Optimization of electricity pricing considering neural network based model of consumers' demand response

Dynamic pricing affects consumers' electricity demands. As the smart grid metering infrastructure enables easy transfer of new tariffs to the consumer (in form of information on consumers' in-home displays or comparable), grid operators can make use of it to improve the current and predicted operational situation of the grid. The problem is to find out an appropriate incentive for the intended modification that obtain consumers' acceptance and cause less costs. To solve this problem, this paper introduces a model of consumers' demand response to changes in prices that - embedded in an optimization loop - can be used to identify optimal dynamic pricing inducing desired response with respect to the defined objective and the given constrains. The model is based on a neural network and outputs the mean value and standard deviation of the changes in load that can be expected as a consequence of the pricing. For optimization, this approach uses heuristic optimization, namely the Mean-Variance Mapping Optimization (MVMO), which provides excellent performance in terms of convergence behavior and accuracy.