SAS – Democratising AI in manufacturing

SAS – Democratising AI in manufacturing

Adriaan Van Horenbeek, Manufacturing Lead Advisory Western Europe at SAS, gives us the facts on the importance of digital transformation and analytics i...

SAS is the leader in advanced and predictive analytics. Since 1976, SAS has empowered and inspired customers with the most trusted analytics, innovative software, and services. From machine learning, deep learning, computer vision, and natural language processing (NLP) to forecasting and optimization, their AI and IoT technologies support diverse environments and scale to meet changing business needs. 

Adriaan Van Horenbeek, Manufacturing Lead Advisory Western Europe at SAS, believes that in the manufacturing industry advanced analytics is now essential for businesses to thrive. "Most companies are now aware that digital transformation is here to stay. One of the key pillars of digital transformation involves making smarter use of data, leading to better decision-making. To do that you need advanced analytics." 

Digital transformation is driven by people, processes, and technology. "People will start using the new technology, so they need to develop new skills and we need to guide them. Their roles might shift a little. If you look at operators for example in the production plant, their role will become a bit more strategic as execution becomes automated by algorithms." 

When it comes to the process, the key is how to derive long-term value from advanced analytics with AI and machine learning. "This means finding out how to scale within the organisation, and for that, you need a clear process in place," he says. 

"We typically have a four-step-approach (link on the green words https://www.youtube.com/watch?app=desktop&v=70HT2HzIYRM) to work with the customer. The first one is to identify the business potential, to see where the largest opportunities are in terms of using analytics. The second one is to create lighthouse projects in a few weeks because this will show the organisation what the possibilities and value of using analytics are for them. The third step is to industrialise which means integrating advanced analytics in the day-to-day processes. The fourth step is scaling, which means accelerating digital transformation by expanding to other business units or their business clients." 

Then it’s time to make sure the right technology is in place to support the people and the process. "We talk about the analytics lifecycle: (link the green words here: https://www.sas.com/content/dam/SAS/documents/infographics/2019/en-analytics-life-cycle-110801.pdf) getting the data, cleaning the data, then going into the analytics and modelling capabilities. 

The final part is deployment, translating the insights that come out of an algorithm into day-to-day decisions. It's only here that you actually capture the value." 

“Covering the entire analytics lifecycle is important”, Van Horenbeek says, “because sometimes we see that technology only covers certain parts and is a patchwork of different technologies which makes user adoption more difficult. We can support our clients to cover the entire analytics lifecycle.” 

We speak the language of the clients we're supporting because for example in my team we come from an industry background, not purely with a data science background. It helps us guide our clients through the entire journey.

Subheading – every journey starts with the right data

Manufacturing organisations typically capture large amounts of data, but that doesn’t necessarily mean it is the right data or there could be gaps. This is where IoT comes in, for example, you can easily place wireless sensors on a production line to capture additional data and store it in a cloud environment

"However, while IoT data capture is very nice, this doesn't bring you value. It only brings you value once you apply analytics and start analysing this data and making smarter decisions,'' Van Horenbeek adds. 

A specific area this can benefit is R&D. "We analyse data to identify new products faster, decreasing the go-to-market time. Another example is the smarter use of data in the supply chain. You can look at demand forecasting, using AI and machine learning techniques to better predict the demand, and inventory optimisation - all are related to smarter scheduling of production. 

"In the production plant, we use AI or machine learning to perform process optimisation in terms of yield and the quality of the product - we can identify the key drivers of variability in the quality of the product that we are producing." 

It’s also possible to find sustainable ways of working, like using data to reduce the energy consumption of production lines. "For example, using less raw materials to produce the same end product,” Van Horenbeek says. 

Advanced analytics is also a valuable tool to serve customers in a better way. "In terms of pricing optimisation and customer intelligence, it can capture customers’ requirements and feed back into the R&D process to develop products that are aligned with the real customer needs." 

SAS supports clients with several Industry 4.0 solutions specific to the particular business problems in their industry. "In addition to our core technology, we've built solutions for dedicated business problems to give our clients a headstart in using advanced analytics according to their business context. In manufacturing, we have specific solutions like SAS® Analytics for IoT, SAS® Production Quality Analytics, and SAS® Asset Performance Analytics which is more targeted towards predictive maintenance. We also have SAS® Field Quality Analytics, targeted to analysing the data of a product during its lifetime. " 

SAS® Viya® platform is built to capture raw data from a variety of different sources and provide dashboarding, statistical analysis, data mining, machine learning, and AI. "Our aim is the democratisation of AI, and we provide visual interfaces throughout the entire analytics lifecycle, so you do not need to be an expert in coding to use it." 

"In digital transformation, we talk about open ecosystems and partnerships. Our technology is also built with that in mind. The SAS® Viya® platform is completely open, so you can integrate open source applications like Python. We want to integrate with the existing ecosystem of the customer that is already in place.

SAS partners with organisations that are leaders in key areas - for instance for cloud technology we partner with Microsoft, to enable our clients to define their cloud strategy. With Ericsson Dedicated Networks - Ericsson, we are leveraging IoT and 5G capabilities. 

We also partner with startups to ensure we are continuously harnessing innovation. One example is BlueChem, https://video.sas.com/sharing?videoId=6213102240001 an incubator in the chemical industry based in Belgium; another is Aerospace Valley Aerospace Valley | Le collectif qui rend compétitif (aerospace-valley.com) in France. "We are looking at how we support these companies with our technology in the first phase of their journey going from a startup to a scaleup. Having these dedicated partnerships also enables us to provide end-to-end solutions to our clients."

“In the next few years”, Horenbeek says, “manufacturing businesses need to move from testing new technologies to using them at scale within their organisations. To transform their way of working and putting analytics and AI in action is key. It will give them a significant competitive advantage.

"There is a discrepancy today between companies using data effectively compared to companies that haven't started yet, and this will only get bigger. I strongly believe that the companies that are not adopting digital transformation and analytics will decline and may even go out of business in the coming years. It's a matter of survival. We saw this with COVID-19, when many companies were not digitally transformed or making use of data, got hit much harder than others. COVID-19 was a bit of a stress test in that respect, clearly showing that companies need to invest more in these technologies." 

Artificial intelligence and machine learning are not new technologies, but today experiments often remain a proof of concept. “The time for experimenting with artificial intelligence (AI) is over. It is time to democratize AI, make the technology widely available on an integrated platform and incorporate techniques into day-to-day operational processes for value.” 

Share