v0.24.0

ModelPredictiveControl v0.24.0

Diff since v0.23.1

• breaking change: MovingHorizonEstimator now default to direct=true
• added: the current/filter formulation of the MovingHorizonEstimator
• doc: explicitly list all the keyword arguments of all the state estimator
• added: soft real time utilities
• added: InternalModel now produces 0 allocation with preparestate! and updatestate! calls
• tests: new integration tests that compare unconstrained MHE to UKF and KF results
• tests: new integration that compare LinModel and NonLinModel

Merged pull requests:

• Added: soft real time utilities (#94) (@franckgaga)
• added: MovingHorizonEstimator support for direct=true, initialized with P̂(-1|-1) (#96) (@franckgaga)

Closed issues:

• Real-time loop tools (#85)

v0.23.1

ModelPredictiveControl v0.23.1

Diff since v0.23.0

• Luenberger, SteadyKalmanFilter, KalmanFilter and UnscentedKalmanFilter now produce 0 allocation with preparestate! and updatestate! calls
• doc: minor corrections

Merged pull requests:

• added: doc correction and reduce allocations for all Kalman filters (#93) (@franckgaga)

Closed issues:

• Support for current estimators and V1 soon? (#89)

v0.23.0

ModelPredictiveControl v0.23.0

Diff since v0.22.1

BREAKING CHANGES

• preparestate! should be called before solving the MPC optimization with moveinput!
• current form by default for all StateEstimator
• ym keyword argument of moveinput! removed (was only for InternalModel, I now use the new preparestate! method)
• MovingHorizonEstimator advanced constructor (with complete covariance matrices): covestim and optim are now keyword arguments instead of positional arguments.

Changelog

• added: current form for all StateEstimator except MovingHorizonEstimator, to improve accuracy and closed-loop robustness
• reduce allocation for SimModeland StateEstimator instances by using internal buffers
• multiple doc correction

Merged pull requests:

• minor doc correction (#87) (@franckgaga)
• minor doc additions (#88) (@franckgaga)
• added: support for the current form for all the state estimators (#90) (@franckgaga)
• Reduce allocations by avoiding empty for default d argument (#91) (@franckgaga)
• added: SimModelBuffer and StateEsimatorBuffer objects to reduce the allocations (#92) (@franckgaga)

v0.22.1

ModelPredictiveControl v0.22.1

Diff since v0.22.0

• added: support for Mwt, Nwt and Lwt keyword arguments in setmodel! (to simplify the API)
• reduce allocations in setconstraint!
• minor doc corrections

Merged pull requests:

• non unicode alternative setmodel MPC (#79) (@franckgaga)
• minor doc correction (#80) (@franckgaga)
• idem (#81) (@franckgaga)
• doc: change nomenclature EMPC (#83) (@franckgaga)
• Minor correction doc (#84) (@franckgaga)
• Added: Mwt, Nwt and Lwt kwargs in setmodel! (#86) (@franckgaga)

Closed issues:

• How to successfully impose constraints to a Control Variable with high priority (#78)

v0.22.0

ModelPredictiveControl v0.22.0

Diff since v0.21.3

BREAKING CHANGE

Migration the new nonlinear programming syntax of JuMP. Some optimizers may not support it yet.

• added: online modification of objective function weights and covariance matrices with setmodel!
• added: accept indices and ranges in plot keyword arguments
• changed: migrate the NLP code to the new syntax
• doc: added plot result previews with static images
• new tests for new plot recipes and weights/covariances modification

Merged pull requests:

• New jump nlp syntax, merge into main? (#31) (@franckgaga)
• Added: online modification of weights and covariances (#77) (@franckgaga)

Closed issues:

• MPC getinfo results (#76)

v0.21.3

ModelPredictiveControl v0.21.3

Diff since v0.21.2

• added: index ranges or vectors for plot keyword arguments (to select variables to plot)
• doc: separate page for sim! and plot functions
• doc: document plot keyword arguments
• doc: plot results in examples

Merged pull requests:

• Doc: minor corrections and add preview cleanup workflow (#71) (@franckgaga)
• cleanup MHE append 0 to deviation vector variables (#72) (@franckgaga)
• added: support indices keyword argument in plot recipes (#73) (@franckgaga)
• bump (#74) (@franckgaga)

v0.21.2

ModelPredictiveControl v0.21.2

Diff since v0.21.1

• added: show sim! % progress and ETA in VS Code status bar
• added: plot estimated state constraints x̂min/x̂max for MHE
• debug: RungeKutta diff cache chuck size nx+2
• doc: added plot recipe documentation (for keyword arguments)
• tests: unconstrained MHE vs KF/UKF tests

Merged pull requests:

• debug: RungeKutta diff cache chuck size nx+2 (#67) (@franckgaga)
• minor doc correction (#68) (@franckgaga)
• added: plot recipes doc and plot MHE state estimate constrains (#69) (@franckgaga)
• Update Project.toml (#70) (@franckgaga)

v0.21.1

ModelPredictiveControl v0.21.1

Diff since v0.21.0

• added: non-Unicode alternative keyword arguments in public functions
• added: modify plots Y-axis labels with setname!
• debug: plots Ru instead of Ry for u setpoint (recipe)
• remove one useless allocation in updatestate!
• various doc corrections and improvements

Merged pull requests:

• doc: correct mistake in MPC prediction matrices (#56) (@franckgaga)
• Doc correction (#57) (@franckgaga)
• doc: clarify notation update_estimate! and imc block diagram (#58) (@franckgaga)
• IMC block diagram (#59) (@franckgaga)
• remove one allocation in updatestate! (#60) (@franckgaga)
• debug: plot Ru instead of Ry for u setpoint in recipe (#62) (@franckgaga)
• added: setname! function for variable names in plot labels (#63) (@franckgaga)
• Offer an alternative non-unicode API for keyword arguments (#64) (@franckgaga)
• Update Project.toml (#66) (@franckgaga)

Closed issues:

• Remove unicode char in keyword arguments (or offer an alternative) ? (#61)

v0.21.0

ModelPredictiveControl v0.21.0

Diff since v0.20.2

BREAKING CHANGE
All the keyword arguments related to initial values e.g. σP0, x0 and x̂0 now require an underscore e.g. σP_0, x_0, x̂_0 (to differentiate from operating point deviation vectors)

• Added: setmodel! for runtime model adaptation of controller/estimator based on LinModel
• Added: linearize and setop! now support non-equilibrium points
• Added: successive linearization MPC with the new setmodel! and linearize functions
• Added: successive linearization MHE with the new setmodel! and linearize functions
• Added: linearize! method for in-place model linearization (to reduce allocations)
• Added: 6 args. LinModel constructor now support scalars (similarly to ss function)
• Added: ExtendedKalmanFilter now compute the Jacobians in-place (to reduce allocations)
• Changed: struct state data x and state estimate x̂ renamed to x0 and x̂0
• Debug: ExplicitMPC with non-Float64 now works
• Debug: accept integers in linearize arguments
• Debug: call empty! on JuMP.Model to support re-construction of MPC instances
• Doc: new setmodel!, setop! and linearize function documentation
• Doc: example of model adaptation with successive linearization on the pendulum (very efficient!)

Merged pull requests:

• Added: adaptatation of LinMPC model through setmodel! (#52) (@franckgaga)
• Added: adaptation of controller/estimator based on LinModel (#54) (@franckgaga)
• Added: setmodel! for MovingHorizonEstimator (#55) (@franckgaga)

v0.20.2

ModelPredictiveControl v0.20.2

Diff since v0.20.1

• added: print info on controller and estimator constraint softening (slack var. ϵ)
• reduce allocations LinMPC, NonLinMPC, MovingHorizonEstimator
• cleanup namespace with imports instead of usings

Merged pull requests:

• Changed: avoid collect in NonLinMPC and MovingHorizonEstimator objective and constraints (#46) (@franckgaga)
• changed: error handling one less alloc (#47) (@franckgaga)
• added: print info on controller and estimator constraints softening (#49) (@franckgaga)
• cleaning the namespace with imports instead of usings (#50) (@franckgaga)
• Namespace, NonLinMPC and MovingHorizonEsitimator cleanup (#51) (@franckgaga)