Hello Siemens Forum Community,

I have been exploring the capabilities of the S7-1500 TM NPU module and am impressed by its integration of AI algorithms with PLC logic. While its applications in visual quality inspection are well known, I am curious about its potential for other types of projects, specifically water quality control.

My idea is to use various sensors to collect data (such as pH, turbidity, conductivity, etc.) and then employ a trained neural network to analyze this data in real time, identifying anomalies and classifying water quality levels. Has anyone implemented or considered a similar project using this module? Are there any specific challenges or recommendations you would suggest for integrating sensor data into the AI model on this platform?

Thank you in advance for your insights and any shared experiences.

Best regards,
Jon.

The TM NPU does not connect directly to sensors. Instead, you can use the S7-1500 PLC to collect sensor data such as pH, turbidity, conductivity, and more. This data is then sent to the NPU for processing with the neural network.

The NPU supports various input formats, including lists, arrays, dictionaries, and ExternalBuffer objects. For example, if your model expects a vector of six values, you can pass your sensor readings as a list like:

[input_ph, input_turbidity, input_conductivity, ..., input_n]

Just make sure that the shape of the data matches what the model was trained on.

You can send data from the PLC to the NPU and also return the results back to the PLC. This two-way communication makes the setup highly flexible and suitable for a wide range of applications.

For more technical details on the supported input formats, I recommend the official Siemens document TM NPU MicroPython:

https://support.industry.siemens.com/cs/de/en/view/109824411

This document clearly explains how to structure input data for the neural network.