Equipment maintenance, quality assurance and supply chain optimization are as essential to industrial and manufacturing processes today as they were over a century ago, but getting them right at the scale and complexity required in today’s global market is a challenge. Thanks to the convergence of data and machine learning, these sustainable industrial manufacturing practices are now in the process of being reinvented.
Every day, organizations generate massive amounts of data at the edge, store that information in the cloud, and use those resources to redesign virtually every process. To derive more insight from their data and ultimately make faster, more informed decisions, companies in manufacturing, energy, mining, transportation and agriculture are harnessing new types of technology for people. machines to improve industrial workloads such as engineering and design, production and asset optimization, procurement. chain management, forecasting, quality management, intelligent products and machines, etc.
From operational efficiency to quality control and beyond, here are four key ways companies are using machine learning to rethink industrial processes:
Predictive maintenance of equipment
Historically, most equipment maintenance has been either reactive (after machine failure) or preventive (performed at regular intervals to avoid breaking machines), both of which are costly and inefficient practices. The best solution, predictive maintenance, gives businesses the ability to predict when equipment will need maintenance.
While most businesses lack the staff and expertise to build their own solution, now there are end-to-end systems that use sensors and machine learning to detect and alert businesses to abnormal fluctuations in vibration. or machine temperature, without machine learning or the cloud. experience required. This kind of technology has helped GE Power, a leading provider of power generation equipment, solutions and services, quickly modernize assets with sensors and connect them to real-time analytics in the cloud, moving from temporal maintenance practices to predictive and prescriptive maintenance practices. And as they evolve, GE Power can use these systems to remotely update and maintain its fleet of sensors, without ever having to physically touch them.
Computer vision anomaly detection
Just as important as ensuring the proper functioning of equipment is ensuring the quality of the products produced by that equipment. Visual inspection of industrial processes typically requires human examination, which can be tedious and inconsistent. To improve quality control, industrial companies are turning to computer vision to provide increased speed and accuracy to consistently identify defects. Once again, complex barriers have prevented companies from building, deploying and managing their own machine learning-based visual anomaly systems. Now, businesses can use high-precision, low-cost anomaly detection solutions capable of processing thousands of images per hour to spot defects and anomalies, then flag images that differ from baseline to that appropriate action can be taken.
For example, Dafgards, a household food manufacturer in Sweden, uses computer vision in the production of its Billy’s Pan Pizza brand, a microwave-safe pizza baked and packaged at the speed of two pizzas per second. While they had previously installed a machine vision system to detect proper cheese coverage on their pizzas, it failed to detect defects on pizzas with multiple toppings. Using a new machine learning service that leverages computer vision, they were able to easily and cost-effectively scale their inspection capability. The company was so successful that Dafgards extended the use of computer vision to several varieties of pizzas, as well as other product lines such as burgers and quiches.
Improved operational efficiency
Many industrial and manufacturing companies are also looking to apply computer vision to help them in their efforts to optimize efficiency and improve operations. Today, companies manually review video feeds at their industrial sites to authenticate access to facilities, inspect shipments, and detect spills or other hazardous conditions. But doing this in real time is not only a difficult task, it is error-prone and expensive. And while companies may look to upgrade existing Internet Protocol (IP) cameras to smart cameras that have enough processing power to run computer vision models, it can be expensive and even with smart cameras, Getting low latency performance with good precision can be difficult. Instead, industrial companies can use hardware appliances that allow them to add computer vision to existing on-site cameras, or even use software development kits (SDKs) to create new cameras capable of running applications. significant peripheral computer vision models.
Global energy company BP is looking to deploy computer vision at its 18,000 service stations around the world. They strive to take advantage of computer vision to automate the entry and exit of fuel trucks to their facilities and to verify that the correct command has been completed. And computer vision can help alert workers to a risk of collision, identify a foreign object in a dynamic exclusion zone, and detect any oil leaks.
Supply chain optimization forecasts
Today’s modern supply chains are complex global networks of manufacturers, suppliers, logistics and retailers, which require sophisticated methods of sensing and adapting to customer demand, fluctuations in availability. raw materials and to external factors such as holidays, events and even the weather. The repercussions of not correctly predicting these variables can be costly, leading to over or under-provisioning and leading to wasted investment or a poor customer experience. To help predict the future, companies use machine learning to analyze time-series data and provide accurate forecasts that help them reduce operating expenses and inefficiencies, ensure greater availability of resources and products. , deliver products faster and reduce costs.
Machine learning has helped Foxconn, the world’s largest electronics manufacturer and technology solutions provider based in Taipei, Taiwan, deal with unprecedented volatility in demand, supply and customer capacity due to the COVID-19 pandemic. The company has developed a demand forecasting model for its plant in Mexico to generate accurate net order forecasts. Using the machine learning model, they were able to increase forecasting accuracy by 8%, resulting in a projected savings of $ 553,000 per year per installation, while minimizing wasted manpower and maximizing customer satisfaction. clients.
To live up to the potential that machine learning can offer to industrial environments, manufactured products, as well as logistics and supply chain operations, companies are increasingly turning to machine learning to make easier, faster and more accurate processes. By combining real-time data analytics in the cloud and machine learning at the edge, industrial companies are gradually turning their aspirations into realities and driving the next industrial revolution.
Swami Sivasubramanian is Vice President at AWS in charge of all Amazon AI and Machine Learning services. His team’s mission is to “put machine learning capabilities in the hands of every developer and data scientist.”