Drives & Controls Magazine May 2023

n TECHNOLOGY BECKHOFF HAS DEVELOPED an AI-supported chatbot add-on for its TwinCat XAE engineering environment that will help users to automate tasks such as creating and optimising PLC function block coding, and generating documents. It can also complete missing code from function blocks, or restructure code for better readability. Beckhoff predicts that it will “open up a whole new world of chatbot possibilities for control engineering”. The TwinCAT Chat Client makes it possible to use Large Language Models (LLMs) – such as OpenAI’s ChatGPT – to help develop TwinCat projects for increased productivity in control programming. It also opens up optimisation potential for direct support. TwinCat product manager Sven Goldstein explains: “While we have been playing around with ChatGPT in recent months, we found out that it has a very nice additional skill – it can also develop source code”. He gives the example of asking ChatGPT to generate the PLC source code for a quick-sort algorithm. This can then be draggedand-dropped to the PLC. LLMs, based on neural networks, are trained using large quantities of text. They can be used for a variety of tasks – including as the basis for chatbots or language translation tools. Beckhoff’s TwinCat Chat connects to the host cloud of an LLM – such as Microsoft Azure, in the case of ChatGPT. It provides a user interface, and communicates with the PLC development environment via an automation interface. This is available via a chat window in Visual Studio, optimising the LLM functions for TwinCat 3 users. It is supplemented by “extensive”TwinCat-specific content. Chatbot add-on automates controller programming Goldstein: a very nice additional skill A CALIFORNIAN INDUSTRIAL software developer has announced a generative AI tool that helps to train and test predictive maintenance models at the industrial edge. EoT (Embassy of Things) says that its patentpending Twin Talk GPT system will help users to improve their operational efficiency and decision-making. The technology can simulate the real-time industrial events and operational data that are used to train machine-learning algorithms that help to optimise equipment helath and perfomance. EoT CEO Matt Oberdorfer describes the product as “a fundamental game-changer for the industrial sector, enabling companies to leverage the power of generative AI to improve predictive maintenance systems with the help of GPTs (generative pre-trained transformers).” Twin Talk GPT is due to be released in November 2023. GPT AI tool will improve predictive maintenance