The invention discloses a method for optimizing power utilization modes of residential users with orientation to demand response. The method comprises the following steps: building load response dynamic models comprising an air conditioner load response model, a water heater load response model and a lighting load response model of typical power utilization equipment of the residential users; determining a target function of a power utilization mode optimizing model by taking the minimum power utilization cost and highest power utilization manner satisfaction degree of the users as targets; building constraint conditions of the power utilization mode optimizing model, which comprise adjustable constraint of indoor temperature, the water temperature of the water heater and a lighting surface; solving the model by using a particle swarm algorithm combined with the grey theory to obtain state control variables of the air conditioner, the water heater and the lighting load in various time periods and determine the optimal operation manner of each power utilization equipment. The method is applied to optimization examples of different power utilization scenes of the residents and provides a specific reference scheme for the participation of the residential users in the demand response projects, thereby achieving the effects of meeting the power utilization demand, reducing the power utilization cost and optimizing the load peak.