Artificial Intelligence (AI) is not just a buzzword; it’s a dynamic field with tangible industrial applications that are already starting to reshape certain industries. In this article we explore the real-world impact of Industrial AI, moving beyond the initial hype.
Transformative objectives
Industrial AI promises to uncover new insights from data and enhance human decision-making. It will empower people to take actions that were previously unattainable using conventional decision-making methods. AI can augment and improve human decision-making by providing better contextual information and guidance. But AI will increasingly start to make decisions for humans – particularly when speed and accuracy are required. Improvements in autonomous vehicle driving systems give us several clues as to what types of decisions can be made by computers, and what types of decisions still need to be made by humans.
The evolution of Industrial AI
Industrial AI is not new. The traditional automation vendors have been positioning themselves for more than a decade now to take advantage of development in fields like expert systems, neural networks and decision support systems. Through mergers, acquisitions and R&D; spend, these companies are presently working to retain their leadership position. The consolidation by acquisition of new innovative technology vendors into these giants is set to alter the control and automation landscape significantly over the next few years.
Industrial AI has made significant strides relatively recently, thanks to factors like the proliferation of cloud computing, affordability of hardware and sensors, and the rise in the number and variety of IoT devices. Big tech like Azure, AWS and Google Cloud have developed robust cloud-based data streaming architectures to collect this data and analyse it with AI technology.
These developments have opened up the possibility of new AI agents running on scalable, powerful hardware capable of deep analysis of industrial data. These use techniques such as predictive forecasting, discovery of correlations, and detection of abnormal patterns. The same concept can be applied to analysing historical and real-time information originating from scada, DCS and PLC systems.
The next logical step in this evolutionary path will be connecting these AI agents to the vast quantities of unstructured information in a business, such as historical documents, email messages and meeting transcripts.
AI beyond generative transformers
The hype around generative transformers has probably muddied the water somewhat. ChatGPT has brought generative transformers into the headlines, and most people have by now experienced some form of AI support in their normal working lives. However Industrial AI encompasses applications that go beyond generative transformers. AI investments are being made in areas such as chemical process modelling, plant simulations, natural language processing, image recognition and decision support systems.
Mainstream adoption and expanded possibilities
Large language models have already entered the mainstream, with tech giants rapidly integrating them into consumer applications. For example, Microsoft Copilot is an AI companion that will span all of Microsoft’s productivity applications, including Microsoft 365, Windows, Edge and Bing. Google Duet has a similar objective to enhance productivity and effectiveness in the workplace. These innovations will spill over into the control room. An example is the analysis of control room logbook data in combination with real-time plant data, together with the information in the company information systems. This analysis can be used for fault finding, forecasting, planning and efficiency improvements.
Let us visualise an example. The control room operator on shift speaks into the AI-enabled logsheet application: “The tanker raw material delivery was late and it seems they rushed to meet our deadline. The delivery note indicates below-spec material. I have submitted a sample for analysis but this will take 24 hours. The tanker is waiting for a decision to offload or otherwise to reject the delivery and fast track an emergency replacement order. The plant urgently needs the material to run over the full weekend”.
The future AI agent can look at the performance history of the supplier, assess the current status of the plant, and assess the impact of a lower feedstock quality on production. The response from the AI agent to the control room operator might be:
“Based on the available information, this supplier often delivers off-spec materials. The latest tanker load is unsuitable for use without further treatment. However, it is likely that a rushed emergency order will have the same quality problem. You currently possess a sufficient inventory of on-spec raw materials in storage, enabling you to blend away the new stock to prevent wastage, and allowing the plant to still meet the weekend production target. I suggest you take delivery of the new stock into the empty raw material tank number three. Concurrently, I will communicate with procurement to initiate the necessary penalty clause and secure a full credit. To maintain production quality, I recommend blending the off-spec materials from the spare tank into the plant gradually over a period of at least 36 hours”.
Evolving applications
Industrial AI’s potential expands into new horizons, including energy optimisation, dynamic process optimisation, supply chain and inventory management and real-time logistics management. These applications all promise efficiency gains and cost savings. For example, AI-based energy optimisation systems can automatically match energy consumption to the current constraints, reducing energy costs and environmental impact. These systems will be differentiated from the current generation of optimisation algorithms by their ability to learn and adapt constantly to changing factors in the environment. Smart building technology innovations are likely to lead the way in this area, with the technology eventually being applied to industrial-scale facilities.
Of course, having a new hammer will make everything look like a nail, and random application of advanced technology to a problem not yet defined is always unwise. Implementing AI successfully will require more than just technology. It will demand a holistic approach, involving an ecosystem of specialist consultants, managed service providers, technical experts and business. Collaboration across this ecosystem will be important to unlocking the transformative potential of industrial AI.
Industrial AI is transitioning from a buzzword to a business reality, with several tangible applications in the pipeline. There are already several established vendors in this space with good solutions. Industrial AI offers industries the potential to revolutionise industrial operations in areas from operations through to energy optimisation. However, as we navigate this transformative journey, business objectives should remain at the forefront. The future is indeed exciting for professionals who possess the necessary skills to apply the new generation of AI in their own factories and mines.
About Gavin Halse
Gavin Halse is a chemical process engineer who has been involved in the manufacturing sector since mid-1980. He founded a software business in 1999 which grew to develop specialised applications for mining, energy and process manufacturing in several countries. Gavin is most interested in the effective use of IT in industrial environments and now consults part time to manufacturing and software companies around the effective use of IT to achieve business results.
For more information contact Gavin Halse, Absolute Perspectives,
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version