question

If you want to export graphs, then I don't know if that is possible outside of using the html report-builder.

What you can do is export the numerical data to an Excel file. The following code looks up the id of the best iteration in the last optimizer run and then exports the data of a state bar graph for that iteration to Excel. (If you don't specify the Excel file path further it will be created in the documents folder)

Database.Connection db = function_s(Model.find("Tools/Experimenter"), "getDBConnection");

int disconnect = 0;
if (!db.isConnected) {
    if (!db.connect()) {
        return 0;
    }
    disconnect = 1;
}

// Query for ID of best iteration
Database.PreparedStatement stmt = db.prepareStatement("SELECT best_solutions FROM optquest_run_results");
var result = stmt.execute();
string lastBest;
while (result.fetchNext()) {
    lastBest = result[1];
}
   
lastBest = lastBest.slice(2, -1);
int bestID = lastBest.toNum();

// Create new Excel workbook
excelcreateworkbook();
excelcreatesheet("StateBarData");
excelsetsheet("StateBarData");
excelwritestr(1, 1, "Object");
excelwritestr(1, 2, "State");
excelwritestr(1, 3, "Time");
excelwritestr(1, 4, "Utilization");
int row = 2;
   
// Query for state bar data and write to excel
stmt = db.prepareStatement("SELECT * FROM [/State Bar] WHERE task_id == ?");
stmt.bindParam(1, bestID, Database.DataType.SmallInt);
result = stmt.execute();
while (result.fetchNext()) {
    string objectPath = StatisticsCollector.getPathFromID(result[2], 1);
    excelwritestr(row, 1, objectPath);
    excelwritestr(row, 2, result[3]);
    excelwritenum(row, 3, result[4]);
    excelwritenum(row, 4, result[5]);
    row++;
}
excelsave("Optimizer_Iteration_" + lastBest + ".xlsx");
excelclose(0);

if(disconnect) {    
    db.disconnect();
}

@Jason Lightfoot actually gave a similar example very recently here.

I used the freeware tool linked below to explore the structure of the sqlite database file and get the correct table and columns to use in the queries.

https://sqlitebrowser.org/about/

The model attached below actually also contains my attempt at finding a solution for your question about dynamically skipping iterations.

My approach was to store the current best parameter values (for the example I just used the sum) in a database which the child processes running the iteration would then access at the start of each replication. If the current parameter set is worse, the run would be stopped immediately with "endreplication(1)".

From this post I got the idea to use the results file itself to store the best parameter set. It does mostly work, however, reading the best values from the database is not always successful (possibly due to parallel access attempts), so some "worse" iteration are still run.

-1 values are skipped iterations. 0 means unsuccessful, >0 successful runs (determined simply by range of parameters).

I originally decided not to post this, as I am not sure if this approach might cause other (stability) issues in larger runs.

The site won't let me upload the sqlite file, so I'll attach the model so you can run the optimization again to test the code above.

experimenter_db_test_3.fsm