.SMArtInt.Blocks

Information

Blocks

This package provides several blocks which can be used to include different types of neural networks within Modelica models. These blocks can be included in models, but their in- and outputs are still generic arrays. The user has to fill the arrays in the same manner as they are used during training in python.

Steps to include a model:

1. Create and train a model in, e.g., TensorFlow
2. Export a trained model as TfLite or ONNX model
3. Place the corresponding block in your model
4. Parametrize the block (provide path, number of in- and outputs, etc.)
5. Connect the in- and outputs of the block. The arrays have the same structure as those in the AI model (e.g. TensorFlow): the inputs have to be connected in the same manner as they are used in the neural network during training.

The examples PipeLocalHeatTransfer_tflite and TF_PI_Stateful_tflite use this approach.

Contents

Name	Description
EvaluateGenericNeuralNetwork
EvaluateSimpleFeedForwardNeuralNetwork
EvaluateRecurrentNeuralNet
EvaluateStatefulRecurrentNeuralNet