(Sorry, that 1.0 announcement I just sent was an accident, a copy of what I sent to the developer list back in May.)
Hi MATPOWER users, I wanted to let you know about some of my recent work on the internals of MATPOWER. I’ve been working on a major overhaul and generalization of the network modeling in MATPOWER that will allow for much easier customization and implementation of new device models, etc, as well as, unbalanced 3-phase modeling. More on that to come in the future. But in the process, I found myself refactoring and enhancing some of the code used for building and solving the mathematical model for the OPF and I decided that it made sense to pull out this code, document it more thoroughly and release it as a separate package called MP-Opt-Model<https://github.com/MATPOWER/mp-opt-model> to be included in future MATPOWER versions, but in its own subdirectory, like MP-Test, MIPS and MOST. So, what is MP-Opt-Model? It consists of the opt_model class for building the model for an arbitrary optimization problem plus all of the code related to interfacing to 3rd party solvers and a few utility functions. To build and solve a model using MP-Opt-Model, you simply create an opt_model object … om = opt_model; Add some optimization variables ... om.add_var('theta', ...); om.add_var('V', ...); om.add_var('Pg', ...); om.add_var('Qg', ...); ... some constraints ... om.add_nln_constraint('Pmismatch', ...); om.add_nln_constraint('Qmismatch', ...); om.add_lin_constraint('branch_angle_diff', ...); ... and some costs ... om.add_quad_cost('genCost', ...); om.add_nln_cost('customCost', ...); ... then call the solve method. [x, f, success] = om.solve(opt); The solver can be selected and its options controlled by the opt struct. MP-Opt-Model currently handles linear (LP, MILP), quadratic (QP, MIQP) and general nonlinear (NLP) programming problems, as well as linear and nonlinear equations, depending on the solvers you have installed. It is not intended to be as comprehensive and flexible as a full modeling language like YALMIP or GAMS, but I suspect it is quite a bit faster. MP-Opt-Model also includes a unified interface to all of the solvers for each class of problems, namely qps_master() (was qps_matpower()) for LP and QP problems, miqps_master() (was miqps_matpower()) for mixed-integer problems, nlps_master() for general nonlinear optimization problems, nleqs_master() for nonlinear systems of equations. MP-Opt-Model<https://github.com/MATPOWER/mp-opt-model> is currently available on GitHub at https://github.com/MATPOWER/mp-opt-model, with the User's Manual<https://matpower.org/docs/MP-Opt-Model-manual.pdf> at available at https://matpower.org/docs/MP-Opt-Model-manual.pdf. The current master branch of the MATPOWER repository<https://github.com/MATPOWER/matpower> has also been updated to include and use MP-Opt-Model. There is still a lot that can be done to improve and extend MP-Opt-Model, and I hope this move will help make it easier for others to not only make use of it, but also help enhance it. Let me know what you think! Ray
