I added some new links to a collection of papers published over the last year in our reports section. This research resulted from a collaboration between Sandia, Michigan Tech, and myself at OptimoJoe.
The papers give an overview of one of modeling and control techniques that we use with our projects. In short, we form a reduced order model (ROM) of the system using circuits or bond graphs, discretize the resulting equations using splines, and then solve for an optimal control using Optizelle. What’s interesting about our current discretization is that it allows us to bound either the state or control over the entire domain. This is unusual because a typical method would only be able to bound the control and do so at only a discrete number of points. For these applications, its important to bound both over the entire domain in order to ensure the efficient and safe operation of the system. For example, it may be important to never exceed the storage capacity of an energy storage device or violate a bound on the voltage of a power bus.
In any case, we have a variety of tools not yet released that we employ on our projects. That includes far more sophisticated optimization algorithms than what we’ve so far open sourced. We’re committed to open software and algorithms, but that takes time. Till then, we’re still available to consult for your projects, which gives access to a broader set of tools.