Simulink Design Optimization 1.1

Calibrating Model Parameters with Test Data

Using Simulink Design Optimization to calibrate parameters with test data involves three steps:

• Importing and preprocessing data
• Estimating parameters
• Comparing and validating estimations

Importing and Preprocessing Data

Simulink Design Optimization can use measured input-output data from actual hardware to estimate and validate the parameters of a Simulink model. Simulink Design Optimization lets you import measured data from MATLAB, Microsoft® Excel®, ASCII, and CSV files, as well as from the MATLAB workspace. Measured data often has offsets, outliers, periods of missing values, and other anomalies that can lead to inaccurate parameter estimation. Simulink Design Optimization lets you preprocess your measured data to remove these sources of error by:

• Detrending to remove data drift and offset
• Filtering to remove noise and band-limited disturbances
• Interpolating to fill in missing values
• Excluding to remove questionable sections of the data set

Estimating Parameters

Simulink Design Optimization lets you set up and maintain multiple estimation tasks. For each estimation task, you can specify which model parameters and initial conditions you want to estimate and which sets of input-output data you want to use. This approach lets you estimate parameters for one section of the model by using one combination of data sets and independently estimate parameters for other model sections by using different combinations of data sets. You can refine the parameter tuning process by using parameter values from previous estimation tasks as initial values for subsequent estimations or set ranges for estimated parameters.

In addition to estimating model parameters, Simulink Design Optimization provides functionality to estimate static lookup table values and a Simulink block for implementing adaptive lookup tables. You can connect your adaptive lookup table directly to an actual system by compiling your Simulink model and implementing the code using an appropriate host, such as xPC Target™.

Comparing and Validating Estimations

Simulink Design Optimization can generate comparative plots of estimation results to help determine which model parameter values result in the best model and measured data fit. Plots include views of parameter sensitivity, measured versus simulated model outputs, and residual values.

Validation involves comparing the model output with an independent set of test data to determine whether the calibrated model accurately represents the system dynamics. Simulink Design Optimization lets you compare multiple model outputs against the validation data set to select the best estimation and parameter sets.

Preprocessing data using Simulink Design Optimization to remove outliers and unwanted trends. Click on image to see enlarged view.