With the development of wireless power transportation technology, wireless rechargeable sensor networks (WRSN) are widely used. To prolong the lifetime of WRSNs and make sure the completion of the long-time tasks, mobile charger (MC) is scheduled to charge sensor nodes wirelessly and prolong their lifetime. Existing studies typically assume that the charging power is fixed, which is unreasonable in real scenarios because current chargers can adjust the charging power. In this paper, we take adjustable charging power into consideration and propose the power level aware charging schedule (PACS) problem, which is proved to be NP-hard. To solve the PACS problem, we discretize the network area into several grids. Then we reformulate PACS as a monotone submodular optimization problem and propose an effective algorithm to solve it. Finally, we conducted experiments to evaluate our scheme. The experiment results show that our algorithm achieves better performance than the comparison algorithms by at least 25.42% and 10% on average in terms of charging utility and survival rate.