Objectives

In this lab you will learn how to:

  • Create an asset instance dashboard to see how an individual asset is performing
  • Create a summary dashboard that aggregates asset performance metrics across multiple assets of the same type
  • Detect anomalies by applying anomaly detection functions
  • Create alerts to prioritize anomalies
  • Monitor alerts across multiple assets in a summary dashboard you create
  • Troubleshoot to find the rood cause of a problem

Before you begin:
This lab requires that you have completed the pre-requisites required for all labs

Watch the 10 min video explaining why anomalies matter to business. Video

Create Simulated Robots to Monitor

In this exercise you quickly create some simulated robot operational data that you will use in your exercises in this Lab. You will use data from simulated industry robots as the assets you are responsible to monitor.

  1. Click Home to see the top level tasks you can do with Monitor. Home  
  2. Click Monitor Entities You will create a simulated set of robot assets (entities) that represent an assembly line of manufacturing robots.
  3. Click Create entity type button. Create an Entity Type  
  4. Select on Sample Robot and click on Nextbutton. Create a Robot Entity Type and simulated robot metrics  
  5. Edit the Entity Type name prepending your own initials Your_Initials_Robots_Type.Create a Robot Entity Type and simulated robot metrics  
  6. Click Submit button. Monitor will create a set of robots you can monitor with simulated random data.

Explore Instance Dashboards

An instance dashboard displays the operational metrics for a single asset, i.e. one specific industry robot. The data metrics of the robot are presented on the dashboard using cards.

  1. On the Data tab, expand Metric to see the metrics that Monitor has created for your Robots with simulated time series data. The outline shows the acc, load, speed and torque metrics in the data items outline. View recent (Less than 24 hours) Robot metrics across all robots  
  2. Expand Dimensions to see the classifications Monitor has created to filter and summarize all the Robots key performance metrics like by Manufacturer. Dimensions are meta data for describing an asset that can be used also be used to filter assets in tables or in functions. View Robot dimensions across all robots  
  3. Expand Metric (calculated) to see the functions used to create the simulated data like generating a random number to cause a fault like abnormal_stop_count. View Robot calculated metric functions
  4. Click the Dashboards tab to to see the list of Entities(Robots) in the Instance Dashboards table.
  5. Click one of the Robot instance dashboards that have been automatically created for you by Monitor. View Robot calculated metric functions  
  6. Click on the Back from calendar drop down to see historical performance metrics of this robot. View Robot calculated metric functions  
  7. Note the different types of cards in your dashboard. Monitor provides multiple types of cards to choose from when configuring a dashboard.
  8. In this exercise the Instance Dashboard for the Robots was automatically created for you. You can also export a summary dashboard and import it by managing instance dashboards as shown in the next exercise. Take a moment to see what cards were created for you in the instance dashboard.

Value Cards A value card shows a single or multiple entity metric value. A value may be attributed with a title, a label and a unit. The value may be given a precision and rules on threshold levels. A value card may be sized as a wide or tall card with big or small sized numbers.

Line Chart Cards A line chart card shows time-series data from a single or multiple entities as a chart with time and value axes. The chart may be given a title and labels on the axes. The line chart may be shown in full screen mode and is added a data table with the metrics values plotted in the chart. A line chart may also be overlaid with an Alert metric indicating any anomalies on the metric.

Table Cards A table card shows tabular data by columns. A table card is configured by referencing some data source for each column. The table may group data source values and present a count rather than each individual value. An Alerts Table is a preconfigured table presenting alert information. You will add one later for tabulating alerts.

Image cards An image card shows as a custom image with configured hotspots. Each hotspot is indicated at a position on the image with an icon and configured as a value card with entity metrics. The value card is shown when clicking on the hot spot. You will add one later for displaying an image of the robot.

Edit Instance Dashboard

In this exercise you will modify the layout of the individual robot instance dashboard. An asset instance dashboard is a configuration of cards, layout and the datasource metrics for a specific asset. One instance dashboard json configuration file is used for all robots of your Robot Entity type. You can export and dashboard configuration file adding new cards. You can also reuse and import dashboard configuration files from others.

  1. Click the gear icon to modify the layout of your robot instance dashboard. This dashboard template is shared by all robots in your Entity Type. Manage Instance Robot dashboard  
  2. Monitor provides a responsive user interface as the display size of the device or browser window changes you can reposition the cards. Change the card layout of your dashboard by dragging cards around with you mouse. View Robot instance dashboard  
  3. Once you are satisfied with the card layout, change the width of the browser window to a smaller size. Note the Dashboard Size of the layout has changed. Once again change the layout of your dashboard by dragging cards around with your mouse. View Robot instance dashboard   The dashboard configuration saves all the adjustments you make to the layout for each resize of your browser window.
  4. Click on Export button. The dashboard configuration file in a json format is saved to your local downloads directory on your computer.
  5. Click on Save to save the new dashboard layout.

Create a Summary Dashboard

In the previous exercise you modified an instance dashboard for each Robot. In this exercise you will create a Summary Dashboard. There is one summary dashboard presenting aggregated and filtered performance KPI across all robots of the type. A summary dashboard allows you to see the metrics for a filtered group of assets. "Show me all robots from a manufacturer and with a specific firmware.""

These filters are called dimensions in Monitor. Later in this Lab you will learn how to assign dimensions to the entities. Our simulated robots already have the appropriate dimensions assigned to each robot. Robot 73000 may have a manufacturer GHI Industries and Robot 73002 may have a manufacturer Rentech. A summary dashboard allows you filter assets and then compute the data metrics aggregations for the applied filter. For example, the Max, Min and Mean torque of all robots for the filtered scope of a selected manufacturer.

A summary dashboard uses time grains when computing the aggregations. Monitor can display four types of summary dashboards. You can choose by hour, day, week, month time grains. A summary dashboard supports the same cards as an instance dashboard. The next steps show you how to create a daily summary dashboard for all your robots.

  1. From the side menu, click Monitor
  2. From the Entity Types tab, search for your Entity Type by typing the name in the search field. Select your entity type and click View
  3. On the Dashboards tab Create summary dashboard  
  4. Click New Summary
  5. On the Definition page, assign the name robot_daily_summary to the dashboard Data filter and data aggregation using dimensions  
  6. Select the daily time grain
  7. Accept the default value for timestamp
  8. The simulated robot data that was created by Monitor also includes dimensions. Select load_rating as the lowest-level dimension and manufacturer as the parent dimension
  9. Select firmware as the mid-level dimension and manufacturer as the parent dimension
  10. Select manufacturer as the dimension and leave None as the parent since it is the top level filter and aggregation we want to see.
  11. Click Next and choose the max, min and mean for aggregation methods for all the number metrics. Assign dimensions for order of aggregations  
  12. Click Exportto save the dashboard json configuration. You will edit it later. You can also import dashboard configurations from this menu. Export the summary dashboard configuration json  
  13. Click Create summary to create the dashboard. Your dashboard should like something like the one below. Expoert the summary dashboard configuration json  
  14. Notice the summary dashboard form automatically added the aggregation functions for min, max and mean to your Metrics (calculated).
  15. Click on Data tab and expand the Metrics (calculated) section of the Data Items to see the daily min, max and mean calculated metrics. They should look something like the figure below. View the min, max and mean calculated metrics  
  16. Optionally repeat the steps above to create summary dashboards to create hourly, weekly and monthly aggregation performance metrics summaries of the all robots.

Add Metric Line Card

You need anomalies to be able to detect anomalies. Monitor provides simulated anomalies that you can apply to learn about anomaly detection. In this exercise you will:

  • Add a simulated anomaly metric named travel_time_anomaly.
  • Add a line card to display the travel_time_anomaly metric an instance dashboard.

The travel_time deviation that was generated earlier for your robot simulated data did not have anamolies in it. The AnomalyGeneratorExtremeValue function adds anomalies to the metric you choose. It abruptly increases or decreases the normal metric values for the metric you choose. In the steps below you will use this function to add a new calculated metric with anomalies named travel_time_anomaly You will add a line chart to display the anomalous data on your instance dashboard. Later you will use Maximo Asset Monitor anomaly detection functions to detect these anomalies. A SCADA system would likely not be able to detect these type of abrupt anomaly increase or decrease in metric values. SCADA systems typically only provide max or min business rule thresholds.

Add Simulated Anomaly Data for Travel Time

In this part of the exercise you will create simulated extreme anomalous values for travel time.
1. Search the function catalog for AnomalyGeneratorExtremeValue to create extreme simulated anomaly data.
Search Function Catalog for Anomaly Generator Extreme Value   2. Choose travel_time for the metric input_item. Extreme anomaly values will be added to this time series metric. Add anomaly metric   3. Set factor to a value of 100. This is how frequently an anomaly will occur. An anomaly will be generated after every 100 data points. 4. Set size to a value of 4. This is how extreme value will be. An anomaly will be 4 times the source travel_time metric. 5. Click Next to configure schedule to calculate historical values Calculating the last field to the name the metric being analyzed for anomalies. Toggle alert schedule and look back period view   6. Click Auto schedule on button to configure the functions schedule. 7. Set the output_metric to travel_time_anomaly. 8. Set the Calculating the last to 2 days to have Monitor generate 2 days of historical simulated anomalous data for travel_time_anomaly. 9. Click Create. After 15 minutes the function should be complete and the anomaly data will be available.

Add Simulated Anomaly Travel Time Line Card to Instance Dashboard

In this part of the exercise you will visualize the simulated travel_time_anomaly in a line card.

  1. Click on Dashboards tab to see the list instance dashboards
  2. Click on one of the robot instance dashboards in the table list.
  3. Click on Gear icon top right. Choose Edit dashboardsEdit Dashboard  
  4. Click on Export button top right. Instance dashboard json file should be saved in your browser downloads directory with a name of Industrial Robot summary-dashboard.json Export Instance Dashboard JSON 
  5. Open the exported instance dashboard JSON file using an Integrated Developer Environment like Atom or copy the file contents into an online Web JSON editor
  6. Read and learn about the JSON structure of Monitor dashboard template  
  7. Study the JSON file cards in the editor. Insert the travel_time line chart card json below into the "cards":[]. { "content": { "series": [ { "dataSourceId": "travel_time_anomaly", "label": "Travel Time" } ], "unit": "sec" }, "dataSource": { "attributes": [ { "attribute": "travel_time_anomaly", "id": "travel_time_anomaly" } ], "range": { "count": -24, "interval": "hour" }, "timeGrain": "day" }, "id": "travel_time_anomaly_line_card", "size": "LARGE", "title": "Travel Time with Anomaly", "type": "TIMESERIES" },
  8. Notice how the JSON nodes for "content": {"series": and "content": {"dataSourceId": don't have an aggregation "aggregator": "min", like other line metrics in summary dashboard JSON. As a result the travel_time_anomaly metric line chart data will display raw grain metric data. travel_time_anomaly will show sudden extreme anomalies  
  9. Notice how the "range": {"count": -24,"interval": "hour"} JSON node lets Monitor know to only display the last 24 hours of data.
  10. Make sure to change all values with card sizes of XLARGE to LARGE and XSMALLWIDEto SMALLWIDE, since these formats have been deprecated.
  11. In the IDE, save your changes to the instance dashboard JSON file.
  12. You can see the finished Industrial_Robot_Instance_Travel_Time.json  
  13. Return to the Monitor dasshboard editor and click on the Import button.
  14. Choose the file you updated with the line card click open to import your updated JSON file.
  15. Your instance dashboard should now look similar to the one below and have a travel_time_anomaly Card with some extreme anomalies like the one in the red rectangle. Robot Instance Dashboard.

Add Anomaly Detection

Maximo Asset Monitor includes models to detect anomalies in the function catalog. The anomaly detection functions can detect many types of anomaly patterns. These include:

  • Varying signal becomes flat line which can be caused by a defective or tampered with sensor.
  • Varying signal becomes a near vertical line which can be caused by a defective sensor or extreme changes to the system.
  • Sudden peak maximum or minimum which can be caused by operating equipment outside normal operating procedures.
  • Flat line becomes a varying signal which can be a system that has been running smoothly and then becomes out of tolerance or unstable.
  • No sensor data is available which can be caused by a defective sensor or network connectivity issues.
  • A predicted value doesn't come within the threshold based on correlated dependent target variables values.

In this exercise you will:

  • Choose an entity to analyze.
  • Choose a metric to analyze.
  • Add anomaly functions to your Robot Entity Type to score how likely an anomaly is occurring.
  • Add a line chart to visually compare if high anomaly model scores correlate with the robot travel_time_anomaly metric on each robot instance dashboard.
  • Add anomaly alert functions with score thresholds that corresponds to a level that indicates an anomaly for a robot metric.
  • Update a function with a new metric name or configuration values if needed.
  • Add alerts to a line chart to visually see where anomalies are occurring on the travel_time_anomaly metric.
  • Add alerts table to robot instance and summary dashboard to summarize anomaly alerts.

Before you begin:

Watch the 14 min video explaining what anomaly detection functions are available in Maximo Asset Monitor and what anomaly patterns are detected. Video

Watch the 19 min video explaining how to configure anomaly detection functions, alerts and dashboards. Video

Choose an Entity to Analyze

Imagine you are managing a fleet of delivery vehicles. One out of the five vehicles is a car instead of a scooter. If you wanted to detect anomalies for excessive miles per gallon performance you should analyze the miles per gallon metric for the car separately from the scooters. Add an EntityFilter function to isolate a function analysis to an individual entity.

  1. While editing your entity type, search the function catalog for EntityFilter function.
  2. On the Data tab, click the '+' button and search in the function catalog for EntityFilter. Add Data 
  3. On the function Configuration tab, enter the list of entity ids separated by a comma or 73000 for just one entity. Choose Entity 
  4. On the function Output tab, enter the name of they filtered entity entity_73000 for the output metric name. Name it entity_73000 

Choose a Metric to Analyze

Add the Filter function you can isolate your analysis to an individual entity and metric. Select the metrics to analyze from the filtered entities by adding Filter.

  1. While editing your entity type, search the function catalog for Filter function.
  2. On the Data tab, click the '+' button and search in the function catalog for Filter. Add Data 
  3. Choose entity_73000 output item from the previous function as source and expression shown below. Choose Filter 
  4. On the function Configuration tab, choose ${entity_73000} == True as the expression. Choose Filter 
  5. On the function Configuration tab, choose travel_time_anomaly as the metric to analyzed for filtered_sources.
  6. On the function Output tab, enter the name of they filtered entity travel_time_anomaly_entity_73000 for the output metric name.

Add Anomaly Functions

You will add an anomaly function that will provide a score of how likely the single input metric is having anomalies during the specified window. The directions that follow will use the metric travel_time_anomaly which will analyze all robots for that metric. If you only want to analyze one robot entity 73000, then select entity filtered metric created in the previous step travel_time_anomaly_entity_73000 instead.

  1. Search the function catalog for anomaly detection functions to detect anomalies for the metric travel_time_anomaly.
  2. On the Data tab, click the '+' button and search in the function catalog for Anomaly. Add Data 
  3. Select K-MeansAnomalyScore. Notice the different kinds of Anomaly Model functions included with Monitor. Click Select button. Choose K-MeansAnomalyScore Anomaly Model 
  4. Select the metric to score for anomalies travel_time_anomaly
  5. Anomaly detection functions require a window size which is the number of samples points to evaluate each time the model is scheduled to execute. Enter a windowsize of 12.
  6. Name the calculated metric travel_time_kmeans_score Add Data 
  7. Click Next to configure schedule, look back period and name the alert.
  8. Click Auto schedule on button to configure anomaly function scoring schedule. Toggle anomaly scoring schedule and look back period view  
  9. Set the anomaly function to execute every 15 minutes. The function will look for new data for travel_time_anomaly that have been added in the last 15 minutes and calculate alerts for those new data items in the windowsize. Configure anomaly scoring schedule and look back period  
  10. Set the look back period to 2 days. This will calculate historical scores looking back 2 days using the new configuration values with the historical metric values for travel_time_anomaly.
  11. Set the output metric name for the alert travel_time_kmeans_score
  12. Click Create to create the alert configuration. It will take 15 minutes for the alert to update. While waiting lets create anomaly functions for the other anomaly detection functions. See the guidance in the table below for scheduling anomaly functions.
  13. Repeat the above steps in this exercise adding, configuring and naming the anomaly scoring models in a similar way for the other models. Use the tool tip suggestions for setting the input arguments default values:
  14. Search for FFTbasedGeneralizedAnomalyScore2 in the function catalog. Configure and name it travel_time_fft_score
  15. Search for GeneralizedAnomalyScore in the function catalog. Configure and name it travel_time_ga_score
  16. Search for SaliencybasedGeneralizedAnomalyScore in the function catalog. Configure and name it travel_time_saliency_score
  17. Search for SpectralAnomalyScore in the function catalog. Configure and name it travel_time_spectral_score

Table - Suggested Scoring Schedule


Data Grain Frequency Sample Window Data Required Schedule Scoring
1 Day 12 24 Days
Noncritical – Once per 12 days

Critical - Once per day
1 Hour 12 24 Hours
Noncritical – Once per 12 days

Critical - Once per 1 hour
5 Minutes 12 2 Hours
Noncritical – Once per 60 mins

Critical - Once per 5 mins
1 Minute 12 1 Hour
Noncritical – Once per 12 mins

Critical - Once per 5 mins

Add a Multi Series Line Chart

Add a line chart** to visually compare if high anomaly model scores correlate with the robot travel_time_anomaly metric on each robot instance dashboard. In this exercise you will add a multi series line chart card to your instance dashboard that plots the anomaly model scores for time_travel_anomaly metric. You can then visually correlate which models are effective at detecting anomalies by seeing which model scores are high when the anomalies happen.

  1. From the instance dashboard click on the gear iconManage Instance Robot dashboard  
  2. Select Edit dashboard

  3. Add the following line chart card Json for travel_time_anomaly metric anomaly function scores to your instance dashboard.

    { "id": "card-anomaly-scores-line-timeseries", "dataSource": { "attributes": [ { "attribute": "travel_time_anomaly", "id": "travel_time_anomaly" }, { "attribute": "travel_time_spectral_score", "id": "travel_time_spectral_score" }, { "attribute": "travel_time_saliency_score", "id": "travel_time_saliency_score" }, { "attribute": "travel_time_ga_score", "id": "travel_time_ga_score" }, { "attribute": "travel_time_fft_score", "id": "travel_time_fft_score" }, { "attribute": "travel_time_kmeans_score", "id": "travel_time_kmeans_score" } ], "range": { "count": -24, "interval": "hour" } }, "content": { "series": [ { "dataSourceId": "travel_time_anomaly", "label": "Travel Time Deviation " },{ "dataSourceId": "travel_time_spectral_score", "label": "Spectral" }, { "dataSourceId": "travel_time_saliency_score", "label": "Saliency" }, { "dataSourceId": "travel_time_ga_score", "label": "Generalized" }, { "dataSourceId": "travel_time_fft_score", "label": "FFT" }, { "dataSourceId": "travel_time_kmeans_score", "label": "Kmeans" } ], "xLabel": "Time", "yLabel": "Score" }, "size": "MEDIUM", "title": "Anomaly Scores for Travel Time", "type": "TIMESERIES" },
    4. View the the anomaly model scores for time_travel_anomaly metric line chart card on each robot instance dashboard. Analyze the anomaly high scores for each model to see which ones happen when the anomaly does for time_travel_anomaly happens. It should look something like the card below Travel Time and Anomaly Scores  5. In the line chart click Spectral label so that you can see only the line chart of each score versus time_travel_anomaly metric value. Travel Time Spectral Anomaly Scores  6. Click on Spectral label again to hide that line. Select the Saliency label to see anomaly score line. Notice how the models score higher they there is an anomaly. Travel Time Saliency Anomaly Scores
    7. Study each instance dashboard to see where the anomaly scores are high and time_travel_anomaly metric appears to have an anomaly. Make note of the score levels that correlate with your Robots anomalies that were generated in the travel_time_anomaly metric. You will use this in the next exercise when you create alerts to notify you that anomalies have occurred. The table below has some suggested starting values:

Anomaly Score Upper Threshold
travel_time_spectral_score 99
travel_time_saliency_score 105
travel_time_ga_score 0.4
travel_time_fft_score 104
travel_time_kmeans_score 8

Add Anomaly Alerts

In this exercise you will create alerts for each anomaly function score identified in the previous exercise. You will set the alert thresholds to values of the visually correlated anomaly scores using the upper_threshold values from the table in Add Anomaly Functions As you learned earlier, alerts are a function in the Maximo Asset Monitor catalog that can be configured to notify you that certain asset conditions require attention. There are a several alert functions included in the catalog like Alert High, Alert Low and Alert Expression Filter. Like other functions in the catalog, these alerts can be scheduled to run every five minutes or less frequently. Add a Alert High Value alert for each anomaly model. Alerts can calculate historical values for a given function configuration.

  1. Click + button access the Function Catalog. Search function catalog for alerts  
  2. Search on Alert
  3. Select the Alert High Value function and then click Select button. Add Alert High Value  
  4. Configure the alert to trigger when the value travel_time_spectral_score reaches 99. An operator would be required to investigate each anomaly alert when they occur. It is important to avoid setting the alert upper_threshold value too high which could cause anomalies to go undetected. Vice versa avoid setting the alert upper_threshold value too low which could cause false alerts. False alerts are those alerts that triggered when there isn't an anomaly. Configure spectral alert expression  
  5. Set input_item to travel_time_spectral_score.
  6. Set upper_threshold value to 99.
  7. Set Severityvalue to High. You can add optionally add separate alerts for Medium and Low thresholds.
  8. Set Status value to New. Alerts don't have any order. They can start or finish an in any order. Users can then transition them to New Acknowledged, Validated and finally Resolved or Dismissed. Monitor does not provide a way to force specific state transition paths for alerts.
  9. Click Next to configure schedule, look back period and name the alert.
  10. Click Auto schedule button to configure schedule. Toggle alert schedule and look back period view  
  11. Set the alert to execute every 15 minutes. The alert will look for new data for travel_time_spectral_score that have been added in the last 15 minutes and calculate alerts for those new data items.Configure alert schedule and look back period  
  12. Leave the look back period set to 1 days. This will calculate alerts looking back 2 days using the historical values for travel_time_spectral_score.
  13. Set the output metric name for the alert travel_time_spectral_alert
  14. Click Create to create the alert configuration. It will take 15 minutes for the alert to update. While waiting lets create the alerts for the other anomaly model scores.
  15. Repeat the above steps in this exercise adding, configuring and naming alerts for each anomaly scoring model using the expression and alert names below.
Search Catalog for Model Upper Threshold Output Metric Alert Name
SpectralAnomalyScore2 99 travel_time_spectral_alert
SaliencybasedGeneralizedAnomalyScore 105 travel_time_saliency_alert
GeneralizedAnomalyScore 0.4 travel_time_ga_alert
FFTbasedGeneralizedAnomalyScore2 104 travel_time_fft_alert
KMeansAnomalyScore 8 travel_time_kmeans_alert

Update Functions

You can update functions with new a new metric name or configuration value. In this exercise you will review and correct any function names that are improperly spelt.

  1. Go to your entity type and click on the Data tab.
  2. Expand the Alert (calculated) section of the Data Items to see the alerts. They should look something like the figure below. View alert data items  
  3. Check the naming of each alert. If they are misnamed, click on the alert name in Data Items column.
  4. On the frame that opens, click the Configure. On the dialog that opens and Next buttons. Change the Output metric name to the correct spelling.

Add Alerts to Line Chart

In this exercises you will add the travel_time_anomaly alerts to the line card for the travel_time_anomaly on the robots instance dashboard. Start by exporting the robots instance dashboard JSON and adding the alerts additionalData JSON below to the travel_time_anomaly metric line chart so that you can see red dots in the areas where there likely is an anomaly.

  1. Edit the robot instance dashboard template JSON.
  2. Export and modify the JSON adding the highlighted JSON below in an IDE to the travel_time_anomaly card that you added earlier. 
        ,
        "additionalData": {
            "type": "alert",
            "dataFilter": {
                "name": [
                    "travel_time_spectral_alert",
                    "travel_time_saliency_alert",
                    "travel_time_ga_alert",
                    "travel_time_fft_alert",
                    "travel_time_kmeans_alert"
                ]
            }
        },
  1. The highlighted gray JSON below shows where you should insert the json above. Add Alerts to Line Card Save the JSON file in the IDE.
  2. Import the saved JSON file into the Monitor Instance Dashboard.
  3. Save the dashboard.
  4. View multiple robot instance dashboards. Analyze the travel_time_anomaly line chart looking for anomalies and alert red dots. When you find one, hoover over it with your mouse cursor. It should look something like the dashboard below Robot Instance Dashboard Json 

Add Alerts Table Card

In the previous exercise you visually represented the anomaly alerts as red points on the line chart. It's helpful to also see alerts organized in a table so that you can easily prioritize and act on them.

  1. Add a table card that displays the travel_time_anomaly_alerts by robots for the last 24 hours. Add the following card to your instance dashboard template json.
{
"id": "travel_time_anomaly_alerts",
"size": "XLARGE",
"title": "Robot alerts from the last 24 hours",
"type": "ALERT",
"dataSource": {
"range": {
  "count": -24,
  "interval": "hour"
},
"timeGrain": "input",
"type": "alert"
}
},

  1. You should now have the table in your summary dashboard showing the following Anomaly detection functions scoring. Alerts Table Card With Anomaly Alerts on Instance Dashboard  

  2. From the calendar control in the top right of the dashboard pick an earlier date to see the historical alerts for the robots on a specific date. Alerts Table Card Without Anomaly Alerts on Instance Dashboard  

Congratulations you have successfully create an asset instance dashboard and a summary dashboard that aggregates asset performance metrics across multiple assets. You also learned how to detect anomalies by applying anomaly detection functions and alerts at scale across all your enterprise assets.