Quick Start 3: Find Patterns In Your Engine Data

APPROXIMATELY 15 MINUTES TO COMPLETE

Important – Make sure you have completed Quick Start 2 before starting Quick Start 3.

For Quick Start 3, you’ve already learned that Cylinder 3 may have issues, so we will look for irregular patterns of behavior.

The best way to do this is to compare operations of Cylinder 3 to the other cylinders. To do this, we’re going to use the compute calculations queries (similar to those created in Quick Start 2) as input to new analytics patterns.

First, we want to compare the real-time performance of Cylinder 3 to its history to see if it is performing outside of the historical averages for this class of cylinder.

Then, we will look for specific deviations or jumps in bearing and exhaust temperatures. Specifically, you will direct AXON Predict to determine if the bearing temperature or exhaust temperature for Cylinder 3 increase unexpectedly.

Analyzing Devices using Patterns

In this Quick Start, you want to determine how Cylinder 3 operates versus the other Cylinders.

To do this, you want to compare the bearing and exhaust temperatures for Cylinder 3 and the other Cylinder devices. You want to determine if the bearing temperature rises 2% within a 10 second window, and if the exhaust temperature rises above 750 degrees for at least 10 seconds at a time. To make these comparisons, you create 4 simple questions and then analyze discovered patterns.

Based on those results, you will see that you need to analyze the data another way to get more information. To do this, you will create more 2 more advanced questions to calculate data averages over time and use those averages to discover additional insight to the health of Cylinder 3.

VIDEO FOR THIS QUICK START

The steps

  1. Browse to your AXON Predict instance (if it's not already open). First, you will create a new dashboard.
  2. Click the Dashboard Options button  and select New Dashboard from the drop down list.
  3. Click Open Blank Dashboard. A new dashboard will open. You're ready to create your first compute question.
  4. Click the Ask a Question about your data button . The Questions panel displays:

    (You can see that it shows the questions you created as part of Quick Start 2.)

  5. Click Add. This opens the question wizard, which guides you through the process for creating analytics questions for your devices and data. Now, you will create the simple patterns for bearing temperature and exhaust temperature data.
  6. In the Patterns Gallery, click Simple and drag the rise/drop pattern template to the workspace.First, you create the two patterns that look for increases in bearing temperature.
  7. Configure the question as follows:
    1. Select the Cylinder 3 device group you created in Quick Start 1.
    2. Under Present time, type 10, and press Enter. (This is set to seconds by default.)
    3. Select BearingTemp as the sensor.
    4. Double-click the rise value, type 2, and press Enter.
    5. Type a descriptive name for the question, like Cyl3_BearingTemp_Rise.
    6. Click Create.
    7. Click Reset.
    (Note: For these Quick Starts, all questions you create need to persist data. You leave the default selection for "Save history for this pattern.")

    When you are finished, the question will look similar to the following:

  8. From Patterns Gallery, select and drag the rise/drop pattern template to the workspace and configure it as follows:
    1. Select the 7 Cylinders group you created in Quick Start 2.
    2. Under Present time, type 10 and press Enter. (This is set to seconds by default.)
    3. Select BearingTemp as the sensor.
    4. Double-click the rise value, type 2, and press Enter.
    5. Type a descriptive name for the question, like 7Cyl_BearingTemp_Rise.
    6. Click Create.
    7. Click Reset.
    Now create the two questions that detect undesired high values for exhaust temperature for these cylinders.
  9. From Patterns Gallery, select and drag the outlier pattern template to the workspace and configure it as follows:
    1. Select the Cylinder3 device group.
    2. Under Present time, type 15 and press Enter. (This is set to seconds by default.)
    3. Select ExhaustTemp as the sensor.
    4. Double-click the outlier value, type 750, and press Enter.
    5. Click At Least, then type 10 to create "at least 10 sec".
    6. Type a descriptive name for the question, like Cyl3_ExhaustTemp_Outlier.
    7. Click Create.
    8. Click Reset.
    When you are finished, the question will look similar to the following:
  10. From Patterns Gallery, select and drag the outlier pattern template to the workspace and configure it as follows:
    1. Select the 7 Cylinder device group.
    2. Under Present time, type 15 and press Enter. (This is set to seconds by default.)
    3. Select ExhaustTemp as the sensor.
    4. Double-click the outlier value, type 750, and press Enter.
    5. Click At Least, then type 10 to create "at least 10 sec".
    6. Type a descriptive name for the question, like 7Cyl_ExhaustTemp_Outlier.
    7. Click Create.
    8. Click Reset.
    You've finished creating the simple patterns for bearing temperature and exhaust temperature. Now you will want to add them to the dashboard and get insight from them.

    Now, you will view the analytics calculated by AXON Predict and compare the results to get insight into your devices. You need to open the realtime dashboard and then add visualization components to the page.

  11. Select Realtime.

    If you see Multi Sensor (defaulted from Quick Start 1 dashboard), then you will remove it from this Quick Start 2 dashboard.

  12. If applicable, locate the Multi Sensor component in the dashboard and click "X" (in the top, right corner) to delete it.
  13. Click Add a Visualization, select Multi Sensor Explore and drag it to the dashboard. Your dashboard should have two empty Multi Sensor Explore components.You need to add raw sensor data for visualization before you add the corresponding questions.
  14. Click DISCOVER DEVICES and, from the displayed list of sensors and data, drag each of the 8 BearingTemp sensors to the first Multi Sensor Explore, one at a time. When done, you can look at the Legend to confirm the sensors have all been added as expected.
  15. Now drag each of the ExhaustTemp sensors to the second Multi Sensor Explore. You can show the Legend and confirm the sensors are all defined in the component.Now you will add the questions for visualization.
  16. Click Ask a Question about your data, and drag the two rise/drop bearing temperature questions you created to the first Multi Sensor Explore, one at a time. (This will be the component to which you added the BearingTemp sensors.)
  17. Then, drag both of the outlier exhaust temperature questions you created to the second Multi Sensor Explore (the component to which you added all the ExhaustTemp sensors).

    In both components, you see the historical data stream for the related sensor and time period. As matches for the configured patterns are detected, they appear in the related component.

    To get a better sense of the data and patterns, you can change the window time and time period. While reviewing the patterns, you see matches showing unexpected data that you want to investigate further.

    You want to review how the cylinders have operated historically on average. To do this, you will create two rise/drop compute "composite" patterns: one will operate against the average bearing temperature for Cylinder 3, and the other will operate against the average bearing temperature for the other seven cylinders.

  18. In the Questions panel, click Add. The question wizard appears. (Click Reset to clear the wizard if it shows content from the previous question you created.)
  19. Click Composite and drag the Rise/Drop compute pattern template to the workspace. The left side of the template is the compute configuration; the right side is the pattern configuration. The output from the compute will be used as input to the pattern. The default template appears as follows:

  20. Configure the left side (compute) as follows:
    1. Click calculator to open the equation builder.
    2. Type 10 (seconds) and press Enter.
    3. Select the Cylinder 3 device group and click Apply.
    4. Select to Group by Device.
    5. Click the equation box and create this equation using the keys: avg(BearingTemp).
    6. Click Create.
  21. Configure the right side (pattern) as follows:
    1. Select min (minute) from the Calculate Every drop down and then type 1 and press Enter.
    2. Select % (percentage) from the RISE drop-down and then type 2 and press Enter.
    3. Type a descriptive pattern name, like Cyl3_BearingTemp_RiseCompute.
    4. Click Create.
    5. Click Reset. The configured question appears similar to the following:

      Now you create the Rise/Drop Compute for the 7 Cylinders group.

  22. Click Composite and drag the Rise/Drop compute pattern template to the workspace.
  23. Configure the left side (compute) as follows:
    1. Click Compute to open the equation builder.
    2. Type 10 seconds and press Enter.
    3. Select the 7 Cylinders group and click Apply.
    4. Click the equation box and create this equation with the keys: avg(BearingTemp).
    5. Click Create.
  24. Configure the right side (pattern) as follows:
    1. Select min (minute) from the Calculate Every dropdown and then type 1 and press Enter.
    2. Select % (percentage) from the RISE dropdown and then type 2 and press Enter.
    3. Type a pattern name like 7Cyl_BearingTemp_RiseCompute.
    4. Click Create.
    5. Click Reset.
    Now, you need to visualize detected patterns to get a better understanding of the devices, specifically of Cylinder 3. You want to compare the real-time bearing temperature for Cylinder 3 to historical bearing temperature values for Cylinder 3, to determine how the values varied over time. AXON Predicts look for pattern matches in historical bearing temperature for Cylinder 3. You will want to investigate the matches and get more details on other device behaviors during those match time periods.
  25. Click Realtime to open the realtime dashboard.
  26. Click Add a Visualization, select Performance Explore, and drag it to the dashboard. Then drag a second Performance Explore. Now you have two empty Performance Explore components in your dashboard.

    In the first component, you will compare the real-time BearingTemp data for Cylinder 3 to the patterns detected for the average of Cylinder 3 bearing temperature (historical data).

  27. Click DISCOVER DEVICES and, under SENSORS, select and drag the BearingTemp sensor for Cylinder 3 to the left side of the component.
  28. Click Ask a Question of your data to see the questions panel with the questions you've created.
  29. Drag the avgTemp_Cylinder3 question (created in Quick Start 2) to the right side of the Performance Explore. (Note that this will automatically drag the corresponding sensor, BearingTemp sensor, to that side of the Performance Explore as well.)
  30. Analyze and compare the data and you see. Now drag the Cyl3_BearingTemp_Rise compute question to the left side of the component (where you already added BearingTemp sensor). See matches.
  31. Now that you have a sense for Cylinder 3 by itself, you want to compare its behavior and data and matches to those of the other cylinders. You will do this in the second Performance Explore you added to the dashboard.
  32. Click the Ask a Question about your data button and drag the Cyl3_BearingTemp_RiseCompute to the left side of the component, and the 7Cyl_BearingTemp_RiseCompute to the right side of the component.
  33. Analyze the data from Cylinder 3 to see how it compares to the pattern matches detected in the Cylinder 3 BearingTemp historical averaged data.You can also compare the results of the two Rise/Drop Compute patterns: the pattern for Cylinder 3 versus the pattern for the other 7 cylinders.
  34. In the second Performance Explore component, drag Cyl3_BearingTemp_RiseCompute to the left side, and drag 7Cyl_BearingTemp_RiseCompute to the right side of the component.
  35. Analyze the matches from both patterns. Does AXON Predict detect matches for both patterns? Are matches detected at the same times and intervals for both questions? Any unexpected match behavior?

    As a last step in the Quick Start, you will save the dashboard so you can reuse it later.

  36. Click the Dashboard Options button  and select Save Dashboard from the drop down list.
  37. In the displayed Save Dashboard page, type a name for your dashboard and click Save.

QUICK START 3 IS COMPLETE!

This Quick Start was intended to show you how you can create and visualize different patterns that AXON Predict uses to detect issues or behaviors in your devices.