The invention discloses an energy-saving optimization method for new energy and thermal power bidirectionally supplying power for high energy carrying loads. The method comprises the steps of building a model to carry out ultra-short period power prediction on wind electric power data in a future preset time slot, adopting an existing load prediction mathematical model to carrying out ultra-short period power prediction on high energy carrying load data in the future preset time slot, judging whether wind electric power meets the demands of normal production operation of high energy carrying loads in the future preset time slot or not, adopting wind electricity to supply power for the high energy carrying loads directly if the wind electric power meets the demands of the normal production operation of the high energy carrying loads in the future preset time slot, and building an energy-saving optimization model of supplying power for the high energy carrying loads through the new energy and the thermal power bidirectionally if the wind electric power cannot meet the demands of the normal production operation of the high energy carrying loads in the future preset time slot. The energy-saving optimization method for the new energy and the thermal power bidirectionally supplying power for the high energy carrying loads has the advantages of being high in reliability and energy utilization rate, and good in energy-saving effect.