How to use multiple Simulink code generated models in a project?

I have an embedded project I am working on where I want to split my simulink model into multiple sub-models and generate code for each separately. If I generate code for each model separately, then the rtU and rtY structures end up get redefined when I include multiple code generated libraries in my project.

Is there a way to configure code generation to avoid this issue?

I have tried changing model settings>code generation>code interface>code interface packaging to "reusable function" and that resolves my concerns with rtU and rtY, but then RT_MODEL is redefined in each generated header file (also I do not know how I am supposed to use the argument "RT_MODEL *const rtM" for the initialization and step functions.

Any inputs/guidance would be great!

Answer 1

[Jacob Mathew]

Hey Hareesh,

As you have mentioned "...where I want to split my simulink model into multiple sub-models and generate code for each separately." it seems to me like you are on track to do Component Based Modeling. You can refer to the following guide to get started on what it is, its advantages and how to approach it based on your requirements:

https://www.mathworks.com/help/simulink/ug/component-based-modeling-guidelines.html

You have already mentioned that you are setting the interface packaging to Reusable Function but just to double check, you can refer to the following to differentiate between Reusable and Non Resuable Functions:

https://www.mathworks.com/help/rtw/ref/codeinterfacepackaging.html

Finally, specific to RT_Model getting redefined, you can refer to the following MATLAB Answers Post to troubleshoot:

https://www.mathworks.com/matlabcentral/answers/1900055-how-can-i-remove-real-time-structures-from-code-generated-using-embedded-coder

[Hareesh]

Hello Jacob,

Thanks for sharing resources for me to look into, I'll try it out and get back at the earliest!

[Hareesh]

Hello Jacob,

I'm a bit confused about how to approach this, as when I set interface packaging to Reusable Function, the embedded coder quickstart auto-selects generating code for multiple instances (which I believe is causing me issues now, by generating tag_RTM RT_MODEL). Is there a way to set reusable function but generate for a single instance? If I try that, the embedded coder quickstart forces me to change interface packaging from Reusable Function to Nonreusable Function.

How can I approach this?

Alternatively, is there any way to have rtU/extU and rtY/extY generated with different names? That way I can use Nonreusable Functions. My primary intention is to have multiple different models in the same project, I do not intend on having multiple instances of the same model

[Jacob Mathew]

Hey Hareesh,

Component based Modeling deals with the identical idea of having multiple different models in your project. The idea is to have certain sub components or sub systems that you can change on the fly as you adapt your design. Using Referenced Models is part of the component based modeling approach and it avoids naming conflicts by encapsulating code for each referenced model using seperate namespaces. You can refer to the following documentaion to guide you:

https://www.mathworks.com/help/releases/R2025a/rtw/referenced-models-1.html

Alternatively, if you want to manually customise the model variable names and namespaces, you can follow the below documentation on Customising Generated Identifiers:

https://www.mathworks.com/help/releases/R2025a/ecoder/identifiers.html

[Hareesh]

Hello Jacob,

Thank you for clarifying, I was a bit lost on how it applies until now. Much appreciation!