v0.19.0

ModelPredictiveControl v0.19.0

Diff since v0.18.1

BREAKING CHANGE
NonLinModel constructor now supposes continuous dynamics by default. Use solver=nothing for discrete-time models.

changelog:

• added: support for NonLinModel with continuous dynamics
• added: 4th order Runge-Kutta solver with 0 allocation
• doc: the pendulum example now use the built-in Runge-Kutta solve
• added some tests with continuous NonLinModel

Merged pull requests:

• starting support of continuous NonLinModel (RK4 only for now) (#38) (@baggepinnen)

v0.18.1

ModelPredictiveControl v0.18.1

Diff since v0.18.0

• remove useless allocations for mutating NonLinModels

Merged pull requests:

• minor doc correction (#36) (@franckgaga)
• Remove useless allocations for in-place NonLinModel (#37) (@franckgaga)

v0.18.0

ModelPredictiveControl v0.18.0

Diff since v0.17.0

• added: support for mutating f! and h! functions in NonLinModel to reduce allocations
• doc: added simple example of the MovingHorizonEstimator for the CSTR in the manual

Merged pull requests:

• Update README.md (#27) (@franckgaga)
• Added a MHE example on the CSTR in the manual (#28) (@franckgaga)
• fix wrong function signature ared (#29) (@baggepinnen)
• dummy commit to trigger doc deploy (#30) (@franckgaga)
• doc: minor correction MHE manual (#32) (@franckgaga)
• Added: mutating NonLinModel (#33) (@franckgaga)
• Re-enable precompilation (#34) (@franckgaga)
• bump version (#35) (@franckgaga)

Closed issues:

• Docs questions (#22)
• Documentation no longer deployed (#26)

v0.17.0

ModelPredictiveControl v0.17.0

Diff since v0.16.1

• added: support for non-diagonal symmetric M_Hp, N_Hc and L_Hp weights, allowing terminal costs and LQR tuning
• doc: added comment on a terminal weight
• reduce allocations for PredictiveControllers and MovingHorizonEstimator (with in-place operations)

Merged pull requests:

• Update README links (#21) (@baggepinnen)
• todo: add terminal costs (#23) (@franckgaga)
• Update README.md (#24) (@franckgaga)
• Added: terminal cost support (#25) (@franckgaga)

v0.16.1

ModelPredictiveControl v0.16.1

Diff since v0.16.0

• added: reduce allocation NonLinMPC
• doc: added internal function optim_objective
• doc: various doc improvements

v0.16.0

ModelPredictiveControl v0.16.0

Diff since v0.15.0

• added: MHE soft constraints with a single slack variable ϵ (a.k.a. equal concern for relaxation, disabled by default thus only hard constraints).
• changed: more concise MPC logging in REPL.
• removed: deprecated arguments for setconstraint! and NonLinMPC.
• doc: hide extended help with details admonition.
• doc: more concise setconstraint! and ExplicitMPC documentation.

v0.15.0

ModelPredictiveControl v0.15.0

Diff since v0.14.1

• added: MHE hard constraints on estimated process and sensor noises
• added: MHEs based on LinModel are now treated as a quadratic optimization instead of nonlinear
• added: Plots recipe avoid repeat xlabels for MIMOs
• added: getinfo method for MovingHorzionEstimator for troubleshooting
• debug: update MHE arrival covariance with I/Os at arrival
• improved coverage

Merged pull requests:

• Added: MHE with quadratic optim + process and sensor noises bounds (#20) (@franckgaga)

v0.14.1

ModelPredictiveControl v0.14.1

Diff since v0.14.0

• changed: default P0 value not function of Q argument

v0.14.0

ModelPredictiveControl v0.14.0

Diff since v0.13.0

• added: MovingHorizonEstimator that can handle constraints on estimation
• added: tests for MovingHorizonEstimator

Merged pull requests:

• Added: MovingHorizonEstimator with state constraints (#19) (@franckgaga)

v0.13.0

ModelPredictiveControl v0.13.0

Diff since v0.12.0

• added: generic number types in all SimModel, StateEstimator and PredictiveController objects (see note below)
• changed: faster ΔU -> U conversion using matrices of floats (instead of booleans)
• new tests with Float32.

Note that most solvers in JuMP only support Float64 numbers. Thus, PredictiveController objects still default to Float64-based optimizers, even if the plant model or state estimator uses other number types.

Merged pull requests:

• Support non-Float64 numbers in all objects (#17) (@franckgaga)