-
Notifications
You must be signed in to change notification settings - Fork 0
changelog
oravec-juraj edited this page Aug 13, 2024
·
19 revisions
wiki20240813.1
We recommend you use the up-to-date version of the MUP toolbox. Therefore, install the current version from download-page (releases), or update the toolbox using tbxmanager:
tbxmanager update mup
https://github.com/oravec-juraj/mup/wiki/MUP_R20240813
- complex compatibility update including:
- (backward) compatibility of mup_rmpcblock_init and mup_rmpcblock_demo is ensured subject to MATLAB R2019a - R2023a
- (backward) compatibility of the SOFT-CON module is ensured subject to MATLAB R2019a - R2023a
- license
- complex compatibility update including:
- (backward) compatibility of mup_rmpcblock_init and mup_rmpcblock_demo is ensured subject to MATLAB R2023a
- fixed bug in README: updated URL linking to the current homepage
- complex compatibility update including:
- homepage (including wiki) of MUP was moved from BITBUCKET to GITHUB
- help/support of mup_rmpc_block was redirected from BITBUCKET to GITHUB
- backward compatibility of mup_rmpcblock_init and mup_rmpcblock_demo is ensured subject to MATLAB R2014a and newer
- extended list of available "Robust MPC Approach":
- added "Shi et al. (2013)"
- extended list of supported solvers:
- added "SCS"
- updated MUPTEST to show the list of available Robust MPC design methods
- fixed bug in sdp_zhang ("Zhang et al., 2013") and wan_zhang ("NSO and SDLF"), the former release was overwriting constraint on semidefiniteness of SDLF matrix (W>=0), but the feasibility check was correct
- fixed bug in feas_huang ("Huang et al., 2011") and feas_mao_huang ("PDLF and WACIS"), the former release did not indicate the corresponding input index in feasibility check, but this bug in verbose mode had no impact on the validity of the feasibility check
- fixed bug in mup_rmpc_demo for WACIS-related alternative RMPC design methods: "NSO and WACIS", "PDLF and WACIS", because parameter BETA was assigned to their previous names ("WAN and HUANG", "MAO and HUANG")
none
- fixed bug in robust MPC design using RMPC_BLOCK, when no soft constraints are considered.
- fixed bug in robust MPC design SOFT-CON DEMO using RMPC_BLOCK.
- RMPC methods "Zhang et al. (2013)" and "NSO and SDLF" have a bug in SDP constraints, there is overwritten constraint on semidefiniteness of the SDLF matrix (W>=0), but the feasibility check is correct
- feasibility check of RMPC methods "Huang et al. (2011)", "PDLF and WACIS" does not indicate the corresponding input index in feasibility check, but this bug in verbose mode has no impact on the validity of the feasibility check
- in mup_rmpc_demo, the parameter BETA is not assigned correctly to WACIS-related alternative RMPC design methods: "NSO and WACIS", "PDLF and WACIS", because ("WAN and HUANG", "MAO and HUANG") fixed bug in mup_rmpc_demo for WACIS-related alternative RMPC design methods: "NSO and WACIS", "PDLF and WACIS", because
- fixed bug in robust MPC design DEMO using CLI.
- bug in robust MPC design using RMPC_BLOCK, when no soft-constraints are considered. Fixed in: MUP r20161207.
- bug in robust MPC design SOFT-CON DEMO using RMPC_BLOCK. Fixed in: MUP r20161207.
- introduced novel SOFT-CON module to implement LMI-based soft constraints on control inputs and system outputs. Currently supported soft-constrained RMPC design methods:
- Kothare et al. (1996) - soft-con,
- Wan et al. (2003) - soft-con,
- Cao et al. (2005) - soft-con,
- NSO and ACIS - soft-con.
- implemented SOFT-CON-related DEMO projects:
- mup_rmpc_soft_con_demo,
- mup_rmpcblock_soft_con_demo.
- bug in robust MPC design DEMO using CLI. Fixed in: MUP r20160529.
- updated list of supported SDP solvers based on the list of YALMIP-supported SDP solvers.
- extended features of RMPC block:
- if feasibility check is enabled then export the current value of optimizer GAMMA,
- export current value of SOLVER-TIME.
- introduced the benchmark uncertain systems:
- benchmark of the paper Kothare et al. (1996): benchmark_kothare,
- benchmark of the paper Cuzzola et al. (2002): benchmark_cuzzola.
- improved muptest.
- fixed bug in the evaluation of control law RMPC methods Cuzzola et al. (2002) and Ding et al. (2007).
- fixed bug in mup_rmpc_demo associated with the data storing.
- fixed bug in mup_rmpc_demo associated with the evaluation of the saturated control input when the control input is unconstrained.
- fixed bug in rmpc_plot that plots figures of RMPC trajectories.
none
- rebuild to take into account symmetric output constraints "y_max" instead of state constraints "x_max"
- adopted RMPC methods,
- adopted MATLAB/Simulink RMPC_BLOCK,
- adopted demos,
- implemented simple self-test run by muptest.
- updated mask of MATLAB/Simulink RMPC_BLOCK to open the MUP's homepage, when pushing the button HELP. If the page is not available, then the README file is opened.
- changed names of robust MPC methods:
- "Mao et Cao (2015)" renamed to "PDLF and ACIS".
- "Mao et Huang (2015)" renamed to "PDLF and WACIS".
- "Wan et Cao (2015)" renamed to "NSO and ACIS".
- "Wan et Huang (2015)" renamed to "NSO and WACIS".
- "Wan et Zhang (2015)" renamed to "NSO and SDLF".
none
- when the control law is designed using RMPC methods Cuzzola et al. (2002) or Ding et al. (2007) then MUP toolbox may crash. Fixed in MUP r20160211.
- when running mup_rmpc_demo then the MUP toolbox may crash when storing the data. Fixed in MUP r20160211.
- when running mup_rmpc_demo considering the unconstrained control input, then the MUP toolbox may crash during control input saturation. Fixed in MUP r20160211.
- when plot RMPC trajectories using rmpc_plot then the MUP toolbox may crash. Fixed in MUP r20160211.
- extended list of available "Robust MPC Approach":
- added "Mao et Cao (2015)".
- added "Mao et Huang (2015)".
- added "Wan et Cao (2015)".
- added "Wan et Huang (2015)".
- added "Wan et Zhang (2015)".
- updated YALMIP functions OPTIMIZE and VALUE, former SOLVESDP and DOUBLE, respectively.
- extended features of CLI
- fixed bug in sdp_ding and feas_ding, former release considered time-invariant uncertainties.
- fixed bug in sdp_li and feas_li one-step-ahead policy of vertex assignment.
- fixed bug in sdp_huang, there was missing the constraint: GAMMA - GAMMA_S > 0.
- fixed bug in sdp_zhang, there was an incorrect index in SDLF assigned.
- extended list of available "Robust MPC Approach":
- added "Ding et al. (2007)".
- added "Zhang et al. (2013)".
- extended list of supported solvers:
- added "MOSEK".
- extended features of CLI:
- enabled inline feasibility check,
- removed opt_$rmpc_name for CLI RMPC design. Robust MPC is designed by sdp_$rmpc_name in both, CLI and in Simulink by rmpc_block.
- extended mup_rmpc_demo
- ensured backward compatibility of mup_rmpcblock_init and mup_rmpcblock_demo with MATLAB R2012a and R2011b using mup_rmpcblock_demo_model_r2012a and mup_rmpcblock_demo_model_r2011b.
- ensured backwards compatibility of plotted figures using rmpc_plot_r13b with MATLAB R2013b and sooner.
- MUP toolbox is available for users of tbxmanager.
- fixed bug in mup_rmpc_demo in addtitional control input saturation
- RMPC method Ding et al. (2007) was implemented so that the time-invariant uncertainties were considered. Transformed for time-variant uncertainties in: MUP r20150616.
- RMPC method Li et al. (2008) has a bug in the one-step-ahead policy of vertex assignment. Fixed in: MUP r20150616.
- RMPC method Huang et al. (2011) has a bug in SDP constraints, there was a missing constraint related to the upper bounds of Lyapunov functions. Fixed in: MUP r20150616.
- RMPC method Zhang et al. (2013) has a bug in SDP constraints, there was an incorrect index in SDLF assign. Fixed in: MUP r20150616.
- fixed bug in the command-line-designed robust MPC using opt_cuzzola.m.
- figures plotted using rmpc_plot.m may not be displayed correctly in MATLAB R2014a and R2014b. Fixed in: MUP v20150202.
- inserted missing m-file get_variat.m.
- when designing the robust MPC directly in the command line by opt_cuzzola.m, then the robust MPC was not designed taking into account the robust invariant ellipsoid. When designing the robust MPC using the rmpc_block in MATLAB/Simulink (i.e., using sdp_cuzzola.m), then the obtained results were correct. Fixed in: MUP v20141127.
- extended list of available "Robust MPC Approach":
- "SLF" renamed to the "Kothare et al. (1996)"
- "PDLF" renamed to the "Cuzzola et al. (2002)"
- added "Mao et al. (2003)",
- added "Wan et al. (2003)",
- added "Cao et al. (2005)",
- added "Li et al. (2008)",
- added "Huang et al. (2011)".
- extended configuration of RMPC_BLOCK:
- added "Optimization Problem Design: initial/reformulation" - added possibility to use repetitive reformulation of the optimization problem.
- added "Parameters".
- extended outputs of RMPC_BLOCK:
- added "feasibility: -1/0/1".
- fixed bug in the feasibility check of "Cuzzola et al. (2002)".
- missing m-file get_variat.m. Fixed in: MUP v20141120.
- initial version
- when "Feasibility check" = "enable" and "Robust MPC Approach" = "PDLF", then the feasibility check may return information about infeasible solution also in case the SDP was in fact feasible. Fixed in: MUP v20140530.