A methodology is proposed to transform the power consumption problem for flash memory storage of a hard real-time embedded system into an imprecise computation model. Also, based on imprecise computation models, two scheduling algorithms have been developed to minimize the total power consumptions of the on-line service requests (i.e., tasks) for a flash memory of a hard real-time embedded system. To the best of our knowledge, it is the first idea to solve the power consumption problem of a hard real-time embedded system using imprecise computation models.