Industry IoT, smart factories and AI in manufacturing
TL;DR;
The world of manufacturing is on the brink of another revolution due to the Internet of Things (IoT) and Artificial Intelligence (AI) applications. Aside from clear use cases like robotics and automation, big data applications are coming into play, thanks to industrial time series data collected by data historians. Thriving on all this data, AI systems can be built to send early warnings, optimise processes, predict maintenance and enforce quality control. By collecting the right data, manufacturers can get really creative with their AI solutions, and it can set them apart from the competition.
There is growing interest in the idea of smart factories with 92% of manufacturing executives believing that this is the way forward, but far less are actually putting in the research for AI solutions, and fewer still are putting those ideas into practice. That said, the Industrial IoT market has been steadily growing over the last few years and as the technology matures, costs are dropping making it easier for companies to access.
Industrial IoT and Industry 4.0
As IoT devices become commonplace items in modern-day society, so does the use of connected machinery and sensors on manufacturing floors, sending a wave of disruption through an industry that has been waiting for a revolution. Sensors and intelligent devices distributed across these shop floors, combined with cloud or edge computing, constantly collect data that can be used to drive AI and machine learning models.
The downside of the IIoT is that it comes with an increased vulnerability to cyberattacks. Cyber attacks are usually centred around stealing data, taking control of operating systems, or spying on the competitor. These attacks can be hugely costly to fix, but luckily AI can play a role in cybersecurity, too.
The best thing about IIoT is that to be a part of the revolution, you don’t have to replace all your equipment for smart machines. You can retrofit your legacy equipment with smart sensors, edge gateways and use video cameras on the plant floor. But what is all the data that these things collect being used for?
Predictive maintenance
Predictive maintenance is one of the most valuable applications of AI in manufacturing. Using machine learning models, your predictive maintenance system can identify when a part is likely to fail based on a combination of historic data and condition monitoring data that suggests whether a machine is functioning within its normal performance. Knowing when to schedule replacements removes the costs of downtime and delays associated with something breaking “out of the blue”. Predictive maintenance has been shown to reduce outages by 7-75% and increase ROI ten-fold. It can be so lucrative that IIoT often pays for itself and still increase profits.
Digital Twins
A digital twin is a digital representation of a process, product or service produced using IoT, machine learning and AI. It has several uses, one of them being that it can support predictive maintenance. The digital simulation can be designed to update and change so that it is always an accurate representation of the physical asset. This can help to identify problems or reveal how to optimise a process and allow the creation of simulations to see how something would work before trying it. It can be great for upkeep on machines, revealing the internal functioning, or for monitoring remote or inaccessible devices, such as remote wind turbines, or pipes under a road.
Not only can digital twinning be great for maintaining machinery, or trying new systems across a whole factory, but it can be used in product development. Using 3D computer-aided design (CAD) models, it is possible to test and improve many aspects of a design before actually producing the product, reducing the need and costs of functional redesigns.
Generative design
Generative design software can be used to speed up innovation. It works by generating multiple outputs based on a set of design requirements. Using this as a base, designers can fine-tune the outputs to create superior designs. When it comes to manufacturing, generative design can streamline the production process by enhancing innovation, boosting productivity and freeing up designers. It can also save costs by reducing the need for redesigns, create more reliable products with designs that are more fit for purpose, or that can be designed for a more positive environmental effect, or to use fewer recourses. This technique will transform product design in the coming years.
Computer vision
As mentioned above, installing HD video cameras in the shop floor is one of the cheaper ways you can add smart elements to a factory. Machine vision tools can be installed to spot minor defects and send alerts when it finds them to reduce problems in production and cutting quality control costs. Visual recognition tools can be used to aid robots and machines with things like label placement, package inspection and sorting.
Computer vision can also be used to track health and safety on the shop floor, from identifying individuals who are not wearing safety equipment to spotting contamination risks, the system may even intervene, blocking access when noncompliance is identified, or halting dangerous machinery to prevent injuries.
Visual recognition systems can be used for text and barcode reading which has several applications in manufacturing, such as checking the right parts are being used or sorting and tracking item through the factory. Computer vision can also be used to guide operators through complicated processes such as assembling an item, by checking codes on parts and interacting with the operator through gestures.
Anticipating the market
Manufacturing AI also has a place away from the shop floor. Market algorithms can evaluate patterns from consumer data and other influences to estimate demand and adapt to an ever-changing market. Social listening can also help determine how customers are feeling about products which may feed into redesigns. This data gives manufacturers the power to be strategic and anticipate changes rather than constantly lagging in response. These predictions can help optimise inventory control, staffing requirements, and energy consumption, which also helps regulate costs.
Risk management
Risk can be a very hard thing to track since some risks aren’t even known until they occur and others have questionable causes that aren’t fully understood. But with AI advancing computers beyond binary thinking, computers can analyse risk data in more dimensions than humans can, changing the game in the risk mitigation sector.
Inventory optimisation
Inventory management can be automated to drastically improve operations, but better still, by applying AI, it can be made to be smart. Learning from patterns in historical data, an AI-powered inventory system can strike the perfect balance between making sure stock is always available and not overstepping any budget or storage constraints. As ever, its the vast amounts of real-time data that makes these insights into supply and demand possible. But to add another smart layer to this process, some factories are even using robots to check and restock their inventories. Using any combination of these systems can boost productivity by up to 40%.
Final thoughts
Unquestionably, the future of factories lies in smart technology. As previously mentioned, transferring to smart manufacturing doesn't have to be done in one costly makeover. Smart devices can be added little by little, and legacy equipment can be freshened up by adding sensors rather than being replaced. As with embracing any form of AI, the key is finding out what processes can be supported by AI, and developing models based on your business needs. AI works best when complimenting workers, and those workers need to embrace the idea too.
Many of the techniques mentioned come together to improve overall functioning in manufacturing such as quality control, reducing operational costs, and relaying data to give smart insights. While more manufacturers are reaching into the smart sector, many are stuck in pilot or proof of concept stage. It is important to get past this stage and scale AI in manufacturing to reap the true benefits.