ACC24 Gekko Tutorial Session

Special Session: Tackling Control Problems with Open-Source Software in Julia and Python

Open-source control and optimization packages are gaining maturity and industrial acceptance. This accelerates developments from university research groups to industrial practice with a common framework for development and deployment. Recent developments for control include classical analysis to new developments such as transformer-based Model Predictive Control (MPC). This presentation reviews Python open-source packages at the intersection of advanced control, optimization, and machine learning. Gekko is highlighted as an open-source algebraic modeling language built in Python that is used to formulate mixed-integer, nonlinear, and differential equations for optimization and advanced control with object-oriented programming. Gekko has been used to optimize grid energy, maximize production, combine control and design optimization, optimize solar-powered aircraft, control the Temperature Control Lab (TCLab), and maximize waste loading for nuclear waste vitrification. Machine learning models like gaussian process regression, support vector regression, and neural networks are integrated into Gekko for gray-box modeling and optimization. Future developments are critical to sustain community momentum and near-term priorities are discussed.

#acc24 #gekko #machinelearning #python #engineering #automation