To reduce the dependence of a control system on the communication network, a DC microgrid (DC-MG) usually adopts the distributed control architecture. However, due to the lack of global information, this control mode is highly vulnerable to cyber-attacks, which leads to greater security risks in the DC-MG. There are many types of cyber-attacks, among which the false data injection attacks (FDIA) account for the highest proportion and pose the biggest danger. On this basis, aimed at the instability problems such as interruption of DC-MG operation caused by FDIA, a DC-MG cyber-attack detection and control method based on feedforward neural network (FNN) is proposed. First, the influence mechanism of FDIA causing the instability of the control system is studied, and a mathematical model of instability caused by FDIA is established. Second, an intelligent estimator is constructed by using FNN to monitor the output current from a distributed power converter in the DC-MG, and the error data values of FDIA are calculated according to the output from the estimator. Third, aimed at the calculated values of error data, a reference tracking method based on a PI controller is introduced to reduce the error data in the attacked converter. In addition, the proposed method can eliminate all the high-domain unba-lanced attacks. Finally, the proposed method was simulated and verified by experiments carried out in MATLAB/Simulink and OPALRT, respectively, and results proved its effectiveness.