According to the technical scheme provided by the invention, a resident user multi-objective optimization power utilization strategy based on data mining comprises the steps of obtaining original power utilization data of resident users; extracting a typical daily power consumption curve of the user from the data, and analyzing power consumption characteristics of the user; establishing a multi-target power utilization model based on the power utilization characteristics; solving the multi-target power utilization model by adopting a genetic algorithm to obtain an optimal power utilization strategy which ensures the power utilization comfort of the user and effectively reduces the power utilization peak-valley difference. According to the technical scheme provided by the invention, the peak-valley difference is reduced while the power utilization comfort of a user is effectively ensured, power generation and power utilization tend to be balanced, and the power supply pressure of a power utilization peak is reduced, so that the problems that the power utilization load is continuously increased, the reliable operation of a power grid is greatly influenced and the peak-valley difference is increased are solved.