The invention relates to a machine learning algorithm based intelligent energy-saving control method and equipment for APs (Access Points) in a WiFi system. The method comprises the steps of: S1: collecting load historical data of each AP in a wireless network system; S2: selecting features in time and space dimensions and carrying out modeling on a load of each AP by utilizing a random forest algorithm to obtain a load model for predicting whether the AP is in an idle state or a non-idle state; S3: based on the load model of each AP, predicting a future state of the AP; and S4: for each AP, extracting time in which the state is continuously idle, and if the time exceeds a preset length, controlling the AP to be closed in the continuously idle time interval. Compared to the prior art, according to the invention, by dynamic control, the idle APs are closed, so that for the large-scale wireless network system, power consumption can be obviously reduced, and energy saving can be implemented.