As an industrial software company serving more than 45,000 users worldwide, the daily use of Artificial Intelligence (AI) and Machine Learning (ML) is not a topic we shy away from. From the development process through to production systems that equip customers with intuitive ways to change the way they work, there’s a place for AI.
AI makes decisions based on input data and patterns, using underlying models like neural networks to drive insights and take actions accordingly. The operation of industrial facilities is underpinned by high safety standards. Thus, when AI comes into play, it is a necessity to also consider the accountabilities, responsibilities, and quality that follows the decision-making process. That leaves us with a very important question: How can we create technology that drives autonomy yet empowers businesses to remain operational within the boundaries of safety?
AI loves data! But where does industrial data come from?
- IoT sensors
- 3rd party libraries and applications
- The enterprise IT landscape including documents, databases, operations data, and more
The Industrial Work Surface is agnostic, ingesting data from various sources and delivering it in the context of industrial business process through an intuitive user interface that reduces complexity. That gives you a virtual environment reflective of any asset, its systems, and its behaviours presented in a way that anyone can understand, as part of the way work is done.
Build trust by selecting the right stakeholders
Stakeholders will vary between use cases, but the mental model stays the same. Any operation, with or without AI, requires stakeholders to be included early in the process. The addition of AI places an even bigger focus on including end-users like engineers who will be using the technology onsite. After all, the decisions that end-users (like engineers and technicians) make have real consequences for the day-to-day operations of assets. And don’t just include them – deliberately institute a sense of ownership over the resulting processes, all the way from model development to training and commissioning. Involving the right people from the start builds trust, giving end-users the confidence they need to turn AI-backed insights and recommendations into actions that drive value.
Change the way you work from start to scale
At Kongsberg Digital, we have for years worked on implementing Hybrid Machine Learning – a practice that combines Machine Learning with synthetic data from 1st principles physics-based simulation models. We have seen first-hand how this way of working can provide more reliable insights for predictive analysis at scale. Data is contextualised and simulation is integrated so that the AI model training process can begin. When trust in the use case has been built, it’s time to scale. Scale in the sense of volume of deployed models, but more importantly scale in the delivery of resulting datasets integrated into the decision-making business process. In other words – changing the way we work. Changing the way we work to include the power of hybrid ML and AI is what makes this technology useful for operations.
You can have all the AI and ML in the world; if it remains contained within the data science department, the value is lost, or at best limited. The only way to unlock change is to infuse intelligence into the industrial business processes.
Understand the process
Transparency is the foundation for functional AI. It’s not a one-button solution – we need to ensure that AI-driven decisions are safe and promote welfare in the context of their intended uses. Throughout the process of building out digital workflows, the appropriate stakeholders want to understand what’s going on. What are the rules, frameworks, and decisions that led to the use case in question? For industrial settings, using 1st principle physics models as training foundations for algorithms means being able to understand and explain how AI has been trained. We can:
- See what data went into the training or decision
- Show the decision-making logic used when training
- Show the course of action taken
- Run and re-run a high volume of tests
- Work to improve the quality of training to continuously avoid biases
Safety is always top of mind for the industry. Collectively, the industry wants to ensure that safe ways of working are the golden standard. Algorithms should always make the choice that best preserves safety in operations and guards against high levels of risk. With dynamic simulation, operators can stay inside the boundaries of safety to maintain their license to operate and more importantly, ensure safer operations for people and the environment.
Keep a human in the look
Kongsberg Digital believes that AI and Machine Learning help people do better, and it’s essential to keep us humans in the loop. The hard part about keeping humans at the center of technology is changing the way people work today. One way to make digital transformation successful is to create technology that is intuitive, user-friendly, and empathetic to users with varying levels of digital skills.
Use case in point
Here’s an example of how our Industrial Work Surface optimizes the entire value chain to improve energy efficiency at facilities that are major energy off-takers: Without digitalisation, the energy nomination process was a resource-intensive manual process. Engineers needed to trend historical data and maintenance records, integrate facility data from planned throughput, and use bespoke offline models and assumptions to predict the facility energy consumption requirements. Offtake nominations were then generated based on these activities.
But what if we could simplify the generation of energy nominations in line with actual plant needs based on an integrated production profile that considered a larger dataset – thus removing the manual data processing step and reducing reliance on ad-hoc interdisciplinary coordination between teams?
And that’s exactly what the Industrial Work Surface does: generate energy nominations based on actual plant needs, easier and faster than ever before. A critical component of this nomination workflow is the combination of AI for the nomination itself, but also the visualization of collective information in one screen that puts the workflow in context alongside historical data and KPIs.
This use case was developed in a collaborative way with the business, bringing in the traceability that stakeholders expect from digital initiatives. Stakeholders for the development process were able to understand the algorithms used for nominations and the choices they made: full explainability. From the outset, pain points were well understood, and the use case was well defined – that’s the responsibility coming into play. It was possible to build trust to the level where technology is used on a regular basis to predict energy consumption for the facility while keeping the human in the loop. And the quality of energy nominations continues to improve in comparison to just a few years ago.
The primary benefits of the energy nomination workflow:
- More responsible energy usage.
- Engineers are freed up to focus on other work.
- From 4+ systems to everything in one environment.
- Penalties associated with selling overcapacity or buying back excess energy are avoided.
- Reduced time spent on energy nominations through a workflow that is close to fully automated.
And that’s just one use case. Technology is a game changer for operational efficiency, presenting new opportunities for businesses to work smarter, safer, and greener. Our verdict? To AI, as long as it continues to be traceable, explainable, accountable, responsible – and keeps the human in the loop.
GET IN TOUCH WITH OUR EXPERTS
Want to know more about how the Industrial Work Surface helps reduce up to 70% of operational inefficiencies within just three months?