It’s not too late to watch our webinar ‘AI in Action - Real-world use cases for food businesses’ featuring an expert panel of speakers including:
- Ted Combs, Industry Principal for Consumer Products at AVEVA
- Glen Brittin, Head of Technical at Gosh! Food
- Craig Leadley, Chief Executive at the Institute of Food Science & Technology (IFST)
- Paul Van Wymeersch, Head of Marketing at Porky Whites
To sign up, just follow this link and watch for free.
If you were able to attend on 10 September, you will know that our panel were unable to get through all the questions posed by the audience on the day. As a result, we asked them to pick out some questions they thought were particularly pressing and offer some additional insight.
Can AI be used to deliver more accurate telemetry of key metrics in food manufacturing and can it eventually predict when processes are likely to fall outside acceptable control limits?
Ted Combs: “The answer is yes, and AVEVA already has such capabilities. You may be interested to know that AVEVA has had forms of AI offerings for fifteen years. In AVEVA’s current advanced analytics solution offering, it offers customers the ability to leverage data captured from machines via PI, processes via MES and more. Modelling and analysis can help an organization learn and predict acceptable control limits and the factors that contribute to variability.
“A large pet food manufacturer implemented AVEVA’s predictive models for middle of line quality for finished product density, moisture, fat and protein content. They now monitor more than 75 process variables from across the production process to make accurate predictions for all four quality parameters in real-time, plus provide recommendations to operators to keep quality on-target. In addition to predictive quality, AVEVA also has capabilities for predictive throughput and predictive energy efficiency.”
Glen Brittin: “I believe AI can be extremely useful for data novices to deliver insights about our factories. AI could perform the job of a data analyst to advise and help decision making. Creating and using data models is not a new concept where these have been created simulate a manufacturing process. Formally they have been referred to as decision support systems (DSS).
“The power of AI is in its ability to process the data. But in order to tap into machine information your factory needs to be able to capture all sources centrally eg. using SCADA. Your second part is talking about SPC - statistical process control. Many production line controllers use PID technique to manage processes. SPC requires sampling on a time base to generate graphs. AI could be used to do this and it can definitely be taught to interpret SPC charts providing warnings on out of control processes. SPC is already a predictive tool.”
When considering the development of AI applications in your organisation, which approach do you see as most relevant?
a. Building and training an in-house AI team
b. Working with external AI-specialised firms or technology providers
c. Collaborating with research institutions and universities
d. A combination of the above
Paul Van Wymeersch: “For Porky Whites (and many other SME manufacturing businesses), I’d say a combination of all three options is preferable. Why? Well, it comes down to several factors that can include budgets, a lack of access to the necessary in-house expertise, priorities and most importantly, the objective.
“Having the idea to integrate AI into your process or systems is just the start of your journey. You need to carefully consider the why, who, how and what you’re trying to achieve. Is this project borne out of the need for greater efficiencies, the opportunity to delve deeper into operational data or something else?
“For me, using a combination of internal and external resource just make sense. Not only will it help you identify how and where AI can add value (whether that be time or money), but this collaboration will also give you access to a range of skillsets. From development, data management/migration and digital project management specialists to testing, refinement and training, finding the right combination of AI partner(s) and/or consultants can be the difference between success, compromise and failure. When it comes to integrating AI – and for that matter any digital project – don’t cut corners.”
Glen Brittin: “Different departments have different requirements. In manufacturing and quality we are utilising functionality which comes with our manufacturing counting system. The AI can be provided with complex guidance to help process massive amounts of data very quickly saving a person significant amounts of time.
“As a people using AI we also need to think about what we are asking it to do - the better the quality and direction of what we are requesting will provide more accurate and precise results. Therefore it requires some practice and learning how to get the best results to act upon.”
I see AVEVA is focused on manufacturing plant optimisation. Which other system do you suggest for managing food safety and quality?
Ted Combs: “We like to say AVEVA helps manufacturers engineer, operate and optimise so really all of these solution areas or systems can contribute positively to the defence side. Food safety extends well beyond the walls of a plant to the entire value chain and AVEVA has solutions to help from farm to fork.
“AVEVA has many agribusiness customers but it also has a relatively new and very strategic solution called connect which enables organizations to capture, share, secure, analyse and visualize disparate data sources throughout the value chain. This allows them to assess, track and manage ingredients and finished goods throughout the connected value chain.”
How can AI help avoid product recalls relating to incorrect allergen labelling?
Glen Brittin: “GPTs could be used to examine trial labels for accuracy in the same way humans do. I have experimented with using AI to calculate proximate nutritional analysis from NPD recipes. A similar principle could be involved particularly if you are specific about the market legislation to be referred back to.
“I would suggest that this is not the main reason for allergen related product recalls rather the potential unintentional presence of allergens. In which case AI has been used to help look at trends in global supply chains to raise alerts - there are companies using and selling this type of technology.”
