private void GenerateGraph()
{
// reset textbox colors in case of prior input errors
textBoxInitialEstimate.BackColor = System.Drawing.Color.White;
textBoxInitialEstimateError.BackColor = System.Drawing.Color.White;
textBoxMeasurementNoise.BackColor = System.Drawing.Color.White;
textBoxProcessNoise.BackColor = System.Drawing.Color.White;
textBoxIntervals.BackColor = System.Drawing.Color.White;
double initialEstimate;
if (!double.TryParse(textBoxInitialEstimate.Text, out initialEstimate))
{
initialEstimate = 0.0;
textBoxInitialEstimate.BackColor = System.Drawing.Color.Red;
}
double initialEstimateError;
if (!double.TryParse(textBoxInitialEstimateError.Text, out initialEstimateError) || initialEstimateError < 0)
{
initialEstimateError = 1.0;
textBoxInitialEstimateError.BackColor = System.Drawing.Color.Red;
}
double measurementNoise;
if (!double.TryParse(textBoxMeasurementNoise.Text, out measurementNoise) || measurementNoise < 0)
{
measurementNoise = 0.025;
textBoxMeasurementNoise.BackColor = System.Drawing.Color.Red;
}
double processNoise;
if (!double.TryParse(textBoxProcessNoise.Text, out processNoise) || processNoise < 0)
{
processNoise = 0.01;
textBoxProcessNoise.BackColor = System.Drawing.Color.Red;
}
int numIntervals;
if (!int.TryParse(textBoxIntervals.Text, out numIntervals) || numIntervals < 1)
{
numIntervals = 100;
textBoxIntervals.BackColor = System.Drawing.Color.Red;
}
KalmanValues results = KalmanSimulator.ApplyKalmanFilter(initialEstimate, initialEstimateError, measurementNoise, processNoise, numIntervals);
chartMain.Series["Actual"].Points.DataBindY(results.ActualSignal);
chartMain.Series["Measured"].Points.DataBindY(results.MeasuredValues);
chartMain.Series["Filtered"].Points.DataBindY(results.FilteredValues);
}
public static KalmanValues ApplyKalmanFilter(double initialEstimate, double initialEstimateError, double measurementNoise, double processNoise, int numIntervals)
{
KalmanValues kalmanValues = new KalmanValues();
// Generate "actual" signal data
kalmanValues.ActualSignal = RandomWalkSignal(numIntervals, processNoise);
// Simulate inaccurate measurements by adding noise to signal
kalmanValues.MeasuredValues = MeasuredSignalWithNoise(kalmanValues.ActualSignal, measurementNoise);
// Apply the filter to the measured data
kalmanValues.FilteredValues = FilterMeasurements(kalmanValues.MeasuredValues, initialEstimate, measurementNoise, initialEstimateError, processNoise);
return(kalmanValues);
}
