A multi-channel adaptive control algorithm (MACA) that can be put into practice in the closed-loop adaptive optics (AO) system is proposed. MACA, unlike the traditional control algorithm, considers the effect of atmospheric turbulence and the system time delay together with the dynamic characteristics of AO system. The control parameters in each control loop can be modified timely during iterative calculation, which can make the system work at the best control state. Taking a 61-element AO system as an example, based on numerical simulation, MACA is studied to control the voltages of deformable mirror in advance. Analyses are made on convergence, control performance and control bandwidth of this algorithm. Compared with the classical proportional integral (PI) control algorithm, simulation results show that the residual error caused by turbulence in the system is reduced more effectively by using MACA which has good adaptability and improves the closed-loop control bandwidth and control properties of AO system.