Use Response Plots to Analyze Batch Linearization Results

This example shows how to plot and analyze the step response for batch linearization results. Batch linearization results refers to the ss model array returned by the slLinearizer interface or linearize function. This array contains linearizations for varying parameter values, operating points, or both. You can use the technique in this example to analyze the frequency response, stability, or sensitivity for batch linearization results.

Obtain batch linearization results.

Load the batch linearization results saved in scd_batch_lin_results1.mat.

load scd_batch_lin_results1 linsys;

linsys, a 4 x 3 x 2 ss model array, contains the closed-loop transfer function of the watertank model from the reference input to the plant output. linsys was obtained for four simulation snapshot times, t = [0 1 2 3], by varying the model parameters, A and b. The sample values for A are [10 20 30], and the sample values for b are [4 6].

Plot the step response for the linearized models.

stepplot(linsys);

View the parameter combination and simulation snapshot time that yielded a specific response.

Click the response.

A data tip appears on the plot, providing information about the selected response and the related model. The last lines of the data tip show the parameter combination and simulation snapshot time that yielded this response. In the previous plot, the selected response corresponds to the model obtained by setting A to 30 and b to 6. The software linearized the model after simulating the model for two time units.

Plot the step response for a subset of the batch linearization results.

Suppose you want to view the responses for models linearized at a specific simulation snapshot time, for example two time units. Right-click the plot and select Array Selector. The Model Selector for LTI Arrays dialog box opens.

The Selection Criterion Setup panel contains three columns, one for each model array dimension of linsys. The first column corresponds to the simulation snapshot time. The third entry of this column corresponds to the simulation snapshot time of two time units. Select only this entry in the first column.

Click OK. The plot displays the responses for only the models linearized at two time units.

Plot the step response for a specific parameter combination and simulation snapshot time.

Suppose you want to view the step response for the model obtained by linearizing the watertank model at t = 3, for A = 10 and b = 4. To do so, you can use the SamplingGrid property of the linearized models, which is specified as a structure. When you perform batch linearization, the software populates the SamplingGrid structure with information regarding the variable values used to obtain the model. Variable values includes each parameter that you vary and the simulation snapshot times at which you linearize the model. For example:

linsys(:,:,1).SamplingGrid
ans = 

       A: 10
       b: 4
    Time: 0

Here linsys(:,:,1) refers to the first model in linsys. This model was obtained at simulation time t = 0, for A = 10 and b = 4.

Extract, from linsys, the model obtained by linearizing the watertank model at t = 3, for A = 10 and b = 4:

sg = linsys.SamplingGrid;

sys = linsys(:,:,sg.A == 10 & sg.b == 4 & sg.Time == 3);

The structure, sg, contains the sampling grid for all the models in linsys. The expression sg.A == 10 & sg.b == 4 & sg.Time == 3 returns a logical array. Each entry of this array contains the logical evaluation of the expression for corresponding entries in sg.A, sg.b, and sg.Time. sys, a model array, contains all the linsys models that satisfy the expression.

View the step response for sys.

stepplot(sys);

Related Examples

Was this topic helpful?