Example Requirements

During this example, Simulink® and Simulink® Coder™ generate code in the code generation folder created in the current directory. If you do not want to (or if you cannot) generate files in this directory, you should change your working directory. Simulink Coder is required to generate model reference binaries to be deployed in standalone applications.

If you plan to alter the example models:

1. Preserve the example in its original state by copying the following files from your MATLAB® installation directory, without changing their names, to a different directory:

toolbox/simulink/simdemos/simfeatures/modelreference/sldemo_mdlref_datamngt.slx
toolbox/simulink/simdemos/simfeatures/modelreference/sldemo_mdlref_datamngt_wsdata.m
toolbox/simulink/simdemos/simfeatures/modelreference/sldemo_mdlref_counter_datamngt.slx
toolbox/simulink/simdemos/simfeatures/modelreference/sldemo_mdlref_counter_datamngt_types.m
toolbox/simulink/simdemos/simfeatures/modelreference/sldemo_mdlref_counter_datamngt_wsdata.m

2. Change your current directory to the directory to which you copied the files.

3. Continue with the example.

Open the Example Model

open_system('sldemo_mdlref_datamngt')

Example Content

This example uses a top model (sldemo_mdlref_datamngt) that contains three Model blocks: Counter1, Counter2, and Counter3. These blocks reference the same model (sldemo_mdlref_counter_datamngt).

The referenced model implements a limited counter algorithm that:

The referenced model outputs a bus signal that contains:

This model is configured to show a number of Simulink features working together:

This example also uses other features that are useful for managing data in Simulink:

Example Outline

The sequence of steps in the example is as follows:

1. Prepare the referenced model (sldemo_mdlref_counter_datamngt) to use structures for the signals and parameters in its external interface.

2. Prepare the top model (sldemo_mdlref_datamngt) to call the referenced model.

3. Simulate the top model and examine the results.

4. Generate code to create a standalone executable for the top model.

Referenced Model: Setting Up Parameters

The referenced model has two model arguments (CounterParams, CounterICs) that parameterize the blocks in the model. Model arguments provide different parameter values to each instance of a referenced model. In this model, the arguments are defined as parameter structures to reduce the number of arguments being passed to the referenced model.

The process for defining the model arguments is as follows:

1. Open the referenced model (sldemo_mdlref_counter_datamngt).

2. Define MATLAB structures to parameterize the referenced model.

CounterParams.Increment  = int8(1);
CounterParams.LowerLimit = int8(-10);
CounterParams.UpperLimit = int8(10);

CounterICs.Count         = int8(0);
CounterICs.OverflowState = SlDemoRangeCheck.InRange;

NOTE: This example uses an 8-bit integer for the counter, so the numeric fields also use 8-bit integers.

3. Use the MATLAB structures as model arguments.

Model arguments are defined as variables in the model workspace of a referenced model. You can initialize the model workspace from a number of different data sources. This example uses a MATLAB script (sldemo_mdlref_counter_datamngt_wsdata) to create the parameter structures for the model arguments. Using a MATLAB script makes it easy to create and modify the parameter structure outside the model. It also facilitates incremental changes, version control, and data differencing.

Set the data source in the dialog of the model workspace.

Alternatively, you can set the data source from the MATLAB command line:

hWS = get_param('sldemo_mdlref_counter_datamngt', 'ModelWorkspace');
hWS.DataSource = 'MATLAB File';
hWS.FileName = 'sldemo_mdlref_counter_datamngt_wsdata.m';
hWS.reload;

You can use the Model Explorer to view and edit the contents of the model workspace.

Select the Argument checkbox to configure variables as model arguments.

Alternatively, you can set the list of model arguments from the MATLAB command line:

set_param('sldemo_mdlref_counter_datamngt', ...
          'ParameterArgumentNames', 'CounterParams,CounterICs');

4. Explore how the referenced model uses the model arguments.

A number of blocks in the referenced model use the model arguments:

You can use Simulink.findVars to discover the blocks that use the model argument.

paramInfo = Simulink.findVars('sldemo_mdlref_counter_datamngt', ...
                              'Name', 'CounterParams');
icInfo    = Simulink.findVars('sldemo_mdlref_counter_datamngt', ...
                              'SearchMethod', 'cached', ...
                              'Name',         'CounterICs');
paramBlks = paramInfo.UsedByBlocks
icBlks    = icInfo.UsedByBlocks

paramBlks =

  3x1 cell array

    {'sldemo_mdlref_counter_datamngt/Increment'                  }
    {'sldemo_mdlref_counter_datamngt/Range Check/Detect Overflow'}
    {'sldemo_mdlref_counter_datamngt/Range Check/Saturate Count' }


icBlks =

  3x1 cell array

    {'sldemo_mdlref_counter_datamngt/Initial Count' }
    {'sldemo_mdlref_counter_datamngt/Previous Count'}
    {'sldemo_mdlref_counter_datamngt/outputs'       }

NOTE: Once you have compiled the model, you can use Simulink.findVars to retrieve the 'cached' variable usage information.

Referenced Model: Defining the Shape of the Output Bus Signal

The referenced model produces two results and packages them into a bus signal:

To define the bus type for the root output of the referenced model:

1. Use the Bus Editor to define the bus object (OutputType).

Alternatively, you can create the bus object at the MATLAB command line:

OutputType = Simulink.Bus;
OutputType.Elements = Simulink.BusElement;
OutputType.Elements(1).Name = 'Count';
OutputType.Elements(1).DataType = 'int8';
OutputType.Elements(2) = Simulink.BusElement;
OutputType.Elements(2).Name = 'OverflowState';
OutputType.Elements(2).DataType = 'Enum: SlDemoRangeCheck';

2. Configure the root outport of the referenced model to output a nonvirtual bus signal based on this bus object (OutputType).

Alternatively, you can set the block parameters from the MATLAB command line:

set_param('sldemo_mdlref_counter_datamngt/outputs', ...
          'UseBusObject',      'on', ...
          'BusObject',         'OutputType', ...
          'BusOutputAsStruct', 'on');

3. Create a MATLAB script to recreate this bus object.

This example uses a MATLAB script (sldemo_mdlref_counter_datamngt_types) to recreate the bus objects used by the referenced model.

You can create this MATLAB script manually, or use Simulink.saveVars:

Simulink.saveVars('sldemo_mdlref_counter_datamngt_types.m', 'OutputType');

4. Configure the referenced model to create the global data that it uses.

It is common practice for a model to recreate all of the global data and types that it uses. In this example, the referenced model's PreLoadFcn is used to execute the MATLAB script that creates the bus objects for the referenced model. The referenced model does not use any other global variables.

To set the PreLoadFcn model callback, open the Model Properties dialog box from the Modeling tab by clicking the Model Settings button arrow, and then under Top Model, selecting Model Properties. On the Callbacks tab of the Model Properties dialog box, in the Model pre-load function box, call the sldemo_mdlref_counter_datamngt_types script, and then click Apply.

Alternatively, set the callback from the MATLAB command line.

set_param('sldemo_mdlref_counter_datamngt', ...
          'PreLoadFcn', 'sldemo_mdlref_counter_datamngt_types');

5. Explore how the bus type is used in the model hierarchy.

This bus type forms part of the interface for the referenced model and is referred to by blocks in both the top and referenced models. Use find_mdlrefs and Simulink.findVars to find out about all the places where this bus object is used in the model reference hierarchy.

models = find_mdlrefs('sldemo_mdlref_datamngt');
open_system(models);
varInfo = Simulink.findVars(models, 'Name', 'OutputType');
blks = vars.UsedByBlocks

models =

  2x1 cell array

    {'sldemo_mdlref_counter_datamngt'}
    {'sldemo_mdlref_datamngt'        }


blks =

  7x1 cell array

    {'sldemo_mdlref_counter_datamngt/Initial Count'          }
    {'sldemo_mdlref_counter_datamngt/Previous Count'         }
    {'sldemo_mdlref_counter_datamngt/Range Check/Bus Creator'}
    {'sldemo_mdlref_counter_datamngt/outputs'                }
    {'sldemo_mdlref_datamngt/Counter1'                       }
    {'sldemo_mdlref_datamngt/Counter2'                       }
    {'sldemo_mdlref_datamngt/Counter3'                       }

Referenced Model: Setting Initial Value for Bus Output

In general, the initial values for bus signals and states can be specified as '0', in which case all of the elements of the bus will be initialized to zero (or the relevant ground value). However, in certain cases, it is desirable to specify nonzero initial values for bus signals and states. In this example, the initial condition of the counter is tunable, so the initial value of output signal must be set consistently.

To specify the initial value of the output of the referenced model:

1. Create a parameter structure that is compatible with the bus signal that you want to initialize. One of the model arguments (CounterICs) is a parameter structure that matches the shape of the output signal. This structure was defined as follows:

CounterICs.Count         = int8(0);
CounterICs.OverflowState = SlDemoRangeCheck.InRange;

Alternatively, you could have used the static method, Simulink.Bus.createMATLABStruct to create the structure using the ground value of the OutputType.

CounterICs = Simulink.Bus.createMATLABStruct('OutputType');

2. Open the dialog box for the root outport and enter the name of the structure (CounterICs) into the Initial output field.

Alternatively, you can set the block parameters from the MATLAB command line:

set_param('sldemo_mdlref_counter_datamngt/outputs', ...
         'InitialOutput', 'CounterICs');

NOTE: In general, when initializing a bus signal or state, the parameter structure does not need to match the bus type exactly, but its fields must be a subset of the elements in the bus object and the attributes of these fields must match the elements in the bus object.

Referenced Model: Masking Referenced Model

It is often useful to mask the referenced model to customize the user interface. You can create a mask on a model and then reference it from a Model block. To mask a model, click Create Model Mask in the model workspace section of Model Explorer for the model.

When you reference a masked model from a Model block, a mask is generated automatically on the Model block. For more information, see the topic Create and Reference a Masked Model.

Top Model: Parameterizing Model Blocks

The referenced model is configured to accept a structure for its model argument. This example calls each instance of the referenced model with different parameter values.

1. Open the top model (sldemo_mdlref_datamngt).

2. Create parameter structures with the same "shape" as the model arguments defined in the referenced model.

Param1.Increment  = int8(1);
Param1.LowerLimit = int8(-20);
Param1.UpperLimit = int8(20);

Param2 = Param1;
Param2.Increment  = int8(2);

IC1.Count         = int8(0);
IC1.OverflowState = SlDemoRangeCheck.InRange;

IC2 = IC1;
IC2.Count = int8(-10);

3. Set the mask parameters on the masked Model blocks to use these parameters.

For example, the parameters of the first Model block ('sldemo_mdlref_datamngt/Counter1') are set as:

Alternatively, you can set the block parameters from the MATLAB command line:

set_param('sldemo_mdlref_datamngt/Counter1', ...
          'CounterParams', 'Param1', ...
          'CounterICs',    'IC1');

3. Create a MATLAB script to recreate these workspace variables.

This example uses a MATLAB script (sldemo_mdlref_datamngt_wsdata) to recreate these workspace variables.

You can create this MATLAB script manually, or use Simulink.saveVars:

Simulink.saveVars('sldemo_mdlref_datamngt_wsdata.m', 'Param*');

4. Configure the top model to create all global variables.

It is good practice for the top model to recreate all of the global data and types used throughout the model reference hierarchy. This configuration avoids loading referenced models when simulating the top model.

In this example, the top model's PreLoadFcn is used to execute the MATLAB scripts that create the workspace data for the top model and the types for the referenced model. The referenced model does not use any global data.

set_param('sldemo_mdlref_datamngt', ...
         'PreLoadFcn', ['sldemo_mdlref_datamngt_wsdata; ', ...
                        'sldemo_mdlref_counter_datamngt_types']);

Top Model: Scheduling Calls to the Referenced Model

As mentioned earlier, the referenced model implements a limited counter algorithm with two trigger inputs. The algorithm detects "rising edges" in the trigger inputs and reacts as follows:

In this example, you drive all three instances of the counter algorithm with the same trigger inputs, generated by the Stimulus subsystem. The period and sample time of the trigger inputs are defined by the values entered for the mask parameters of the Stimulus subsystem:

NOTE: Reset and increment periods must be at least 2 times the sample time.

Simulating the Top Model

1. Save or close the referenced model.

You need to save the referenced model before it can be used by the top model. Alternatively, you can close the referenced model and use the original version of the model provided with this example.

2. Simulate the top model by selecting Simulation > Run.

sim('sldemo_mdlref_datamngt')

NOTE: Simulating the top model automatically generates a simulation target for the referenced model.

3. Review the simulation results as displayed by the Scope blocks.

Generating Code for the Top Model (Requires Simulink Coder)

To generate code and build a standalone executable for the top model, open the top model, and then press Ctrl+B.

Alternatively, use rtwbuild at the MATLAB command line.

rtwbuild('sldemo_mdlref_datamngt')

NOTE: When you generate code for the top model, Simulink Coder automatically generates code for the referenced model. Use Embedded Coder™ to generate embedded code for production deployment.

Controlling the Type Name for the Model Argument

By default, Simulink Coder generates an automatic name for the type of the parameter structures. This name is unique and deterministic, but not easy to recognize. You can control this type name by using Simulink.Parameter objects to define the parameter structures and Simulink.Bus objects to define the type.

1. Define the bus type for the model argument by using the Bus Editor or by typing the following commands at the MATLAB command line:

CounterParamType = Simulink.Bus;
CounterParamType.Elements = Simulink.BusElement;
CounterParamType.Elements(1).Name     = 'Increment';
CounterParamType.Elements(1).DataType = 'int8';
CounterParamType.Elements(2) = Simulink.BusElement;
CounterParamType.Elements(2).Name     = 'LowerLimit';
CounterParamType.Elements(2).DataType = 'int8';
CounterParamType.Elements(3) = Simulink.BusElement;
CounterParamType.Elements(3).Name     = 'UpperLimit';
CounterParamType.Elements(3).DataType = 'int8';

2. Modify the existing MATLAB script to recreate this bus object.

This bus type must be defined in the base workspace, along with the bus type for the referenced model's output. You can manually modify the MATLAB script that defines the types for the referenced model or use Simulink.saveVars to append this variable to the existing MATLAB script.

Simulink.saveVars('sldemo_mdlref_counter_datamngt_types.m', 'CounterParamType', '-append');

3. Create Simulink.Parameters to define the model arguments.

To associate the bus types with the model arguments, you must use Simulink.Parameter objects in place of the MATLAB structures in the model workspace of the referenced model.

CounterParams = Simulink.Parameter;
CounterParams.Value.Increment  = 1;
CounterParams.Value.LowerLimit = -10;
CounterParams.Value.UpperLimit = 10;
CounterParams.DataType = 'Bus: CounterParamType';

CounterICs = Simulink.Parameter;
CounterICs.Value.Count         = 0;
CounterICs.Value.OverflowState = SlDemoRangeCheck.InRange;
CounterICs.DataType = 'Bus: OutputType';

NOTE: When you set the data type of a parameter object to be a bus, you can use double-precision values for the numeric fields of the parameter structure. Simulink converts these double-precision values to the appropriate numeric data type during model compilation.

You can manually modify the MATLAB script that defines the model argument or use Simulink.saveVars to update the existing MATLAB script.

Simulink.saveVars('sldemo_mdlref_counter_datamngt_wsdata.m', '-update');

NOTE: When you use the -update option, you do not need to specify the variables to be written out because Simulink.saveVars writes out only the values of variables already defined in the MATLAB script.

5. Reload the contents of the model workspace.

You have modified the data source that is used by the referenced model's workspace. If this model is open, you must reload the contents of the model workspace.

You can reload by using the model workspace's dialog box or at the MATLAB command line.

hWS = get_param('sldemo_mdlref_counter_datamngt', 'ModelWorkspace');
hWS.reload;

Controlling Representation of Parameters in the Top Model

You can use a similar approach to convert the parameter structures in the top model to use Simulink.Parameter objects. This configuration gives you control over the type name and enables you to control how these parameters appear in the generated code.

1. Create Simulink.Parameter objects with the ExportedGlobal storage class to define the parameter structures for the top model as a exported, tunable variables in the generated code.

Param1 = Simulink.Parameter;
Param1.Value.Increment  = 1;
Param1.Value.LowerLimit = -10;
Param1.Value.UpperLimit = 10;
Param1.DataType = 'Bus: CounterParamType';
Param1.CoderInfo.StorageClass = 'ExportedGlobal';

Param2 = Param1.copy;
Param2.Value.Increment = 2;

IC1 = Simulink.Parameter;
IC1.Value.Count = 0;
IC1.Value.OverflowState = SlDemoRangeCheck.InRange;
IC1.DataType = 'Bus: OutputType';
IC1.CoderInfo.StorageClass = 'ExportedGlobal';

IC2 = IC1.copy;
IC2.Value.Count = -10;

2. Modify the existing MATLAB script to recreate these workspace variables.

You can manually modify the MATLAB script that defines the parameters for the top model or use Simulink.saveVars to update the existing MATLAB script.

Simulink.saveVars('sldemo_mdlref_datamngt_wsdata.m', '-update');

NOTE: When you use the -update option, you do not need to specify the variables to be written out because Simulink.saveVars writes out only the values of variables already defined in the MATLAB script.

Regenerate Code for the Top Model (Requires Simulink Coder)

To generate code and build a standalone executable for the top model, open the top model, and then press Ctrl+B.

Alternatively, use rtwbuild at the MATLAB command line.

rtwbuild('sldemo_mdlref_datamngt')