My idea is to have/create my teams own chatbot to respond to issues with equipment/ HUD screens when maintenance puts in issues. The information that is provided comes from the chat bot is what my team feeds it. If the chatbox seems that the issues is major or does not have a result. Then is messages the person on call from my team. Is this doable within ignition, if so could I be pointed in the direction or would I need to use outside resources?
You likely need to spend more time actually drawing out what you expect from this because, not trying to be mean, but right now this sounds like the typical half baked 'just slap AI on it' answer that is growing in businesses. Ignoring the complexity of the topic there is even security concerns such as allowing your companies production data to be scrapped by external LLMs unless you're planning on hosting your own internal one.
There is an entire field of predictive analytics that builds statistical models by feeding training data into machine learning algorithms for anomaly detection but that's far above a 1 person side project. At best you could feed your tag names, descriptions, equipment IDs, etc into a local model so that operators can use plain text prompts to pull up trends to look at sections of a plant.
5 Likes
Appreciate the feedback. I have been brain storming and was just reaching out to see if anyone has or is doing something similar. That way I could bounce ideas and see what learning pains they went through.