Distribution networks are commonly used to demonstrate low-voltage problems. A new method to improve voltage quality is using battery energy storage stations (BESSs), which has a four-quadrant regulating capacity. In this paper, an optimal dispatching model of a distributed BESS considering peak load shifting is proposed to improve the voltage distribution in a distribution network. The objective function is to minimize the power exchange cost between the distribution network and the transmission network and the penalty cost of the voltage deviation. In the process, various constraints are considered, including the node power balance, single/two-way power flow, peak load shifting, line capacity, voltage deviation, photovoltaic station operation, main transformer capacity, and power factor of the distribution network. The big M method is used to linearize the nonlinear variables in the objective function and constraints, and the model is transformed into a mixed-integer linear programming problem, which significantly improves the model accuracy. Simulations are performed using the modified IEEE 33-node system. A typical time period is selected to analyze the node voltage variation, and the results show that the maximum voltage deviation can be reduced from 14.06% to 4.54%. The maximum peak-valley difference of the system can be reduced from 8.83 to 4.23 MW, and the voltage qualification rate can be significantly improved. Moreover, the validity of the proposed model is verified through simulations.