1.04.00

Flexible user notifications and integration with Rockwell and third-party systems

: FactoryTalk Analytics GuardianAI now supports flexible user notifications and integration with other systems. Insights such as anomalies, failure risks, and other diagnostic information can be written to a Logix-based controller (ControlLogix or CompactLogix) using a newly available custom UDT (User-defined Data Type) structure. A new stepper is added in the configuration as Output Controller to select the custom UDT tag defined from the controller tag list.

Once FactoryTalk Analytics GuardianAI findings are written to the UDT, any application or system capable of communicating with the controller can access and act on this data. This includes capabilities such as historizing the data in a historian solution or automatically generating work orders in both Rockwell and third-party Computerized Maintenance Management Systems (CMMS).

To accelerate value of this feature, use the provided HMI faceplates available in Rockwell Automation’s AI Device Library compatible with FactoryTalk Optix, FactoryTalk View ME/SE and Studio 5000 View Designer. These faceplates allow operators to view diagnostics from their FactoryTalk Analytics GuardianAI system directly from the HMI. For more information, see the AI Device Library Reference Manual DEVICE-RM700 (https://literature.rockwellautomation.com/idc/groups/literature/documents/rm/device-rm700_-en-p.pdf).

Controller Profiles

: Users can now add and manage multiple controller profiles, each with its respective CIP path, from a centralized location within the Source Controller Connection page.

A new stepper is added in the configuration as Output Controller to select the custom UDT tag defined from the controller tag list. This helps in providing notifications to the operator about the component and its associated deviations and failure risks.

Health Metrics

: Integrated directly into the application, health metrics capture key diagnostic data to support troubleshooting. These include:

Data buffers received per minute
Data buffers processed per minute
Data buffers rejected per minute
Deviation events per day
Model memory usage
Database space usage
Timestamp of the last data buffer received

This information is also written to the selected output controller's UDT tag, making it accessible to any system or application that communicates with the controller.

Baseline Quality Score

: A quality indicator is now available for each completed training frequency bin, showing the status as either Good or Uncertain. Clicking the indicator reveals detailed baseline quality metrics, including the normal cluster score and noise score, to help assess the quality of the training.

Minimum Sample

: A new optional field called Minimum Sample has been added under Advanced Settings. This value influences how deviations are detected. A higher value produces less distinct deviations with more occurrences, while a lower value produces more distinct deviations with fewer occurrences. For most use cases, the default value of 40 is recommended as the optimal setting.

The Machine Learning (ML) Engine algorithm has been upgraded within the application to deliver enhanced performance. As part of this update, the ML Engine is now deployed in a separate container during installation. FactoryTalk Analytics GuardianAI communicates with the ML Engine using the NATS.io message broker. Please note that upgrading to 1.4.0 will require a retrain due to the enhancements of the ML algorithm.

Timestamps for Raw data and Recording data are available now.

Improved the layouts of Deviation and Failure Risk charts. Now the charts display the data points at respective timestamps. Additionally, Multi Deviation and Failure Risk charts provide filters to display the required deviations or failure risks.

A toggle has been introduced to control the bearing monitoring. By default, they are optional. Enabling the toggle makes all field mandatory for bearing model inclusion.

Browsing support for array tags now enables selection of individual tags within arrays in State Definition and Monitoring On/Off pages.

Users can modify the email notification frequency from Advanced Setting as needed for every new Deviation and Failure Risk that FactoryTalk Analytics GuardianAI detects in a day.

PowerFlex 755 Series A and B may exhibit reduced performance when used with FactoryTalk Analytics GuardianAI.

Direct upgrade from FactoryTalk Analytics GuardianAI version 1.2.0 to version 1.4.0 is not supported. To successfully upgrade to version 1.4.0, users must first upgrade their application to version 1.3.0. Once it is successfully completed and verified, then the user can proceed with the upgrade to version 1.4.0. This staged upgrade ensures compatibility with intermediate changes introduced in version 1.3.0 and helps prevent potential issues during the installation process.

After upgrading to version 1.4.0, users must reinitiate training for all monitored components due to enhancements in the machine learning algorithm. If multiple states are available, all the states need to be trained again.

Avoid using fractional horsepower motors rated below ½ HP. The deviation/resolution in the output current from these types of motors is too low. As a result, FactoryTalk Analytics GuardianAI may not detect anomalies reliably in such motors. For more information on best practices and recommended specifications in using FactoryTalk Analytics GuardianAI, refer to the KB article: FactoryTalk Analytics GuardianAI Recommended Specifications and Best Practices.