Skip to content

Commit

Permalink
Refactoring ClosedLoopIdentifier ahead of rewrite.
Browse files Browse the repository at this point in the history
  • Loading branch information
Steinar Elgsæter authored and Steinar Elgsæter committed Dec 4, 2024
1 parent 95e3126 commit baca029
Show file tree
Hide file tree
Showing 5 changed files with 273 additions and 238 deletions.
72 changes: 43 additions & 29 deletions Dynamic/Identification/ClosedLoopUnitIdentifier.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -108,8 +108,8 @@ public static (UnitModel, double[]) Identify(UnitDataSet dataSet, PidParameters
double[] u0 = SignificantDigits.Format(dataSet.U.GetRow(0), nDigits);
double y0 = dataSet.Y_meas[0];
bool isOK;
var dataSet1 = new UnitDataSet(dataSet);
var dataSet2 = new UnitDataSet(dataSet);
var dataSetRun1 = new UnitDataSet(dataSet);
var dataSetRun2 = new UnitDataSet(dataSet);
FittingSpecs fittingSpecs = new FittingSpecs();
fittingSpecs.u0 = u0;

Expand All @@ -123,18 +123,17 @@ public static (UnitModel, double[]) Identify(UnitDataSet dataSet, PidParameters
{

/////////////////
// step1:
// run1, step1:
DisturbanceIdResult distIdResult1 = DisturbanceIdentifier.EstimateDisturbance
(dataSet1, null, pidInputIdx, pidParams);
(dataSetRun1, null, pidInputIdx, pidParams);

dataSet1.D = distIdResult1.d_est;
var unitModel_run1 = UnitIdentifier.IdentifyLinearAndStatic(ref dataSet1, fittingSpecs, doTimeDelayEstOnRun1);
// nGains = unitModel_run1.modelParameters.GetProcessGains().Length;
dataSetRun1.D = distIdResult1.d_est;
var unitModel_run1 = UnitIdentifier.IdentifyLinearAndStatic(ref dataSetRun1, fittingSpecs, doTimeDelayEstOnRun1);
idDisturbancesList.Add(distIdResult1);
idUnitModelsList.Add(unitModel_run1);
isOK = ClosedLoopSim(dataSet1, unitModel_run1.GetModelParameters(), pidParams, distIdResult1.d_est, "run1");
isOK = ClosedLoopSim(dataSetRun1, unitModel_run1.GetModelParameters(), pidParams, distIdResult1.d_est, "run1");

// "step2" global search" for linear pid-gainsgains
// run1, "step2" : "global search" for linear pid-gainsgains
if(unitModel_run1.modelParameters.GetProcessGains()!= null)
{
wasGainGlobalSearchDone = true;
Expand All @@ -152,9 +151,7 @@ public static (UnitModel, double[]) Identify(UnitDataSet dataSet, PidParameters
var max_gain = Math.Abs(pidProcessInputInitalGainEstimate * initalGuessFactor_higherbound);
var min_gain = - max_gain;

// when debugging, it might be advantageous to set min_gain equal to the known true value
//TODO:Remove!!!
// min_gain = 0;
// min_gain = 0; // when debugging, it might be advantageous to set min_gain equal to the known true value

// first pass(wider grid with larger grid size)
var retPass1 = GlobalSearchLinearPidGain(dataSet, pidParams, pidInputIdx,
Expand All @@ -181,43 +178,48 @@ public static (UnitModel, double[]) Identify(UnitDataSet dataSet, PidParameters
}
}
}
bool doStep3 = true;

// ----------------

// - issue is that after run 4 modelled output does not match measurement.
// - the reason that we want this run is that after run3 the time constant and
// ----------------
// - issue is that after run 1 modelled output does not match measurement.
// - the reason that we want this run is that after run1 the time constant and
// time delays are way off if the process is excited mainly by disturbances.
// - the reason that we cannot do run4 immediately, is that that formulation
// - the reason that we cannot do run2 immediately, is that that formulation
// does not appear to give a solution if the guess disturbance vector is bad.

const bool doRun2 = true;
const double LARGEST_TIME_CONSTANT_TO_CONSIDER_TIMEBASE_MULTIPLE = 30;//too low

// run2: do a run where it is no longer assumed that x[k-1] = y[k],
// this run has the best chance of estimating correct time constants, but it requires a good inital guess of d
if (doStep3 && idUnitModelsList.Last() != null)

// run 2 version 1: try looking for the time constant that gives the smallest average disturbance amplitude

if (doRun2 && idUnitModelsList.Last() != null)
{
var model = idUnitModelsList.Last();

DisturbanceIdResult distIdResult2 = DisturbanceIdentifier.EstimateDisturbance
(dataSet2, model, pidInputIdx, pidParams);
List<double[]> estDisturbances = new List<double[]>();
List<double> distDevs = new List<double>();
var distIdResult2 = DisturbanceIdentifier.EstimateDisturbance
(dataSetRun2, model, pidInputIdx, pidParams);
var estDisturbances = new List<double[]>();
var distDevs = new List<double>();

estDisturbances.Add(distIdResult2.d_est);
var timeBase = dataSet2.GetTimeBase();
var timeBase = dataSetRun2.GetTimeBase();
double candiateTc_s = 0;
bool curDevIsDecreasing = true;
double firstDev = vec.Sum(vec.Abs(vec.Diff(distIdResult2.d_est))).Value;
distDevs.Add(firstDev);
while (candiateTc_s < 30 * timeBase && curDevIsDecreasing)
while (candiateTc_s < LARGEST_TIME_CONSTANT_TO_CONSIDER_TIMEBASE_MULTIPLE * timeBase && curDevIsDecreasing)
{
candiateTc_s += timeBase;
var newParams = model.GetModelParameters().CreateCopy();
newParams.TimeConstant_s = candiateTc_s;
var newModel = new UnitModel(newParams);
DisturbanceIdResult distIdResult_Test = DisturbanceIdentifier.EstimateDisturbance
(dataSet2, newModel,pidInputIdx,pidParams);
var distIdResult_Test = DisturbanceIdentifier.EstimateDisturbance
(dataSetRun2, newModel,pidInputIdx,pidParams);
estDisturbances.Add(distIdResult_Test.d_est);
double curDev = vec.Sum(vec.Abs(vec.Diff(distIdResult_Test.d_est))).Value;
var curDev = vec.Sum(vec.Abs(vec.Diff(distIdResult_Test.d_est))).Value;
if (curDev < distDevs.Last<double>())
curDevIsDecreasing = true;
else
Expand All @@ -235,11 +237,23 @@ public static (UnitModel, double[]) Identify(UnitDataSet dataSet, PidParameters
step2params.TimeConstant_s = candiateTc_s;
var step2Model = new UnitModel(step2params);
idUnitModelsList.Add(step2Model);
DisturbanceIdResult distIdResult_step4 = DisturbanceIdentifier.EstimateDisturbance
(dataSet2, step2Model, pidInputIdx, pidParams);
var distIdResult_step4 = DisturbanceIdentifier.EstimateDisturbance
(dataSetRun2, step2Model, pidInputIdx, pidParams);
idDisturbancesList.Add(distIdResult_step4);
}

{
// version 2: choose the time-constant that gives the best closed-loop plant fit
/* var pidModel = new PidModel(pidParams, "PID");
(var plantSim, var inputData) = PlantSimulator.CreateFeedbackLoop(unitDataSet_raw, pidModel, unitModel, pidInputIdx);
plantSim.Simulate(inputData, out var simData);
var test = plantSim.PlantFitScore;
*/
}


// debugging plots, should normally be off
if (false)
{
Expand Down
Loading

0 comments on commit baca029

Please sign in to comment.