The manufacturing industry is undergoing a massive transformation, as advanced technologies are streamlining the way complex processes are managed. In particular, manufacturers can improve their operations by minimizing machine downtime, predicting maintenance needs, and optimizing factory floor resources. It is in these areas that artificial intelligence (AI) and machine learning are helping to disrupt the manufacturing industry – by making it easy to collect data on machine performance and immediately find solutions.
Key AI Segments That Impact Manufacturing
AI is a collective term for learning system capabilities that are perceived as representing intelligence, including image and video recognition, prescriptive modeling, smart automation, advanced simulation, and complex analytics, among many others, according to Cap Gemini. In the context of manufacturing processes, AI use cases revolve around the following technologies:
- Machine learning: Using algorithms and data to automatically learn from underlying patterns without being explicitly programmed to do so.
- Deep learning: A subset of machine learning that uses neural networks to analyze things like images and videos.
- Autonomous objects: AI agents that manage tasks on their own, such as collaborative robots or connected vehicles.
AI for manufacturing is expected to grow from $1.1 billion in 2020 to $16.7 billion by 2026 – an astonishing CAGR of 57 percent. The growth is mainly attributed to the availability of big data, increasing industrial automation, improving computing power, and larger capital investments.
A Game Changer for Heavy Asset Manufacturing
With the help of AI, heavy manufacturers have changed their tune on keeping operations efficient. They used to finance improvements with capital expenditures (i.e. they spend lots of money on new equipment to replace malfunctioning equipment). AI is a less costly alternative, letting these companies 1) more effectively analyze machine data to find proactive maintenance solutions, and 2) replace manual monitoring activities by machine operators with automated AI decision-making on equipment status.
A recent report from McKinsey illustrates a use case for a cement manufacturer that created an AI-driven “asset optimizer” to improve output from a vertical raw mill. The solution first quickly captured millions of lines of data from hundreds of process variables. It then mapped the data to automated workflows and applied neural networks and analytical algorithms to optimize process controls. The optimizer was able to go on autopilot, working autonomously without operator intervention to control the manufacturing process. The results: the solution delivered an 11.6 percent improvement over the manual mode after eight months, helping to raise profits for the manufacturer.
The Importance of Predictive Maintenance
Ongoing maintenance of production machinery and equipment is a major expense for manufacturers and can have a dramatic effect on operational profits. Unplanned downtime costs manufacturers $50 billion annually, and asset failure is the cause of 42 percent of unplanned downtime.
The concept of predictive maintenance uses AI algorithms to anticipate when machines and equipment might fail. AI can be trained to continuously monitor sensors on equipment, predict when they are likely to fail, and recommend proactive, condition-based maintenance schedules. Examples are deviations from material formulas, subtle changes in equipment behavior, or changes in raw materials. The result is less downtime and extending the remaining useful life of equipment.
A key use case from General Motors highlights a predictive maintenance process. The auto company analyzes images from cameras mounted on assembly robots to identify signs of failing robotic components. In a pilot test, the AI-powered system detected 72 instances of component failure in 7,000 robots, finding the problem before it could impact production with an unplanned outage.
Improving Product Safety and Enabling Smart Factories
Tires on automobiles are safety components, so manufacturers like Bridgestone strive for perfection when it comes to quality. To reach that goal, the company created a machine learning system to provide automated control over quality assurance, usually dependent on human monitoring. The system uses sensors to measure 480 quality items and automatically controls the machine in real time. Product uniformity was improved by 15 percent over the human-controlled process, and productivity was doubled by removing bottlenecks in the molding process.
Machine learning will also be at the heart of tomorrow’s smart factories. Nokia, for example, recently introduced a video app that leverages machines learning to alert manufacturing operators when the production process experiences anomalies so that they can be corrected in real-time. And with a 5G network, the assembly line can communicate the problems directly to the datacenter more than 600 km away.
Master Deep Learning, Machine Learning, and other programming languages with Artificial Intelligence Engineer Master’s Program
Build Your AI Skillsets Today
AI is now at the heart of the manufacturing industry, and it’s growing every year. Skillsets are still in short supply, so there is value in training for AI engineers who can create practical applications using a wide range of intelligent agents; machine learning experts who are trained in supervised and unsupervised learning, mathematical and heuristic techniques and hands-on modeling; and deep learning experts who learn to master TensorFlow, the open-source software library designed to conduct machine learning and deep neural network research.