The invention discloses a sliding mode variable structure Buck circuit control method based on an extreme learning machine, and the method comprises the following steps: S10, carrying out the analysis of a Buck circuit topological structure, and building a system mathematical model through a state space averaging method; S20, defining a sliding mode surface of a controller, designing an extreme learning machine adaptive law, and designing the controller according to the system mathematical model in the S10; S30, verifying the existence of the sliding mode and the stability of the system by using the Lyapunov theorem according to the controller in the S20. Voltage and current signal changes of the circuit are monitored at the same time, and duty ratio signals output by a system are corrected in real time; the system can rapidly reach the sliding mode surface from any initial state, loads are switched when the circuit is in a steady state, the system can be stabilized to a steady-state output voltage value in a short time, the robustness of a controller is guaranteed, and system buffeting is weakened.