Method and system for generating an artificial intelligence model

A computer implemented method for generating an artificial intelligence (AI) model for an engineering system, the computer implemented method comprising:

• determining an AI function in an engineering framework system, wherein the AI function is a specific function block comprising a definition of operations of algorithms;
• defining an inference path for the generation of the AI model starting from the determined AI function by a computation graph, wherein the inference path for reaching the AI model is described by operation steps such as convolution or by providing function blocks that correspond to parameterizable layers or by a single function block representing a complete neural network such as a convolutional neural network;
• converting the AI function and the inference path into a processing format;
• sending and exporting the converted AI function and the converted inference path to an extraction and extension module of an AI workbench module, wherein the extraction and extension module of the AI workbench module has a functionality to extract and extend the computation graph architecture of the previously defined inference path by operations of algorithms for model training;
• deploying the converted AI function on a processing module of the engineering system, another processing device, or the processing module and the other processing device;
• transmitting the extended computation graph of the inference path from the extraction and extension module to an AI framework module;
• communicating of a communication adapter with the processing module continuously by using a supporting communication protocol for receiving training data as input for the AI function and forwarding the training data to the AI framework module;
• and
• transferring learned parameters of the AI model from an API interface of the AI framework module to the communication adapter for updating the AI model.

A system for generating an artificial intelligence (AI) model, the system comprising:

• an engineering framework system configured to:
• determine an AI function, the AI function being a specific function block comprising a definition of operations of algorithms;
• define a inference path for generation of the AI model starting from the determined AI function by a computation graph, wherein the inference path for reaching the AI model is described by operation steps such as convolution or by providing function blocks that correspond to parameterizable layers or by a single function block representing a complete neural network such as a convolutional neural network;
• convert the AI function and the inference path into a processing format, and export the converted AI function and the converted inference path to an extraction and extension module of an AI workbench module, wherein the extraction and extension module of the AI workbench module has the functionality to extract and extend the computation graph architecture of the previously defined inference path by operations of algorithms for model training and is configured to transmit the extended computation graph of the inference path to an AI framework module;
• a processing module of the engineering framework system, another processing device, or the processing module and the other processing device configured to deploy the converted AI function;
• and
• a communication adapter configured to communicate with the processing module, the other processing device, or the processing module and the other processing device continuously using a supporting communication protocol for receiving training data as input for the AI function and forwarding the training data to the AI framework module,
• wherein the AI framework module is configured to transfer learned parameters of the AI model using an API interface of the AI framework module to the communication adapter for updating the AI model.