Skip to content

Posts from the ‘Manufacturing’ Category

10 Ways Machine Learning Is Revolutionizing Manufacturing

machine learningBottom line: Every manufacturer has the potential to integrate machine learning into their operations and become more competitive by gaining predictive insights into production.

Machine learning’s core technologies align well with the complex problems manufacturers face daily. From striving to keep supply chains operating efficiently to producing customized, built- to-order products on time, machine learning algorithms have the potential to bring greater predictive accuracy to every phase of production. Many of the algorithms being developed are iterative, designed to learn continually and seek optimized outcomes. These algorithms iterate in milliseconds, enabling manufacturers to seek optimized outcomes in minutes versus months.

The ten ways machine learning is revolutionizing manufacturing include the following:

  • Increasing production capacity up to 20% while lowering material consumption rates by 4%. Smart manufacturing systems designed to capitalize on predictive data analytics and machine learning have the potential to improve yield rates at the machine, production cell, and plant levels. The following graphic from General Electric and cited in a National Institute of Standards (NIST) provides a summary of benefits that are being gained using predictive analytics and machine learning in manufacturing today.

typical production improvemensSource: Focus Group: Big Data Analytics for Smart Manufacturing Systems

  • Providing more relevant data so finance, operations, and supply chain teams can better manage factory and demand-side constraints. In many manufacturing companies, IT systems aren’t integrated, which makes it difficult for cross-functional teams to accomplish shared goals. Machine learning has the potential to bring an entirely new level of insight and intelligence into these teams, making their goals of optimizing production workflows, inventory, Work In Process (WIP), and value chain decisions possible.

factory and demand analytics

Source:  GE Global Research Stifel 2015 Industrials Conference

  • Improving preventative maintenance and Maintenance, Repair and Overhaul (MRO) performance with greater predictive accuracy to the component and part-level. Integrating machine learning databases, apps, and algorithms into cloud platforms are becoming pervasive, as evidenced by announcements from Amazon, Google, and Microsoft. The following graphic illustrates how machine learning is integrated into the Azure platform. Microsoft is enabling Krones to attain their Industrie 4.0 objectives by automating aspects of their manufacturing operations on Microsoft Azure.

Azure IOT Services

Source: Enabling Manufacturing Transformation in a Connected World John Shewchuk Technical Fellow DX, Microsoft

  • Enabling condition monitoring processes that provide manufacturers with the scale to manage Overall Equipment Effectiveness (OEE) at the plant level increasing OEE performance from 65% to 85%. An automotive OEM partnered with Tata Consultancy Services to improve their production processes that had seen Overall Equipment Effectiveness (OEE) of the press line reach a low of 65 percent, with the breakdown time ranging from 17-20 percent.  By integrating sensor data on 15 operating parameters (such as oil pressure, oil temperature, oil viscosity, oil leakage, and air pressure) collected from the equipment every 15 seconds for 12 months. The components of the solution are shown

OEE Graphic

Source: Using Big Data for Machine Learning Analytics in Manufacturing

  • Machine learning is revolutionizing relationship intelligence and Salesforce is quickly emerging as the leader. The series of acquisitions Salesforce is making positions them to be the global leader in machine learning and artificial intelligence (AI). The following table from the Cowen and Company research note, Salesforce: Initiating At Outperform; Growth Engine Is Well Greased published June 23, 2016, summarizes Salesforce’s series of machine learning and AI acquisitions, followed by an analysis of new product releases and estimated revenue contributions. Salesforce’s recent acquisition of e-commerce provider Demandware for $2.8B is analyzed by Alex Konrad is his recent post,     Salesforce Will Acquire Demandware For $2.8 Billion In Move Into Digital Commerce. Cowen & Company predicts Commerce Cloud will contribute $325M in revenue by FY18, with Demandware sales being a significant contributor.

Salesforce AI Acquisitions

Salesforce revenue sources

  • Revolutionizing product and service quality with machine learning algorithms that determine which factors most and least impact quality company-wide. Manufacturers often are challenged with making product and service quality to the workflow level a core part of their companies. Often quality is isolated. Machine learning is revolutionizing product and service quality by determining which internal processes, workflows, and factors contribute most and least to quality objectives being met. Using machine learning manufacturers will be able to attain much greater manufacturing intelligence by predicting how their quality and sourcing decisions contribute to greater Six Sigma performance within the Define, Measure, Analyze, Improve, and Control (DMAIC) framework.
  • Increasing production yields by the optimizing of team, machine, supplier and customer requirements are already happening with machine learning. Machine learning is making a difference on the shop floor daily in aerospace & defense, discrete, industrial and high-tech manufacturers today. Manufacturers are turning to more complex, customized products to use more of their production capacity, and machine learning help to optimize the best possible selection of machines, trained staffs, and suppliers.
  • The vision of Manufacturing-as-a-Service will become a reality thanks to machine learning enabling subscription models for production services. Manufacturers whose production processes are designed to support rapid, highly customized production runs are well positioning to launch new businesses that provide a subscription rate for services and scale globally. Consumer Packaged Goods (CPG), electronics providers and retailers whose manufacturing costs have skyrocketed will have the potential to subscribe to a manufacturing service and invest more in branding, marketing, and selling.
  • Machine learning is ideally suited for optimizing supply chains and creating greater economies of scale.  For many complex manufacturers, over 70% of their products are sourced from suppliers that are making trade-offs of which buyer they will fulfill orders for first. Using machine learning, buyers and suppliers could collaborate more effectively and reduce stock-outs, improve forecast accuracy and met or beat more customer delivery dates.
  • Knowing the right price to charge a given customer at the right time to get the most margin and closed sale will be commonplace with machine learning.   Machine learning is extending what enterprise-level price optimization apps provide today.  One of the most significant differences is going to be just how optimizing pricing along with suggested strategies to close deals accelerate sales cycles.

Additional reading:

Cisco Blog: Deus Ex Machina: Machine Learning Acts to Create New Business Outcomes

Enabling Manufacturing Transformation in a Connected World John Shewchuk Technical Fellow DX, Microsoft 

Focus Group: Big Data Analytics for Smart Manufacturing Systems

GE Predix: The Industrial Internet Platform

IDC Manufacturing Insights reprint courtesy of Cisco: Designing and Implementing the Factory of the Future at Mahindra Vehicle Manufacturers

Machine Learning: What It Is And Why It Matters

McKinsey & Company, An Executive’s Guide to Machine Learning

MIT Sloan Management Review, Sales Gets a Machine-Learning Makeover

Stanford University CS 229 Machine Learning Course Materials
The Economist Feature On Machine Learning

UC Berkeley CS 194-10, Fall 2011: Introduction to Machine Learning
Lecture slides, notes

University of Washington CSE 446 – Machine Learning – Winter 2014

Sources:

Lee, J. H., & Ha, S. H. (2009). Recognizing yield patterns through hybrid applications of machine learning techniques. Information Sciences, 179(6), 844-850.

Mackenzie, A. (2015). The production of prediction: What does machine learning want?. European Journal of Cultural Studies, 18(4-5), 429-445.

Pham, D. T., & Afify, A. A. (2005, July). Applications of machine learning in manufacturing. In Intelligent Production Machines and Systems, 1st I* PROMS Virtual International Conference (pp. 225-230).

Priore, P., de la Fuente, D., Puente, J., & Parreño, J. (2006). A comparison of machine-learning algorithms for dynamic scheduling of flexible manufacturing systems. Engineering Applications of Artificial Intelligence, 19(3), 247-255.

Advertisements
%d bloggers like this: