Today, Oracle announced its new Oracle Adaptive Intelligent Applications (AIA) aimed at helping the manufacturing sector leverage analysis and actionable insights to reduce costs and increase yields. Available now, the cloud-based applications combine machine learning (ML) and artificial intelligence (AI) technologies to process data from a variety of sources to identify and solve issues that limit efficiency in the manufacturing process.
“We have been working with customers over the last three and a half years to figure out what type of business problems they are trying to solve,” said Ramchand Raman, vice president, Oracle Product Development. “In talking to the industry, we found that that a need to improve yields and a focus on quality/cycle times were nearly universal. We found a lot of heterogenous systems. Our goal was to look at the manufacturing supply chain industry and see what techniques could be applied.”
By focusing on these two KPIs, Oracle looked at how data streams from Internet of Things sources, machine logs, and data from company systems could be applied to create meaningful analytics, Raman explained. Oracle AIA gathers structured, semi-structured and, unstructured data from a variety of sources such as manufacturing execution systems (MES), quality management, enterprise resource planning (ERP), human capital management (HCM), customer relationship management (CRM) and operational technology (OT) systems. Possible inputs include sensor and log data from equipment and machines as well as external environmental data such as humidity, temperature, etc.
By identifying anomalies during production, drilling down to find the root cause of the issue, and predicting events before they happen, the applications, which are built on the Oracle Cloud Platform, can monitor all stages of manufacturing from production through customer delivery.
“Traditionally, pattern and correlation analysis, genealogy and traceability analysis and predictive analysis are done by a small group of specialist data scientists” said Raman. “Oracle Adaptive Intelligent Applications for Manufacturing dramatically simplify the output of complex machine learning and AI algorithms and present these insights to average business users to drive better, faster decision making.”
The applications include:
- Pattern and correlation analysis: The software can identify key patterns and correlations in multiple factors (including manpower, machine, method, material, and management related information). These insights, when combined with with manufacturing business metrics such as yield, quality, cycle time, cost, scrap, rework, and returns, can help users to help quickly identify root causes of issues.
- Genealogy and traceability analysis: “Smart recall” analysis provides the ability to do backward and forward tracing of products and processes to figure out which products, services, and customers might be impacted by an issue.
- Predictive analysis: This solution predicts the likelihood of occurrence of critical outcomes such as yield, defects, scrap, rework, cycle time and costs for production activities to that users can intervene early to minimize losses.
“The system provides contextualization,” said John Barcus, vice president of Manufacturing Industries Business Unit at Oracle. “It allows users to do things holistically and tie pieces of data together in ways that provide a much better answer. Customers start to do machine learning and then want to do more artificial intelligence to engage and predict results in order to look at a much bigger picture. This solution has the opportunity to simplify a lot of things where once organizations needed individual expertise around a lot of areas. Many companies want to do something but aren’t prepared to staff up with a data scientist.”
The system allows users to work with the data themselves to understand the relationship between different variables. “The applications allow you to see dependencies and correlations not typically visible to a normal person,” said Raman. ”You can find hundreds of factors that impact the business and then drill down to find the five or ten primary attributes that influence the outcome. Then they can take corrective action before the problem arises.”
The software is currently available. Pricing is set at $100 per user per month with a 30-user minimum and includes 10TB of data storage. An extra 10TB of data storage costs $950 per month. The price includes all the necessary software, hardware and hosting support, Raman added.
— Hailey Lynne McKeefry, Editor in Chief, EBN