The problem of automated scheduling a constellation of satellites to achieve maximum information content is a challenging problem. This optimization problem is further complicated when one attempts to meet collection requirements on various sites all while operating within a power budget. The goal of this research is to find the schedules for a set of satellite sensors that observe a set of fixed ground locations while incorporating visibility, solar angles, time of day, site priority, desired goal time between collects for each site, and satellite power. Our solution approach utilizes a top-down approach that accounts for information over the entire scheduling window. The higher layers use relaxed satellite information in ever decreasing time windows, while the lowest resolution scheduling layer uses a Lagrangian relaxation based approach which incorporates the power constraint in the objective function. The final step of our approach is to search locally for better solutions using a k-switch local search method to improve on the optimization objective function. This paper will focus on the technical discussion of the hierarchical approach and generation of the final solution via Lagrangian relaxation. The information content or benefit of the top-down, Lagrangian relaxation technique will also be compared to other scheduling techniques to provide a measure of performance. We will provide performance results for our approach using simulated data and sensors.
展开▼
