Designing a Task Scheduler for Off-Grid Systems

Abstract

Off-grid systems that use solar power can be significantly effected by cloudy days or limited resources. To minimize the use of back-up generators, residents may try to lower their consumption until the sun comes out again, which can be a great inconvenience. A scheduler that uses forecasting methods to predict battery storage levels can be useful in informing a user about the optimal times to complete tasks. The scheduler design uses generation and consumption models from previous work and data from an actual off-grid facility to determine optimal. The scheduler appears to be more efficient than immediate or random task scheduling over the course of a week. There is plenty of room for improvements, as the scheduler makes many assumptions about the states and structures of off-grid systems.