How to use LSTM based trained Recurrent Neural Network in Simulink?

9 views (last 30 days)
I have trained recurrent neural network and using "Predict" command it gives the required output for the trained input set but i am facing error on using it in "user defined function" in simulink, importing net.mat file via "from work space" block or "load ('net.mat');" command.
ERROR:
Unsupported input format for From Workspace block 'SHAPFpqFinal/RNN/From Workspace'. Available formats are double non-complex matrix, a structure with or without time, or a structure with MATLAB timeseries as leaf nodes. All formats require the data to be finite (not Inf or NaN).
Component:Simulink | Category:Block error
Kindly do guide me in this regard. Thanks in advance....

Answers (1)

Arkadiy Turevskiy
Arkadiy Turevskiy on 18 Oct 2021
Hi,
We added a library of blocks to integrate deep learning networks into Simulink models starting with 20b. Support for LSTM and other recurrent networks was added in 21a.
To bring trained LSTM into a Simulink model, please use Stateful Classify block. In the block dialog specify .MAT file with your LSTM network.
See this example to help you.
All deep learning blocks are decribed here.
HTH.
Arkadiy

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!