The manufacturing industry is currently facing a major upheaval, especially with the current supply chain problems.
“Manufacturing production in the United States increased 3.50 percent year-on-year in December of 2021, the smallest growth rate since the recovery started in March 2021.”[1] This statistic does not show great promise.
What pain points are manufacturers currently facing?
· Demand Forecasting
· Inventory Management
· Predictive Maintenance
· Pricing Volatility
· Quality Assurance
· Supply Chain Disruption
· Workforce Disruption
· Workplace Safety
· Indirect Cost Control
· A/R Collections
Many of these pain points can be addressed by the advanced technologies of artificial intelligence (Ai), computer vision, and machine learning. For example, Ai and machine learning are starting to deliver amazing results for the manufacturing industry.
· Demand forecasting is an extensively used machine learning application in supply chain planning today.
· Medical device manufacturer Cerapedics is using Ai and predictive maintenance to improve batch yields, lower costs, and mitigate risks. [2]
What is computer vision and machine learning?
Computer vision is a machine learning field that enables processing of images and video at a detailed level. Using those images or videos, the computer tries to understand what it is seeing and to identify anomalies, generate alerts, and extract structured data. It can do this much faster and more accurately than a human typically can.
The computer identifies and classifies objects in the image or video, such as a human being, piece of equipment, animals, and it can also recognize faces. Machine learning then helps the computer intelligently react to the given information by continually learning.
In the following sections, we’ll discuss how computer vision and machine learning can help the manufacturing industry.
The images show an application of advanced machine vision algorithms in operations monitoring that help producers understand how their facilities are affected, allowing swift preventative measures or corrective action to betaken.
Automatic Analysis of Behavior
The videos that these still images are taken from are used to check the interaction between humans and animals, analyze animal behavior, and check whether farm employees are treating the animals humanely and following safe workplace practices.
Faster and More Efficient
Our solution can monitor 30 cameras generating 720 hours of real-time video every 24 hour cycle and only taking 30 min of possible humane or safety breaches and putting on a dashboard for the farm manager to review. This amount of video would be almost impossible for humans to monitor accurately over the same time period and same amount of video captured.
Computer Vision Applications
There are many applications that this computer vision algorithm can detect, such as animals, tracking objects, human and animal behavior and interactions, behavior model diagnosis (is an animal walking normally), animal welfare, laboratory research.
All these applications can be applied towards manufacturing use cases, such as a workplace safety machine learning model that can detect:
· Attendance Tracking
· Workplace Productivity
· Safety Zone Incursion
· PPE Compliance Monitoring (Helmets, Goggles, etc.)
· Risky or Unsafe Behavior
Computer vision can be used to analyze closed-circuit television (CCTV), drone, and other video sources to enhance infrastructure surveillance and inspections, such as:
· Internal and External Inspections
· Building Inspection
· Equipment and Machinery Inspection
· Perimeter and Building Surveillance
Using Computer vision, manufacturing and other industries can speed up and improve the accuracy of building and infrastructure inspections by an order of magnitude, saving time and money and detecting more risks.
Machine Learning has many uses to monitor manufacturing equipment and prevent costly repairs by identifying a problem before a costly failure can happen.
Intelligent Infrastructure Monitoring
· Constantly monitor critical equipment/infrastructure
· Combine IoT, asset management and service management data for a more complete view
· Predict catastrophic failure for early intervention
· Predict and optimize maintenance needs for more efficient maintenance scheduling
· Use Computer vision to monitor moving machinery for faults and quality issues
Supply chains have come under unprecedented pressure in recent years with issues we never could have anticipated in normal times. Examples of this are:
· Component subcontractor factories shut down in Asia from zero-tolerance COVID-19 policies [3]
· Giant container ships getting stuck in the Suez Canal and backing up an estimated $14-15billion in goods on ships unable to pass through the canal [4]
· Huge price inflation in shipping container prices wreaking havoc with shipping calculations [5]
· Having to change transportation modes to achieve targets (i.e., rail, air, truck, sea)
· Just-in-Time (JIT) assumptions have been thrown out the window leaving manufacturers short on critical components such as vehicle management chips bringing entire car production lines to a halt or scaled back production.
All these factors have forced manufacturers and distributors to take a fresh look at their supply chain processes and assumptions. They must now consider other variables and data points when doing their calculations which include factors such as:
· Demand Forecasting
· Inventory Management
· Supply Chain Lead Times
· Delivery turnaround times
· Order to Cash (OTC) timing
Sufficient demand data comes from recurring customers and an abundant historical demand information. A bottom-up approach predicts each customer’s demand likelihood and is aggregated by a statistical machine learning model. Using a forecasting model, strategic planning, and pinpointed marketing tactics, the following impact can be delivered:
· Forecasting at various product levels and time horizons
· Enhancing pricing strategies
· Targeted sales to specific customer groups
· Increase profit margins and optimize inventory
· Seasonal adjustments
· Precisely designing promotions
Using sparse demand data can be more difficult. The demand comes from new customers and there is often not enough historical demand information. Using a top-down approach, you can forecast based on similar products and estimate through driving factors. Through a forecasting model, strategic planning, and pinpointed marketing tactics, the following impact can be delivered:
· Highly precise demand estimations
· Enhanced pricing strategies
· Targeted sales to specific customer groups
· Streamlined supply chain
· Precisely designed promotions
How ElectrifAi Can Help
How do you remain competitive in a global marketplace? By partnering with experienced Ai experts who can significantly improve your business operations.
ElectrifAi has been helping people do more with their data since 2004. We are on a mission to accelerate your time to value with our vast library of pre-built machine learning models. Rather than spend months or years trying to create a machine learning model that may never work – along with the hefty costs associated to build that model – we can get you the insights you need in weeks.
We sift through massive amounts of unruly and disparate data to home in on what matters. We place a thin layer of intelligence atop the chaos, and output the clear, crisp, and practical business knowledge you seek. We deliver the visibility and insights you need to reduce risk, work efficiently, and produce cost savings. It’s that simple.
Interested in learning how your data can begin working for you? Contact us today for a free consultation.
[1] Tradingeconomics.com, TRADING ECONOMICS, 2022,tradingeconomics.com/united-states/manufacturing-production.
[2] “Data Quality Key to Leveraging AI, Machine Learning.” Manufacturing.net,30 June 2021, www.manufacturing.net/technology/blog/21533379/data-quality-key-to-leveraging-ai-machine-learning.
[3]Naoki Matsuda. “China COVID Crackdown Closes Several Factories in Industrial Hub.” Nikkei Asia, 14 Dec. 2021,asia.nikkei.com/Spotlight/Supply-Chain/China-COVID-crackdown-closes-several-factories-in-industrial-hub.
[4] Russon, Mary-Ann. “The Cost of the Suez Canal Blockage.” BBC News, 29 Mar.2021, www.bbc.com/news/business-56559073.
[5] Rao, Sujata, and Jonathan Saul. “Analysis: Shipping Costs - Another Danger for Inflation-Watchers to Navigate.” Reuters, 10 Dec. 2021,www.reuters.com/markets/commodities/shipping-costs-another-danger-inflation-watchers-navigate-2021-12-10/.
