Issue link: https://maltatoday.uberflip.com/i/1284572
3.9.2020 8 OPINION Alexiei Dingli Prof Alexiei Dingli is a Professor of AI at the University of Malta and has been conducting research and working in the field of AI for more than two decades, assisting different companies to implement AI solutions. He forms part of the Malta.AI task-force, set up by the Maltese government, aimed at making Malta one of the top AI countries in the world T he use of Artificial Intelligence (AI) in Industry 4.0 has been baking for quite a while, but technology is now ripe enough to turn it into reality. It involves the infusion of data and smart au- tomation within the manufacturing pro- cess together with human input. To do so, it makes use of various technologies such as the Internet of ings, Cloud Comput- ing, Blockchain and others. Let us try for a second to understand why the use of AI is important. If you are the owner of a manufacturing plant, then you're probably familiar with having various machines working in tan- dem and producing thousands of devic- es every day. Monitoring the different processes which go on simultaneously is already a huge headache. Most probably, you are at the mercy of the various oper- ators in order to ensure that the level of production is at least maintained. How- ever, even though you are taking the best possible measures to mitigate any risks and ensure that everything runs smooth- ly, things still go wrong. When they do, havoc breaks loose until the main issues are sorted out. ings get further com- plicated when problems arise during a weekend or a public holiday since the availability of skilled professionals is very limited and costly. When produc- tion is halted, it obviously results in fi- nancial losses which might also impact production deadlines. Large produc- tion processes would normally depend on warehouses of raw materials, parts or other items. All of these cost a lot of money and involve huge risks if they are mismanaged. e handling of these as- sets alone is already a huge headache for management. And we haven't even start- ed considering how to improve the plant; maybe through optimisation or changing the basic configuration. But sometimes, when dealing with legacy systems, any change can be catastrophic and as such, management might be slow to respond towards the changing needs of the mar- ket. Sometimes, it is simply too late and the lack of decisive action leads the plant towards certain doom. Do these scenarios sound too familiar? Well, the good thing is that most of them can be handled quite easily with today's technologies. To understand what we mean, let us give you some examples. Micro-monitoring at a distance Consider managing a plant with hun- dreds of machines, working like clock- work. To complicate matters, let's im- agine that there is more than one plant distributed around the world and locat- ed in different time-zones. How can you manage all of that? e answer is that one can't microman- age it! You'll have people to do that, espe- cially when the numbers start growing. But we all know that people are not infal- lible. e solution to that is to install an Intelligent Digital Twin (IDT) system ca- pable of monitoring all of your machines with 24/7 constant coverage, irrespective of where they are physically located. You can forget about it and let the system take care of the issues which arise, whilst only alerting you when your attention is really needed. Furthermore, it will give you the peace-of-mind which you need since you can zoom into the inner work- ings of any machine when you want and in real-time. Optimising your plant Plant optimisation and improvement is a big game-changer. However, it is fre- quently ignored because as the saying goes, "if it ain't broke, don't fix it". But how true is that? Did anyone ever try calculating the opportunity loss of doing things in a different way? Most probably, it never happens because nor- mally, you would have to rely on experts to drill down to that information. With AI, this scenario changes because the system actually performs these cal- culations without needing to be prompt- ed by anyone. It will constantly analyse the supply chain and quickly simulate the ideal and most efficient scenario for any particular job. It does so by looking at the current production, predicts po- tential downtime, slot in new tasks in the pipeline and make suggestions based upon deadlines such as delivery targets, the material available, etc. Predicting the future Predicting the future is not something easy but it is a very desirable feature es- pecially when one considers that issues such as machine downtime, is one of the largest contributors to loss of pro- duction. In fact, 80% of companies are unable to calculate their true downtime costs. On average, companies record 4 hours of downtime daily! Many have tried to tackle this feat but very few predictions can really survive the test of time. However, when deal- ing with a closed system, past events can give a very good indication of what will happen in the future. Machines in a manufacturing plant can be considered as a closed system which has been run- ning for months and in some cases, for years. Being mechanical in nature, they constantly log heaps of precious diag- nostic data which is probably lying in some database just in case the engineers need it. In reality, sifting through all that data is rather painful, precisely because of its volume, so people seldom refer to it. However, AI systems excel at process- ing huge volumes of data and analysing it. Specifically, they find it rather easy to locate patterns and these patterns can give a good indication of what is going to happen. Hence, an AI system can easily make predictions on when the machine is going to fail and for what reason. is will allow the plant to organise prescrip- tive maintenance which shifts the man- agement's strict dependence on planned maintenance, to being able to take re- al-time action based upon actual events. Furthermore, the system will not only predict a potential failure but it will also help management identify the root cause of the issue. us, some of the failures might be eliminated from the produc- tion cycle once and for all, whilst oth- ers will be handled in a timely manner. In so doing, unplanned downtime can be heavily reduced or even eliminate in some cases, thus maximising profitabil- ity and equipment reliability. A factory made of Lego In an ideal world, one would gather the machines and move them around un- til the right configuration is found. is would minimise costs and optimise the available space. Furthermore, when the plant starts growing and new machines are added, different configurations might be beneficial. In reality, this does not happen because it takes a lot of time and effort to do so, probably so much that it outweighs the benefits. However, with an IDT system, this can be achieved rather easily. An IDT sys- tem already has a virtual replica of all the machines and their status is updated in realtime. Changing the configuration of a system thus becomes child's play since moving virtual machines and changing configurations is similar like playing with Lego. Machines can be added or removed and their output can be easily simulated using the historical informa- tion which the system already possesses. us, different trials can be set up, either by the managers or by the AI system controlling the plant, in order to find out the best configuration possible. Once done, the value of such a configuration is assessed and if it outweighs the costs, it can be deployed in the physical system on the factory floor. Asset-management made easy Managing all the assets in an organisa- tion is a big headache especially in plants having hundreds, if not thousands, of machines working simultaneously. Be- cause of this, predictions become even more valuable, especially when manag- ing the purchasing of spare parts in or- der to replace defective components or when planning for forthcoming machine reconfigurations. All the components, both the pur- chased ones and those produced through the manufacturing process can be tracked through their unique identi- fier. Some of them will be tagged using technologies such as Radio Frequency IDs (RFIDs) which seamlessly provide information to the central system with regards to the physical movement of components. us, with the click-of-a- button, the person managing the system will have an up-to-date overview of all the physical stock available in the plant at any one point in time. Long gone are the days of never-ending stocktaking. e system will provide additional informa- tion such as products on order or those in transit. Furthermore, if the tracking module is built upon a blockchain sys- tem, one can ensure full traceability for any product; starting from the raw ma- terials which make up the product, up to the end of the product's lifecycle in the recycling stage. Better decision making IDT systems will be there to assist management in the provision of timely information, in the sifting through the sea of data and in the creation of summa- rised viewpoints over the operations of the various plants. Without doubt, they will become the personal assistants of management capable of not only giving accurate snapshots but of also highlight- ing areas of concern in real-time. Furthermore, the time will come when the management of the organisation will be entrusted in the hands of the AI system thus automating the entire deci- sion-making process. Decisions will be taken at the speed of light and correc- tive actions will be dispatched as soon as something happens. Managers will be relieved from the day-to-day microman- agement, they can look at the plant from the macro perspective and spend more time on planning future improvements in line with the direction imparted by the board of directors. Conclusion e future looks bright and exciting for the industries of tomorrow. But the fu- ture starts today since most of the tech- nologies mentioned above are already available. So what are you waiting, jump on the bandwagon of change and pre- pare your organisation for the challenges of the future! The most significant change since the Industrial Revolution