artificial neural forecasting system to optimise the power dispatch using hybrid evolutionary algorithms. The performance of the optimisation algorithm is compared to the widely used method of dynamic programming.