Insight - Predictive Analytics | April 2020

Posted: May 11, 2020

Online asset health and performance monitoring

Welcome to our AVEVA Insight blog, where we discuss all things Insight, including new features, updates, tips and tricks. Today I am very pleased to announce that a new service is now available for AVEVA Insight - Predictive Analytics.

Adding cloud-based Predictive Analytics to your AVEVA Insight solution provides increased visibility of asset performance and early detection of anomalies to everyone in your Insight environment. Helping to avoid unscheduled downtime by making proactive responses possible. In this blog we explore this exciting development in more detail:

• What is AVEVA Insight - Predictive Analytics?
• What operational challenges can it help with?
• Browser-based model builder and monitoring screens
• How do I enable it for my system?

What is AVEVA Insight - Predictive Analytics?

Our new Predictive Analytics capabilities extends your AVEVA Insight system so that you can not only consolidate your on-premises data into one central cloud-based location for operational continuity, but also expose the results of the predictive analytics engine to more people in your organization. 

Predictive Analytics uses Supervised Machine Learning to understand an asset's individual operating history and develops a series of normal operational profiles for each specific piece of equipment. It then compares the known operational profiles with real-time operating data to detect the subtle changes in system behavior that are often the early warning signs of pending equipment failure. An advanced alert manager and email notification provide near real-time updates of how well a plant or system is functioning.

Investigating alerts is made much simpler with Insight! The system identifies which sensors contribute to the anomalies, possible faults, and prescriptive information related to the type of asset. Each alert can be further document in a case management system, contributing to an extensive knowledge base on asset performance.  

What operational challenges can it help with?

Like traditional predictive analytics capabilities, AVEVA Insight - Predictive Analytics helps you to better understand your individual assets and how they are performing. This initial release uses supervised machine learning algorithms that learn normal operating parameters and provide early warning if running conditions deviate. This is particularly important for high-value assets as it gives you an opportunity to:

• Minimize unplanned downtime
• Better plan your maintenance windows
• Tune your asset strategy to a "just in time" repair policy
• Improve inventory management of spare parts

And because it's completely cloud-based:

• It can be easily deployed across your entire business
• The need for specialized IT resources to maintain the system is minimal
• You gain more complete and accurate analytics by leveraging the combined data sources in AVEVA Insight
• You have secure, remote, browser-based access to the Model Building and Monitoring Client so you don't have to be onsite to see the results
• Your predictive analytics results will be stored as tags within Insight so you can see them in charts, newsfeeds and alerts.

In the example below, you can see that the results are accessed by searching and navigating as per any other tag. Here we are looking at Compressor 1, and once we have located the tag, simply click through for more detailed information.

The new module also includes an advanced analysis application for performing Root Cause Analysis on why an asset is not performing as expected. The software is not specific to certain types of equipment or processes, and the core technology and algorithms have been used in a wide variety of equipment health and performance monitoring applications. Artificial Intelligence, advanced pattern recognition and sophisticated data mining techniques help to determine when a piece of equipment is performing poorly or is likely to fail. The early identification of equipment problems leads to reduced maintenance costs and increased availability, reliability, production quality and capacity.

Keeping Data Scientists happy with model builders and analytics
You can see that this new capability within Insight has really made predictive analytics accessible to everyone, but what about the detailed models? Well, we have developed dedicated screens where Data Science teams can build and monitor the algorithms. 

First we have the Predictive Analytics Model Builder where you can define models and projects. In the example here we are building from a predefined project for Compressor 1, and if you click through you can see that there is already data from AVEVA Insight in the model.

Next is the Predictive Analytics Monitoring screen. Here the subject matter experts, your Data Science team, can monitor the detailed results of the model. Users can view the anomaly alerts and then click through to the Overall Model Residual (OMR) for the relevant asset. Continuing with the Compressor 1 example, you can see that the model is currently in alert, and that we can click for more information on the different sensors that are relevant for this model.

This puts the detailed information on individual asset performance into the hands of your experts quickly and easily. With a unique asset signature established you are able to move from a reactive to proactive maintenance approach.

How do I enable it for my system?

If you would like more information, or want to enable Predictive Analytics for your AVEVA Insight subscription please contact your local distributor, or speak to our team at feedback@aveva.com

If you don't yet have an AVEVA Insight account, you can get started today with our 45-day free trial at https://insight.connect.aveva.com