Abstract: Provided are methods and apparatus for controlling an energy supply system. In an example the energy supply system includes a battery. In an example, provided is a computer-implemented method for controlling a battery. At least a portion of the method can be performed by a computing device including a processor. The method can include (i) creating, using a machine-learning model, a load prediction curve for a load for a current day, (ii) analyzing, to control peak energy use from an electric power source at a specific time, the load prediction curve to identify the specific time and a respective quantity of energy to transfer between the battery and a bus coupled between the electric power source and the load, and (iii) generating a battery controller input based on the identified respective quantity of energy to transfer. Various other methods, systems, and computer-readable media are also disclosed.