Existing research on wireless rechargeable sensor networks has focused on scheduling mobile chargers to recharge fixed or mobile nodes, while a few studies have begun to investigate the problem of charging nodes with uncertain mobility by static chargers. The main challenge in charging nodes with non-deterministic mobility is to determine the locations where the static chargers are deployed. Thus, once mobile nodes move within the effective charging range of the charger, they can be charged. The current studies have deconstructed this problem in two steps: first, generate a candidate set of chargers' locations, and then select some of them as the final deployment solution. The latter is called the charger selection problem, which is a typical multi-objective optimization problem with two goals: (1) prolonging the network lifetime, and (2) reducing the cost of network deployment. In our work, we propose a genetic-algorithm-based multi-objective optimization scheme (GAMOS) to address the charger selection problem. First, construct a D-dimensional vector, where $D$ is the cardinality of the candidate set. Each dimension of the vector represents a candidate charger and takes the value of 0 or 1, while 1 means that the candidate location is selected as the deployment solution and 0 means not. We initialize several vectors randomly in the D-dimensional solution space, and then iteratively optimize them. Finally, we use a dual indicator selection method for final solution selection. We pioneered the use of multi-objective optimization algorithms to solve the energy supply problem in WRSN, what's more, the results of simulation experiments based on the NCSU dataset[1] show that the average network survival rate of GAMOS is 90.60% and the average number of chargers is 7.20, outperforming other existing algorithms overall.