Hybrid microgeneration system with neural network predictive control

The paper describes the results from a comparative study of an Artificial Neural Network (ANN) based approach and a conventional on-off control applied to the operation of a ground source heat pump (GSHP)/photovoltaic thermal (PVT) system serving a single house located in Ottawa (Canada). The hybrid renewable microgeneration system was simulated in TRNSYS dynamic simulation environment. Six ANN control logics were analyzed and compared with a conventional on-off strategy both in terms of primary energy consumption and ability to maintain the desired indoor comfort levels during two specific weeks (May 1st-May 7th and July 1st-July 7th). The simulation results showed that, in comparison to the on-off control, the GSHP-PVT system with ANNs based strategies saved primary energy (up to 11% during the week of May 1st-May 7th and 36% during the week of July 1st-July 7th), while maintaining acceptable indoor conditions.