How Pattern Protections Transforms Static Asset Data Targeted Action

In the ever-evolving world of asset performance management (APM), organizations are constantly seeking innovative solutions to proactively manage their equipment and systems. HxGN APM, Hexagon’s asset performance management platform, offers one integrated application to define and operationalize asset strategies, enabling businesses to assess risks, protect assets, monitor performance and act on emerging threats before they escalate. At Hexagon, we have our finger on the pulse of this continuous evolution and through listening to you, our customers, we continue to focus our development efforts on delivering value that best meets your asset management needs. In this blog post, we’ll go over the main improvement in our most recent quarterly release.

Practical Application in Manufacturing

Imagine a manufacturing plant that relies on high-speed rotating machines to maintain production efficiency. Unexpected downtime for these machines could lead to significant financial loss and missed production targets.

In the past, the plant struggled with delayed detection of issues, false alarms and a lack of historical context for monitoring asset health. By implementing HxGN APM’s enhanced Pattern Protections, the plant transformed its approach to asset performance management, detecting problems early, reducing false alarms and improving reliability.

These results were made possible by the latest enhancements to HxGN APM. Let’s explore these new features that are driving these improvements.

Enhanced Pattern Protections

We’ve made enhancements to Pattern Protections with the introduction of the Historical Pattern Type, Input Tolerance and Confidence Threshold. Pattern Protections take a significant step forward in enabling organizations to optimize their operations and reduce downtime. These enhancements are particularly valuable in industries where asset reliability and performance are mission-critical, such as manufacturing, energy, mining and logistics.

Historical Pattern Type

The Historical Pattern Type introduces a new way to evaluate data by leveraging both recent data points and a historical range of previous points. This approach is particularly beneficial for:

• • High-density data feeds where persistent patterns need to be recognized. 
  • Situations where patterns emerge more gradually, such as temperature trends or vibration anomalies.

In contrast to the already established Most Recent Pattern Type, which evaluates only the latest data points, the Historical Pattern Type provides a more comprehensive view of asset behavior. It ensures that subtle but critical patterns are not overlooked, enabling organizations to address issues before they escalate.

For example, in a mining operation, tracking exhaust temperatures over time can reveal early signs of engine problems. By using the Historical Pattern Type, engineers can capture and act on these patterns with greater accuracy.

Input Tolerance

Input Tolerance enables you to evaluate multiple data inputs as a group, so that you can compare relative data at the same time. For example, input readings may stagger by a few seconds, so you may want to set the tolerance to within one minute of each other.  Doing so ensures evaluation of data from a similar time period as well as greater accuracy.

Confidence Threshold

The Confidence Threshold is an advanced feature that fine-tunes how patterns are matched. It allows users to set a minimum confidence level for pattern recognition, ensuring that only patterns with a high likelihood of accuracy trigger pre-defined actions.

Here’s how it works:

• • Each pattern match is assigned a confidence level based on how closely the incoming data aligns with the trained pattern. 
  • The Confidence Threshold filters out matches where the confidence level is below the specified threshold, reducing false positives and ensuring that only actionable insights are surfaced.

For example, if a confidence threshold of 80% is set for a pattern, only pattern matches with a confidence level of 80% or higher will trigger an action such as a notification or advisory. This capability is particularly useful in complex systems where multiple patterns may overlap, allowing organizations to focus on the most critical conditions.

The Problem Solved

Going back to our real-life manufacturing example, in the past, the plant relied on manual monitoring and periodic maintenance checks to identify potential problems. However, this approach had several limitations:

• • Delayed Detection: Subtle variations were often missed because they were not immediately apparent during manual inspections. 
  • False Alarms: Simple threshold-based alerts can trigger false positives, leading to unnecessary inspections and downtime.
  • Lack of Historical Context: Without analyzing historical trends, it was difficult to distinguish between normal operational variations and early signs of failure.

The plant’s engineering team used Pattern Protections to create a proactive monitoring system. Here’s how the Historical Pattern Type, Input Tolerance and Confidence Threshold made a difference:

Historical Pattern Type for Persistent Pattern Recognition

The team used the Historical Pattern Type to evaluate both recent and past data for the bearings. 

By training the model on historical data, they identified a pattern where the data shows an early indication of bearing wear.

This persistent pattern would have been missed using only the Most Recent Pattern Type, as the condition changed gradually over time.

Input Tolerance for Relative Comparison

The team set input tolerances to ensure accurate groupings of input data for evaluation.

Confidence Threshold for Precision Monitoring

The team set a Confidence Threshold of 85% to ensure that only highly accurate pattern matches would trigger alerts. 

This eliminated false positives caused by temporary or insignificant fluctuations, allowing engineers to focus on genuine risks.

With Pattern Protections in place, the plant achieved the following benefits:

Early Detection of Bearing Failure

The Historical Pattern Type recognized the gradual degradation and provided early indication of failure. 

This early detection allowed the team to schedule maintenance and avoid disrupting production.

Reduced False Alarms

By setting a Confidence Threshold, the system filtered out low confidence matches that were not indicative of actual failure conditions. 

This reduced unnecessary inspections, saving time and resources.

Improved Decision-Making

The ability to simulate patterns before activation gave the team confidence in the reliability and accuracy of the model. 

They fine-tuned the protection to ensure it aligned with the plant’s unique operating conditions.

Increased Asset Reliability

By proactively addressing bearing issues, the plant reduced unplanned downtime and extended the lifespan of its equipment.

The combination of Historical Pattern Type, Input Tolerance and Confidence Threshold in Pattern Protections transformed the way the manufacturing plant managed its critical assets. These features provided the precision and flexibility needed to detect subtle patterns, reduce false positives and take timely action.

For organizations where asset performance is critical, such as manufacturing, energy or mining, these capabilities represent a game-changing approach to predictive maintenance. By leveraging machine learning and advanced analytics, Pattern Protections enable businesses to move from reactive to proactive asset management, ensuring operational excellence and cost efficiency.

Conclusion

With the introduction of the Historical Pattern Type, Input Tolerance and Confidence Threshold, enhanced pattern protections enable precise monitoring and early detection of potential issues, reducing downtime and improving decision-making. HxGN APM plays a pivotal role in these advancements, providing a comprehensive platform for organizations to leverage digital twins to optimize performance, mitigate risks and achieve operational excellence. By adopting these tools, businesses can unlock new possibilities for predictive maintenance, enhance asset reliability and drive sustainable growth in an increasingly competitive landscape.

To learn more about enhanced Pattern Protections and more, watch our latest webinar-here.