How Anomaly Detection Can Save Your Manufacturing Company Time and Money

Blog: The Tibco Blog

Reading Time: 3 minutes

During times of inflation, volatile supply chains, labor shortages, and increased environmental concerns, manufacturers are witnessing industry-wide change. To get ahead of these challenges, manufacturers (and other leading industries) must expand revenue streams, reduce operating costs, evolve strategic capabilities, and analyze risks. Anomaly detection is a critical step towards resilience for market leaders.

With mature anomaly detection capabilities, you can achieve substantial operational cost reductions and product quality improvements. Some of the benefits include reduced defects, lowered equipment downtime, and optimized energy consumption. Anomaly detection allows you to optimize operations, providing a needed hedge against supply chain and labor risks, volatile energy costs, and new entrants.

Interested in applying anomaly detection to your company? Read on to learn about anomaly detection methods and how your competitors are already innovating with TIBCO technology.

How Anomaly Detection Works

An anomaly is an unexpected change or deviation from an expected pattern in a dataset. It’s a great method for instantly detecting deviations from expected results. In manufacturing, anomalies can exhibit in machine sensor data, process measurements, or product characteristics. Analytics can investigate anomalies to determine their effect on machine downtime, process variability, and product reliability.

When data shows an anomaly, it often indicates something has changed or is behaving abnormally—leading to actionable insights—so you can correct any issues before they become widespread or time-consuming.

Common Methods of Anomaly Detection

Manufacturers have many different techniques and methods for using anomaly detection. The three main methods are visual discovery, supervised learning, and unsupervised learning.

Visual Discovery

With visual discovery, data or business analysts build data visualizations to find unexpected behavior, often requiring prior business knowledge and creative thinking to find the answers with the right data visualizations. While humans can easily spot patterns in data, at scale this may not be a timely endeavor.

Supervised Learning

Supervised Learning has analysts label a set of data points as normal or anomalous. A data scientist then uses this labeled data to build machine learning models to predict anomalies on unlabeled new data. Supervised learning is a great option for known patterns.

Unsupervised Learning

In unsupervised learning, analysts build unsupervised machine learning models with unlabeled data to predict new data. Since the model is tailored to fit normal data, the small number of anomalous data points stand out. This is a great option when new patterns are continuously emerging, like in fraud or manufacturing.

There are several key unsupervised learning techniques:

Autoencoders
Clustering
Support vector machine
Time series techniques

Optimize Operations like Hemlock

Hemlock Semiconductor (HSC) is one leading manufacturer that utilizes process control, predictive maintenance, and anomaly detection solutions from TIBCO to streamline and control its semiconductor manufacturing.

By introducing near real-time alerting for individual manufacturing processes, employees can automatically compare key parameters against pre-defined thresholds, statistical rules, and optimal patterns uncovered through machine learning and AI methods. Using TIBCO Spotfire Automation Services software, HSC receives automatically generated alerts as soon as a process falls outside of acceptable parameter bands, notifying manufacturing personnel that something required their attention. Personnel can then readily access data to see precisely which variable may have caused the problem. Once these cause-and-effect relationships are identified, teams take action to prevent process defects from happening again.

Because of TIBCO innovations, HSC has reported saving $300,000 a month in optimized resource consumption. “We now have visibility into processes that were very difficult to see before and can pursue answers to the really complex questions that enable us to optimally tune each process,” says Kevin Britton, program manager at Hemlock Semiconductor Operations.

Don’t Let Anomalies Impact Your Performance

Anomaly detection—and the critical insights it provides—can save you time, money, and effort. TIBCO has deep domain expertise in implementing and improving anomaly detection processes across a wide range of industries. Capabilities that were once only possible in semiconductor manufacturing are now adopted at scale in other industries.

With mature anomaly detection capabilities, you can achieve substantial operational cost reductions and product quality improvements.
Click To Tweet

Unleash the power of your data by reading this whitepaper about Anomaly Detection in Manufacturing.

The post How Anomaly Detection Can Save Your Manufacturing Company Time and Money first appeared on The TIBCO Blog.