generating/extracting a simulink restart state from system output? ?problem with 'discrete' blocks?

I need to periodically interrupt a SIMULINK model, and restart it with the the state it was in at some previous time (with some changes to model parameters using environmental variables).

The concise questions is: how do i take 'XXX = sim(gcs)' and define a restart state from a timeslice of 'XXX'?

The gist of what i'm trying to do is as follows:

%% LOAD SYSTEM
if ~isempty(gcs); close_system(gcs,0); end
SIMSYS = load_system(systemname)   ;           

setActiveConfigSet( SIMSYS, 'Configuration1' );
cset =  getConfigSet( gcs, 'Configuration1' );    
set_param(cset,'SaveFormat',    'StructureWithTime' );
set_param(cset,'StateSaveName', 'state_history');                                

% SIMULATE over intervals
for kk = 1:5; %length(tstarts); %1
    tstart   = tstarts(kk);
    tend     = tends(kk);    
   
    set_param(cset,'StartTime',     string(tstart) );
    set_param(cset,'StopTime',      string(tend)   );             
    set_param(cset,'SaveState',     'on');
     
    if kk==1;   % set initial details and paramter values
        set_param(cset,'Solver','VariableStepAuto')
        set_param(cset,'SolverName','VariableStepAuto')
        set_param(cset, 'LoadInitialState','off');        
        set_param(cset, 'LoggingToFile','off');        
        B.R = 3;
        B.C = 0.12;
        B.P = 17.0;   
    end     

     try 
         simOUT = sim( gcs, cset ); 
     catch ME
         rethrow(ME)
     end
     
     % modify things for next iteration
     set_param(cset, 'LoadInitialState','on');     
     
     % MAKE a variable 'rststate' from the 
     rststate   = simOUT.state_history;
     % get record time closest to next 
     [~,rst_kt] = min( abs(rststate.time - tstart) );
     
     %***
     
     for rstkk = 1:length(rststate.signals);
                 rststate.signals(rstkk).values = rststate.signals(rstkk).values( rst_kt );
     end
     rststate.time = rststate.time(rst_kt);
      
      % SET intial state as restart state            
      set_param(cset, 'InitialState',    'rststate' );
      %[B.R, B.C, B.P] = parameter_update(some system monitors)
end                                   
system_close(gcs,0) 

Unfortunately, this fails at the SECOND call to 'sim' with the error

Size mismatch in the initial state for model 'your_model'. The signals(136).values field in the initial state structure has
1 element(s), where as the corresponding state 'Discrete' in block 'your_model/Solver Configuration' has 2 element(s)

What I've done before is manually remove these blocks from the restart state at line *** a la:

rststate.signals(136)=[]  % or similar

but this works inconsistently and is model specific. Namely, when i do remove entry 136, which corresponds to the Solver Configuration, I get the following error which is absolutely perplexing:

Unable to load the specified initial state for model 'your_model'. Cannot find a matching block state
corresponding to element 135 of signals structure array

To me , it seems like the indexing ouf the output state and the initial state are different. THat's not a typo; removing element 136 from the initial signal gives a model that element 135 is not provided. The following don't work either, in the sense that they remove the 136th signal element, and produces and undefined element 135.

%rmblk = find( ismember( {rststate.signals(1:end).label}, 'Discrete') )
rmblk = find( contains( {rststate.signals(1:end).blockName },'Solver'));
rststate.signals(rmblk) = [];

Answer 1

[Fangjun Jiang]

Just got this:

https://www.mathworks.com/help/simulink/slref/saving-and-restoring-simulation-operating-point.html

[jessupj]

Thanks for that link, and for your attention to my plight. It nearly answers the question. I'll work with it and see if i can find a way to keep the simscape solver block from breaking the restart since that's where the problem seems to lie.

In all those examples, they restart the model at the end of the previous run e.g. a run for t \in [0,5] and a restart it at t=5. What i need is a run over t \in [0,10] and a restart at t=5, or some other point.

i guess one could run run over t \in [0,5], drop a restart/final state for t=5, then run t \in [5,10], concatenate the two separate histories, perfrom analysis, and then restart at t=5. also: i don't need a 'perfect restart', and don't always know a priori at what time point i will be restarting the model. i simply want to grab a restart state from some intermediate point of the simulation output.

[jessupj]

Thanks for the help, both here and in the related/consequent problem https://www.mathworks.com/matlabcentral/answers/594232-simulink-simscape-final-state-cannot-find-a-matching-block-state-to-element-135-output-by-the-same