Edge Computing is Now Really at the Service of Predictive Maintenance

One of the main challenges for companies and organizations engaged in the digitization process is connecting operational technology (OT) – the workshop, to be clear – to the IT infrastructure. Better connectivity between OT and IT leads to clear advantages by improving the way we design, build and manage industrial services and machines. What are these advantages and, above all, how can they be obtained? We asked Marco Filippis, Export Marketing Coordinator at Mitsubishi Electric, a company that is certainly at the forefront and is showing how its solutions as well as its e-F@ctory partners’ solutions can help to achieve this.

di Fabrizio Garnero

“Technology can create a competitive advantage by providing maintenance and detect anomalies to improve productivity or efficiency of existing equipment,” begins Mr Filippis. “The first step in optimizing OT/IT connectivity necessarily requires the processing and use of that large amount of data that is now made available by the equipment on our machines, whether these are robots, machine tools or production lines. Data can be pre-processed locally and aggregated to create valuable information, connecting seamlessly the factory level to higher-level IT systems such as MES and ERP platforms. Now, companies with this type of need have a new edge computing solution called MELIPC, presented at the SPS in Nuremberg at the end of last year. It makes it possible to optimize processes such as predictive maintenance or quality control with real-time data analysis and timely feedback for the operator”.

The intermediate level of the pyramid
According to its “vision”, Mitsubishi Electric considers the industrial world as a pyramid; starting from the so-called shop floor, you get to the IT system by passing through the intermediate level of edge computing. What has been done over the years by Mitsubishi Electric was first of all to integrate the shop floor thanks to an iQ-R unique automation platform that acts as a master for all the automation devices at this level. Focusing attention on robotics, the first major innovation compared to other manufacturers was therefore to consider the robot “simply” as one of the elements in a wider solution, despite its key role. “Now, what we have done inside the smart factories, starting from this concept of horizontal integration on the shop floor (where Mitsubishi Electric has been technologically ready for about fifteen years), is creating a complete vertical integration that brings the shop floor to IT systems. To do this, meanwhile, we have released this product, called MELIPC, which integrates algorithms for predictive maintenance and quality control, creating an architecture capable of processing data from the field to return them with timelines close to real time, so that we are able to intervene in a predictive way, while also informing the various company IT systems”.
“Oversimplifying, it is as if having real-time data that I have to return to the field to intervene and correct any anomalies very quickly, it is possible to do so by acquiring, among other things, all the surrounding environment variations thanks to the artificial intelligence algorithms that we have developed. Instead, if I have to transfer that data to a higher system, that can be an ERP, the cloud or whatever is at IT level, I filter it and transfer it without any problem”.

A new awareness in data management
MELIPC is a very powerful product that manages many more machines and, therefore, not only robots, even if, once dropped into the world of robotics, this platform has a particular value for the simple fact that robots have always provided data outside. What changed is the awareness of how they are managed. In continuous production applications, it can improve processes and adapt to variations. Efficiency is improved by generating a predictive model in real time, which can be updated and fine-tuned according to the new characteristics of the process. For example, a machine for assembly or packaging in an application for the food & beverage sector can foresee problems and avoid errors, thus improving quality. The MELIPC solution is capable to recognize anomalies in the machine operation by using artificial intelligence (AI). In an automotive painting plant, for example, it can be used to avoid downtime. Detailed status information is reported to the automation level so that machine operators can perform quick adjustment or proactive maintenance to maintain efficiency and increase availability.
Processing of production data can also improve process safety and integrity. In a life science market application, the MELIPC solution can be used to store and analyse sensitive information at the OT level, rather than being sent to the cloud. This also provides much faster real-time diagnosis, reducing IT infrastructure costs and requirements.
An applicable and scalable approach to any company
Of all that concerns the concept of integration, speaking of Mitsubishi Electric, we could safely say that it is the direct result of the first-hand experience gained at its production plants, which has been then “translated” and transferred to the production realities of customers, from large industry to the classic Italian PMI. The software component can indeed be tailored to the specific organization in order to have control of production means and eliminate all the inefficiencies within the factory.
“The concept that Mitsubishi Electric wanted to develop after creating a layer that would allow us to be fully integrated on the shop floor, is therefore tied to a vertical connection which enable us to respond effectively and efficiently to the current requests of smart factories, as this is a totally customizable and scalable approach according to the specific customer needs” explains Marco Filippis. “They would seem to apply exclusively to large industry but I assure you that this is a very interesting approach also for small business, which perhaps can reap the greatest benefits, obviously with a customized and tailored solution for their production problems.
At the shop floor level, it is possible to act through dedicated hardware and sensors directly on the robot so that it can adapt to the surrounding environment in which it operates, improving performance. It is thanks to a platform like MELIPC that even a small, evolved reality has the ability to take data from the field to manage it remotely with different purposes, including, for instance, that of predictive maintenance, up to the IT system where robots, machines and plants are visualized, analysed and managed in order to get reports to plan any type of optimization, efficiency and improvement in the field of production performance”.
“In this context – continues Mr Filippis – another key aspect on which Mitsubishi Electric is working is training, not only from a technical point of view, but also – above all, I would say – by implementing a “training” of the market through a dense modulated communication activity based on the interlocutor, perhaps without taking for granted some concepts that for some big companies are already in daily use, while for many small companies – which make our production system great – it is not so. Although much has been done with the National Industry 4.0 Plan, there is still a long way to go. Basically, however, Mitsubishi Electric’s approach is applicable and scalable to any type of company”.
Predictive maintenance is certainly one of those issues on which Mitsubishi Electric has been focused and will continue to do so in the near future because “we believe that the market is now ready, at least to a great extent” says Mr Filippis. “We are convinced that it is therefore looking for tools and solutions that can help the user to understand the robot state of health in advance and not having to manage an emergency but instead planning interventions in a wisely and smartly without negative effects on productivity”.