RA TechED 2019 - CL05 Reduce Waste with Logixai
0 likes378 views
The document introduces FactoryTalk Analytics LogixAI, which uses artificial intelligence and predictive analytics to help manufacturers discover new insights from industrial data. It can perform descriptive, diagnostic, predictive, and prescriptive analytics to monitor equipment performance, identify issues, predict faults and quality problems, and recommend actions. The key benefit is it can build these predictive models without requiring data science expertise by automating the data preparation and model training process. The presentation also provides an overview of how FactoryTalk Analytics LogixAI works and its different modes of operation for anomaly detection and acting as a virtual sensor.
RA TechED 2019 - CL05 Reduce Waste with Logixai
- 1. Automate the discovery of new insights with FactoryTalk® Analytics™ LogixAI™
- 3. PUBLIC | TechEd | #ROKLive | Copyright ©2019 Rockwell Automation, Inc. 3 Agenda FactoryTalk® Analytics™ LogixAI™ Hands-On Lab 1 2
- 4. The discovery, interpretation and communication of meaningful patterns in data. Analytics relies on the application of statistics, computer programming and operations research to quantify performance. ANALYTICS
- 5. Scalable analytics The key differentiator for Rockwell Automation DEVICESYSTEMENTERPRISE Am I running ok? Why did a fault happen? I predict a fault will happen soon. What action helps to avoid the fault? Is Line 1 running ok? Why is Line 1 quality poor? I predict that Line 1 quality is moving out of tolerance. What action helps the operator to avoid poor quality? What plant performed the best? Why is site A throughput below plan? Will I meet plan today? Tomorrow? How can I change operations to improve profitability? Yield? Quality? DESCRIPTIVE PRESCRIPTIVE PREDICTIVEDIAGNOSTIC
- 6. Scalable analytics Where FactoryTalk® Analytics™ LogixAI™ fits into the ecosystem DEVICESYSTEMENTERPRISE Am I running ok? Why did a fault happen? I predict a fault will happen soon. Is Line 1 running ok? Why is Line 1 quality poor? I predict that Line 1 quality is moving out of tolerance. What plant performed the best? Why is site A throughput below plan? Will I meet plan today? Tomorrow? How can I change operations to improve profitability? Yield? Quality? + What action helps to avoid the fault? What action helps the operator to avoid poor quality? DESCRIPTIVE PRESCRIPTIVEPREDICTIVEDIAGNOSTIC
- 7. Building predictive models and change management 3% 5% 4% 9% 19% 60% Building training sets Other Refining algorithms Mining data for patterns Collecting data sets Cleaning and organizing data What data scientists spend the most time doing Project definition Data exploration Data preparation Model creation, testing & validation Scoring & deployment Model management Other 13% 18% 25% 23% 12% 9% 8% Groups spend this percentage of time on each phase in a predictive analytics project Based on 166 responses, adapted from Eckerson (2007)
- 8. FactoryTalk® Analytics™ LogixAI™ Descriptive | Diagnostic | Predictive | Prescriptive Configure Identify data Model Monitor Integrate DATA SCIENCE EXPERTISE NOT REQUIRED
- 9. How FactoryTalk® Analytics™ LogixAI™ works You select output variable that you want to model and monitor You select potential inputs or state variables that could impact your output variable FactoryTalk® Analytics™ LogixAI™ automated modeling cleanses the data and determines the relationships as a mathematical function You train the model using the User-Defined Tag Structure You set the model to calculate predictions Use your automation skills to integrate predictions Summary
- 10. PUBLIC | Copyright ©2019 Rockwell Automation, Inc. | 10 FactoryTalk® Analytics™ LogixAI™ Modes of operation Anomaly Detection Virtual Sensor Learn what's normal, find deviations from normal. Anomalous behavior that is not understood and cannot be alarmed upon through typical automation system capabilities can cause scrap, defects, poor quality or undesirable operations. A predictive model can be utilized to support understanding if an output variable could exceed acceptable operational boundaries. Pinpointing slow changing deviations that cannot be detected by other means enables proactive thinking to address conditions before production is impacted. Estimate values from incomplete data. A predictive model can act as a virtual sensor (aka soft sensor) by estimating output values. Having a predicted, estimated output value reduces the need for manual effort required to capture and record readings. A virtual sensor can reduce the need to instrument the operation with expensive or difficult to implement instruments or sensors. A virtual sensor can be used to check the reliability of a physical sensing device, where environmental conditions may cause degradation of the physical sensor devices’ reading (i.e. caused by heat, humidity, debris). Ultimately, virtual sensors can enable tight control or proactive deployment of resources when the estimated output variable is expected to reach an out of specification condition.
- 12. PUBLIC | TechEd | #ROKLive | Copyright ©2019 Rockwell Automation, Inc. 12 Share your feedback Please complete the session survey on the mobile app Select TechEd and login Use your email and last name that you used to register for the event. Click on Schedule on the main menu • Select the session you are attending • Click on the survey tab • Complete the survey and submit 2 3 Download the Events ROK mobile app 1