Driving Better Decisions
Predicting outcomes of decisions that impact every phase of production is one of the challenges manufacturers face today. Many existing analytics apps and techniques look for causality in the data first, missing patterns in the data that provide greater predictive insight that manufacturers need. Machine learning apps are designed to optimise decisions and outcomes based on predictive patterns found in large-scale data sets. Instead of looking for causation, machine learning looks to find greater predictive accuracy in the data, delivering better decisions in the process. Machine learning’s advantages are a perfect fit for the challenges manufacturers face.
Machine learning algorithms are iterative in nature, continually learning and seeking optimal outcomes of a given query or decision. Every time an outcome is reached that is less than optimal for the given data sets and query, the algorithm again seeks to find the best possible outcome. These iterations often happen in milliseconds, simplifying complex patterns in large-scale data sets and delivering insights not available before.
Revolutionising Manufacturing Scalability
Manufacturing analytics and intelligence are driving a revolution in global manufacturing. Any given manufacturer’s ability to scale operations with greater accuracy and speed directly reflects how effective their analytics, manufacturing intelligence, and predictive insights are. Machine learning is providing greater speed, simplicity, and scale to global manufacturing by providing predictive insights that break down system barriers keeping shop floor and business systems apart. Machine learning and manufacturing intelligence are creating an entirely new platform for unifying the manufacturing IT landscape, delivering excellent results to customers while reducing costs and wasted time.
Five ways machine learning is revolutionising manufacturing:
Creating smarter factories from the machine- and shop-floor level to the top floor with more effective use of predictive insights, analytics and manufacturing intelligence.
Many manufacturers are in a race to stay in step with their customers, and none can afford to miss a product generation. Every manufacturer is also a global competitor now. Under constant pressure to respond reliability to customers with greater accuracy, clarity, and speed, manufacturers are looking to machine learning for insights on how to better manage every aspect of manufacturing. The Fourth Industrial Revolution is underway and may provide a framework to streamline manufacturing and become the global language of production. The following graphic provides an overview:
Source: BCM Public Relations
Gaining greater predictive accuracy of supplier’s quality levels, production yields and cost of quality and their impact on company-wide financial performance.
Supply chains are among the most vital series of systems and processes any manufacturer relies on to consistently deliver high-quality products. Using machine learning to select the optimal series of suppliers and scheduling the optimal series of machines and crews to build a highly customised jet can lead to significantly higher production yields. As machine learning is adept at analysing metadata of processes, capturing data on a jet wing’s electrostatic and electromagnetic properties provides real-time feedback on jet wing production yields. An example of this test is shown below. Once this data is captured and integrated back into the supplier quality management system, costs of going with one series of suppliers versus another are compared. Machine learning streamlines this entire workflow, leading to greater enterprise-wide quality management at lower cost.
Capitalising on machine learning’s iterative algorithms to measure and continually improve equipment, product line, and plant effectiveness.
One of the most common metrics used in manufacturing is Overall Equipment Effectiveness (OEE). Machine learning algorithms and the apps that are built on them are ideal for improving OEE as they can iteratively seek to optimise equipment uptime predictions based on integrated sensor data. As the Industrial Internet of Things (IIoT) sensors become more commonplace in manufacturing equipment, machine learnings’ contributions analysing the data and providing respective guidance will increase. Tata Consultancy Services found that OEE could be increased from 65% to 85% using these technologies. The following graphic illustrates machine learnings’ contribution to increasing OEE plant-wide, which has a direct effect on higher profitability.
Moving beyond the constraints of traditional Manufacturing Execution Systems (MES) and create greater consistency between the shop floor and top floor with machine-driven manufacturing intelligence.
Plant-centric MES systems by their very nature are well-suited for managing production operations including machine and team scheduling. These systems, however, lack the needed predictive insights to optimise the most challenging decisions of running multi site manufacturing operations. Using machine learning with enterprise Manufacturing Operations Management systems to optimise production scheduling across multiple operations, including real-time supplier orchestration has the potential to trim days off of production schedules, significantly reducing operating costs.
Gaining greater accuracy into pricing, production, sales and services forecasts by using machine learning to analyse sales and production data more efficiently.
Manufacturers in commoditised industries often assume that dropping their prices will increase demand, as many believe demand is still elastic or influenced by price alone. Machine learning is giving manufacturers a glimpse at the demand curve for their products and services, with many finding that dropping prices has the opposite effects in sales. One manufacturer running a pilot with machine learning algorithms designed to predict total market demand told me that increasing prices on service led to more sales. He explained that their machine learning apps had found that overall demand was flat, and price reductions would have no effect on selling more. Raising prices and augmenting services led to more sales and margins.
This article is re-published from the Dassault Systemes’ Navigate the Future blog.
At The AI Summit in San Francisco on 28-29 September, CxOs from the world’s leading enterprises will gather to explore the huge opportunity that AI presents the manufacturing industry. To find out more, and to join us at the Fort Mason Center in September, visit: theaisummit.com