AI and Machine Learning for Manufacturing Industry: Use Cases
Solution consultant at Sahaj Software Solutions
Artificial Intelligence(AI) has already proven to solve some of the complex problems across the wide array of industries like automobile, education, healthcare, e-commerce, agriculture etc. and yield greater productivity, smart solutions, improved security and care, business intelligence with the aid of predictive, prescriptive and descriptive analytics. So what can AI do for Manufacturing Industry?
AI sometimes misleads many manufacturers, suggesting a large end-to-end system. In reality AI is more often a collection technologies like Computer Vision, NLP(Natural Language Processing), Speech Recognition, Conversational AI (chatbots), Analytics and Automation - each with its own strengths and applications.
The Manufacturing industry has always been available to embrace the innovative technologies. Drones and industrial robots have been a part of the manufacturing industry since 1960's. From machinery inspection and diagnostics to production planning, AI-powered analytics enables manufacturers with the improvements in efficiency, product quality and safety of the employees.
Opportunities and Forecast:
- Manufacturers contribute over $2.3 trillion to the economy of USA every quarter (Bureau of Economic Analysis).
- According to an IFS survey of 750 manufacturing companies across 16 counties, 81% of the manufacturers have already embraced some type of transformational technology to digitise business processes.
- In a 2018 Forbes Insights Survey on Artificial Intelligence, 44% of the respondents from the automotive and the manufacturing sectors classified AI as "highly important" to the manufacturing function in next five year, while almost half (49%) said "it was absolutely critical to success".
- Global AI in Manufacturing Market is expected to reach $15.2 billion by 2025 from $513.6 million in 2017, growing at a CAGR of 55.2% as per Allied Market Research.
show that unplanned downtime costs manufacturers an estimated $50 billion annually, and the asset failure is the cause of 42% of this unplanned downtime. While Sensors, IoT and connectivity can fetch you operational data, advanced AI algorithms in the form of Machine Learning and Artificial Neural Networks help you to predict the next failure of a part, machine or system.
Predictive maintenance eliminates the guess work and promises cost savings over routine or time-based preventive maintenance, because tasks are performed only when warranted. It allows convenient scheduling of corrective maintenance, and to prevent unexpected equipment failures. A report by McKinsey and Company notes that AI-driven predictive maintenance can increase asset productivity by up to 20% and reduce maintenance cost up to by 10%.
This solution works like a diagnostic tool-kit. By acting upon high-volume data, it can provide actionable insights on daily basis - including root cause analysis - there by increasing the yield, reducing the expenditure and improving the overall ROI. In cases where maintenance is unavoidable, field technicians and operating teams will be provided with specific instructions on what should be fixed and when, resulting in very focused repairs.
If the machine is prone to wide range of un-understood failure modes(say some of them are very rare), it can be impractical to create sufficient models of high quality to predict them all.
Digital twin is a 'digital replica'/'virtual model' of an entity(a product, process or service). AI technology can build the accurate models using the CAD and AI tools to include theoretical and real-world data. This combination of data is enabling us to build digital twins more accurately. Digital twin leverages Internet of Things (IoT), but requires the skill of machine learning and AI. Having a digital model helps us predict the wear, movement and interactions with other devices. It also enables us to find limitations, bottlenecks, mistakes or better features to accelerate time to market.
Pairing of virtual and digital worlds allows engineers to see and test parts, entire machines and even productions lines, all digitally.
"Lessons are learnt and opportunities are uncovered within the virtual environment that can be applied to the physical world-ultimately to transform your business," Bernard Marr, author of Big data writes in Forbes.
Digital twin enables design engineers to know how the production line, supply chain and maintenance will be affected with the change in design specifications.
Worker/Work-place Safety Monitoring
Monitoring the employees and work-place/job-site helps to reduce the risk of accidents, there by yielding the increased work-force productivity. AI-powered computer vision solutions like safety gear detection and recognition in conjunction with human tracking can be used to ensure safety gear compliance. HAR(Human Activity Recognition), through spatial and temporal analysis (video analytics) can be used to monitor employees to prevent unauthorised access and avoid accidents, thereby providing the safety.
- Safety Gear Verification: Verify whether employees are wearing appropriate safety gears for a given job-site like safety helmets, vests, glasses/goggles, shoes or other protective gears, accordingly generate alerts and reports to enable safety gear compliance.
- Perimeter Protection: Monitor unauthorised access and prevent unauthorised people entering hazardous areas within the manufacturing plant.
- SOP Compliance: Monitor Standard Operating Procedures within the warehouse facilities and manufacturing plants, any anomalous step/activity that can lead to safety hazards can be reported and alerted in real time.
- Security Surveillance: Identify and report the employee fallen on the floor, vehicle breakdown in the premises, unauthorised crowding, camera blockage/tampering.
- Other Analytics: Detect and report the theft and misplacing of items within the warehouses and manufacturing plants.
In this fast-paced technology driven world, it's highly necessary for manufacturers to maintain high levels of quality and to comply with quality regulations and standards to meet short time-to-market deadlines. AI can be used to optimise time-consuming product validation process in the design of new products such as processors. This application of AI helps discover hidden bugs earlier in the design process. AI can be used to find microscopic defects in products such as circuit boards at resolutions well beyond human-vision, using a machine learning algorithm. AI can be helpful in notifying manufacturing teams of emerging production faults that are likely to cause product quality issues.
This method involves inputting of parameters defined by designers and engineers into an AI algorithm called "Generative Design Software".Parameters include cost and time constraints, material types and available manufacturing methods etc. The Algorithms explores all the possible configurations and generates design alternatives. The proposed alternatives can be tested using Machine Learning, the process will be iterated through all alternatives till the optimal and best solution is found. Good thing about this approach is there's no human bias factor which means no assumptions are taken into consideration and there's no logical starting point.
At a given time supervisor can only watch over few workers. AI with the help pf video analytics can be used to monitor and track employees, flag their failures and enable their learning from their better colleagues, there by leading to increased productivity and efficiency.
Track the breaks(smoking, lunch, washroom etc.) taken by employees and report productive working hours.In a manufacturing plant, at the assembly line track the assemblers' work which includes timing each step and checking for mistakes.Provide supervisors and workers with feedback so that they can avoid repeating an error.Activity analysis of employees within the manufacturing plant or warehouse can enable intelligent assignment of work/task.
Apart from the use cases discussed, AI can still bring lot more to the table and offer the competitive edge. AI-powered self driving vehicles can be used within the manufacturing plants to automate movement of materials. Voice-driven solutions may lead the charge. AI in conjunction with RPA(Robotic Process Automation) can provide much more advanced solutions to the smart warehouses. We can gather enough data from humans in manufacturing plant to train machines to mimic them. A report from London Business School shows that companies save little with robotics alone, but those that retool business processes with AI and robotics combined have seen cost savings as high as 30-70%.
"If you're stuck to the old way and don't have the capacity to digitalise manufacturing processes, your costs are probably going to rise, your products are going to be late to market, and your ability to provide distinctive value-add to customers will decline," says Stephen Ezell, an expert in global innovation policy at the Information Technology and Innovation Foundation.
The technology is moving faster, missing the AI wave could mean getting stranded and it's harder to compete if company falls behind. AI is not just a key to competitive advantage, it is now a requirement for survival in many industries. On the other hand companies are beginning to realise that they need not climb an AI mountain, they've to keep taking right, small and necessary steps to reach new heights. However many companies have legacy equipment that doesn't provide the data or send the data to the other locations and also other factors like confidence, maturity, scaling, ROI and connectivity are slowing AI's adoption in some cases.
AI may be efficient at creating the things, improving the things, automating the things and making the things cheaper, but there's no replacement for human ingenuity in dealing with unanticipated changes in tastes and demands. AI-powered extreme monitoring in a manufacturing plant (low-wage work environment) can be inhumane.
Subscribe to get your daily round-up of top tech stories!