How to improve the simulation speed of a system slowed by the addition of a Fuzzy System

The simulation became slow after adding fuzzy system. How can I fix this without effect my system please ?

[Daniel]

@Ahmed Alshamrany, could you provide a little more detail on the fuzzy system?

Does it use fixed-point data types, for instance?

Does it set a particular step size in your model that might be faster than the model is running otherwise?

Is the fuzzy system itself complicated?

Is the system a separate file, such as a referenced model or a referenced subsystem?

A screenshot, model files, or indicating what blocks you removed and added might help as well.

Answer 1

[Sam Chak]

Hi @Ahmed Alshamrany

If your fuzzy system is of the Mamdani type (as is the case for 99% of beginner users), and it has many inputs, membership functions, and rules, this can significantly slow down the simulation, especially when running in Simulink. Simplifying the fuzzy system by reducing the number of membership functions and rules, or even converting it to a Sugeno-type fuzzy system generally helps.

In the past, users of fuzzy systems in Simulink were advised to approximate the fuzzy control surfaces using 2-D Lookup Table or n-D Lookup Table blocks to improve execution speed. Since R2024b, the Fuzzy PID Controller block has been introduced, which is capable of converting a fuzzy system into a lookup table with precomputed outputs.

Example 1

Example 2