Getting Started with Analytic Instances

If you have a SmartSignal analytic instance locked for editing, you can reset customized values back to the template values. You can reset customized values for all or some of the customized tag values.

• Reset customized values for some tags.
  1. Access a SmartSignal Analytic.
  2. In the Actions menu, select Changes > Select Values to Reset to Template.

     The Analytic Instance tag is placed in the Reset Tag Attributes to Template mode. In this mode, a check box appears for each tag that has been customized and for each attribute column that supports the reset feature.

  3. Select the row, column, or select individual cells for reset.
  4. Select Reset Selected Cells to save your changes.
  5. Select button to exit the reset to template mode of operation and return to standard editing.
     Note: Your changes will not be placed into production until you deploy the locked instance. You can cancel your changes by using the Discard Changes action instead of deploying.

     The Select Values to Reset to Template action is also available at the model level and the mode of operation is same as listed in this procedure.

• Reset customized values for all tags.
  1. Access a SmartSignal Analytic.
  2. In the Actions menu, select Changes > Select Values to Reset to Template.

     The Reset All Tag Attributes to Template window for the selected instance appears, displaying a set of options that allow you to control the tags and models that will be reset. By default, all tags and model instances will be selected for reset. The information icon next to each option shows the list of specific attributes that are eligible for being reset.

  3. Validate your selections and select Reset button to trigger the update of the analytic instance.
     When the action is completed a status indicator is displayed in the window to show the results.
  4. Select Close button to exit the action.
     Note: Your changes will not be placed into production until you deploy the locked instance. You can cancel your changes by using the Discard Changes action instead of deploying.

When you access a SmartSignal analytic instance, you can view the history of actions taken regarding the selected instance, and you can also export the analytic instance history to an .xls file.

1. Access a SmartSignal analytic.
2. In the Actions menu, select Changes > View Instance History.

   The Analytic Instance History window for the selected instance appears, displaying a table containing details about actions taken regarding the analytic instance during the dates specified in the date boxes. The history for the most recent changes appear by default.

   
   
   

   The View Instance History window displays details for a variety of actions that can be taken regarding an analytic instance, including the following actions:

   • Deploying or discarding changes
   • Synchronizing an instance with a template version
   • Activating or inactivating an instance
   • Locking an instance for editing
   • Importing an instance from a SmartSignal Classic transfer file
3. Optional: If you want to view the history for a different date range, in the date boxes, enter or select the dates for which you want to view the history, and then select Get History.
   Note: If you are using Internet Explorer 11, instead of the calendar functionality that appears in other browsers, a drop-down list box appears in which you can select one of the following values:

   • 7 days
   • 30 days
   • 180 days
   • 365 days
   • All
4. Optional: If you want to specify which columns appear in the window, select , and then, in the Column Chooser window, modify the selections as needed.
5. Optional: If you want to export an .xls file containing the data displayed in the Analytic Instance History table, select .

When you access a SmartSignal analytic instance, you can copy it to create another instance. You can use the instance copy to try out different configurations using the same input as the original.

1. Access a SmartSignal Analytic.
2. In the Actions menu, select Changes > Copy Analytic.

   The Copy Analytic window appears with the analytic Name box auto-populated with the selected analytic name with "_copy" appended in the end.

3. Optional: You can change the name and description of the copied analytic instance.
4. Optional: You can select the Copy Training Data checkbox to include the training data in the copy of the analytic instance.

When you access a SmartSignal analytic instance that is not locked for editing and you have the permission to modify the instance, you can delete the selected instance.

1. Access a SmartSignal analytic.
2. In the Actions menu, select Changes > Delete Analytic.
   The Delete Analytic window for the selected instance appears, prompting you to confirm if you want to delete the analytic instance.
3. Select Delete Analytic.

1. Lock for editing the analytic instance that you want to modify.
2. In the Actions menu, select Settings > Edit Sensor Health Configuration.
   The Sensor Health Configuration window appears.
3. Enter values as described in the following table.

   Parameter	Description
   SHM Enabled	Select this option to enable Sensor Health Monitoring for the analytic instance. The Reward Window parameter determines the duration, in hours, of good data required to restore the sensor health state.
   Alert Suppression Enabled	Select this option to enable Alert Suppression for the analytic instance.
   NAN	Select this option if you want sensor health to be checked based on whether sensor providing a NAN value (Not a Number) will be considered. The Penalty Window (Hours) parameter determines the number of hours of persistent NANs required to trigger a bad sensor health state.
   Outlier	Select this option if you want sensor health to be checked based on whether the sensor values are significantly higher or lower than normal. The Penalty Window (Hours) parameter determines the number of hours of persistent outliers required to trigger a bad sensor health state.
   Flatline	Select this option if you want sensor health to be checked based on whether the sensor values are completely the same. The Penalty Window (Hours) parameter determines the number of hours of persistent flatline required to trigger a bad sensor health state.
   Paintbrush	Select this option if you want sensor health to be checked based on whether the sensor values are significantly different alternating values (erratic). The Penalty Window (Hours) parameter determines the number of hours of persistent paintbrush required to trigger a bad sensor health state.

1. Lock for editing the analytic instance that you want to modify.
2. In the table, in the row containing the tag for which you want to configure the Sensor Health Monitoring parameters, in the SHM Configuration column, select .
   The SHM Configuration window appears.
3. Enter values as described in the following table.

   Parameter	Description
   SHM Enabled	Select this option to enable Sensor Health Monitoring for the analytic instance tag. The Reward Window parameter determines the duration, in hours, of good data required to restore the sensor health state.
   NAN	Select this option if you want sensor health to be checked based on whether sensor providing a NAN value (Not a Number) will be considered. The Penalty Window (Hours) parameter determines the number of hours of persistent NANs required to trigger a bad sensor health state.
   Outlier	Select this option if you want sensor health to be checked based on whether the sensor values are significantly higher or lower than normal. The Penalty Window (Hours) parameter determines the number of hours of persistent outliers required to trigger a bad sensor health state.
   Flatline	Select this option if you want sensor health to be checked based on whether the sensor values are completely the same. The Penalty Window (Hours) parameter determines the number of hours of persistent flatline required to trigger a bad sensor health state.
   Paintbrush	Select this option if you want sensor health to be checked based on whether the sensor values are significantly different alternating values (erratic). The Penalty Window (Hours) parameter determines the number of hours of persistent paintbrush required to trigger a bad sensor health state.

In the SmartSignal Maintenance module, you can generate a Sensor Health Monitoring report to view a timeline of Sensor Health Monitoring behavior. This provides insight into the initiating cause of the sensor to go into SHM Alert state, and then shows the length of the Sensor’s SHM Alert.

1. Access a SmartSignal analytic.
2. In the Actions menu, select Health > Sensor Health Report.

   If there is no timeline present, this means no sensor has initiated an SHM Advisory event. Adjust the custom date range or choose a present date range from the drop down menu

   

   A timeline report will be generated with the following UI capabilities:

   Report Parameter	Description
   Source Tag	Default tag view will be Tag Alias. Turn Source Tag to ON to view source tag instead
   Sort By	Allows user to sort by Impacted Tags (default), Tag Alias Alphabetically, or Blueprint Ordinal Position
   	Zoom in y axis, showing less sensors
   	Zoom out y axis, showing more sensors
   	Exports current view as PNG/PDF/JPEG/SVG/GIF File
   	Switch to Grid view. Enables capability to filter/sort by Source Tag, Start Date, End Date, Outlier %, Paintbrush %, Faltline %, and NaN %. Also enables export grid view to xlsx.
   	Resets to default view

In the SmartSignal Maintenance module, you can generate a report to indicate if the sensors are not working as expected or producing any bad quality data.

1. Access a SmartSignal analytic.
2. In the Actions menu, select Health > Sensor Health Report.

   If there are no existing reports, then the Generate Sensor Health Report window appears for you to specify the report parameters.

   

   The report will be generated based on the following parameters:

   Report Parameter	Description
   DATE RANGE	The report uses the tag data that has been processed by the analytic (i.e., observations) over the specified date range.
   % NAN	The percentage tag values that are NAN (i.e., Not A Number) that can be tolerated before it is added to the Watch list.
   % FLATLINE	The percentage tag values that are flat-lined (i.e., is not changing) that can be tolerated before it is added to the Watch list.
   % OUTLIER	The percentage of tag values that are outside actual high/low limits that can be tolerated before it is added to the Watch list.

3. Select Generate.

   The job starts and a progress bar appears that provides the status of the steps performed in the report generation process. The duration of the job depends on the date range specified and the number of tags that are active. After the report is generated, the report is displayed in a table that is divided into three rating sections (i.e., Watch, Warn, Good). You can change the sequence of the rating with the help of the Rating button. You can also expand and collapse each section by selecting the Rating bar.

   

   The generated report contains the following columns.

   Column Name	Description
   Tag	The tag alias name.
   Source Tag	The time series link of the tag.
   Description	The description of the tag.
   Units	The units of measurement for the source data of the tag.
   Confidence	A high percentage indicates confidence in the quality of the tag. A lower percentage indicates that the tag may have quality issues. The percentage is calculated using a combination of the following: The thresholds given for percentage of NaN, Flat-line, and Outlier. The tag's estimates are closely matching the sensor values (for the tags that are modeled).
   Within 5% of Actual	The percentage of tag values, where the estimated value is within 5% of the actual value.
   % NaN	The percentage tag values that are NaN (i.e., Not A Number).
   % Flatline	The percentage tag values that are flat-lined (i.e., is not changing).
   % Outlier	The percentage of tag values that are outside actual high/low limits.

4. Select to export an .xls file containing the data displayed in the report.
5. Select Close after the review is complete.

In the SmartSignal Maintenance module, you can generate a report that indicates how well the diagnostic rules defined in the template are being covered by this instance.

1. Access a SmartSignal analytic.
2. In the Actions menu, select Health > Diagnostic Coverage Report.

   The Diagnostic Coverage Report window appears for you to specify the report parameters.

   
   
   

   The report will be generated based on the following parameters:

   Report Parameter	Description
   Ignore inactive models	By default, the report will not include inactive models when determining rule coverage by excluding the inactive models in report. You can clear this check box to see the coverage using all models defined in the template.

   Note: If the analytic instance is locked for editing, the report will be generated based on the configuration details of the locked instance.

   
   
   

   Tip: If you want to generate the report based on the configuration details of the deployed instance, select the Show Deployed Report check box.

   
   
   
3. Select Generate.

   The job starts and a progress bar appears that provides the status of the steps performed in the report generation process. After the report is generated, the report is displayed in a table that is divided into three sections (i.e., Fully Covered, Partially Covered, Not Covered). You can use the check box above each section to filter the diagnostic rules that appear in the grid below. You can also expand and collapse each diagnostic rule to see the underlying contributing tags.

   

   The generated report contains the following columns.

   Column Name	Description
   Diagnostic Rule Name	The name of the diagnostic rule as defined in the corresponding analytic template.
   Tag Alias	The tag alias name.
   P1 – P5	Visual indication of each priority level configured in the rule and whether it is covered by the tag and model configuration.
   View	Press the calculator icon to see the full diagnostic rule expression.

4. Select the row to expand the grid to see the tags that contribute to diagnostic rule.
   Each tag has a corresponding icon that signifies the priorities of the tag contributes and whether the tag is active for that priority.

   

5. Select to export an .xls file containing the data displayed in the report.
6. Select Close after the review is complete.

You can clean the training data to reduce its disk space usage and to improve the performance of the SmartSignal Maintenance module.

1. Access a SmartSignal analytic.
2. In the Actions menu, select Model Training > Clean Training Data.
   The Clean Training Data window appears.
3. Modify the percentage to which you want to reduce the size of the training data.
4. Select Continue.
   The job configured to clean the training data starts and a window appears to display the progress of the job.

In the SmartSignal Maintenance module, you can modify the settings associated with an analytic instance, including the instance name, description, and data collection settings.

1. Lock for editing the analytic that you want to modify.
2. In the Actions menu, select Settings > Edit Settings.

   The Edit Settings window appears.

3. As needed, modify values in the available sections.

   Setting	Description
   Active	Switch this toggle to determine whether the analytic instance is active (that is, operational).
   Name	In this box, you can modify the name of the analytic instance.
   Description	In this box, you can modify the description of the analytic instance.
   Collection Type	In this drop-down list box, you can select the data collection type that you want to use for this analytic instance. SmartSignal supports four modes of data collection: Snapshot This collection type ingests the most recent data for all tags in the analytic instance. Use this collection type if you have a continuous, uninterrupted flow of data. This type is not recommended when data is ingested irregularly or in batches. Batch This collection type ingests data at the collection interval when readings for all tags in the analytic instance have the same time stamp. Use this collection type if your data is time aligned (that is, the readings for your tags for a collection interval have the same time stamp) and readings are ingested together. This type is not recommended if the readings across all tags used in the analytic instance are not aligned to the same time stamp. Sample and Hold Like the Snapshot collection type, this type uses the most recent data for all tags used in the analytic instance. However, unlike Snapshot, Sample and Hold also supports data that is ingested irregularly or in batches. Use this collection type or the Interpolated collection type in most cases. This type is not recommended if precisely time-aligned tag readings are required. In the Sample and Hold collection mode, the maximum difference between two tag readings can be up to one hour. If the most recent reading value before the specified collection time is at least one hour old, no value (a NaN) is returned. Interpolated This collection type returns a linearly interpolated value between two tag readings if no reading value is available at the specified collection time. Otherwise, if a reading value is available at the specified collection time, that value is returned. However, if the most recent reading value before the specified collection time is at least one hour old, no value (a NaN) is returned. Although the Interpolated collection mode is recommended for tags used in a SmartSignal model, it is not recommended for tags with quantized outputs (that is, outputs that are confined to a relatively small set of discrete, or distinct, values).
   Collection Interval	In this box, you can modify the interval at which data is collected for all tags used in the analytic instance.

4. Select Close.
5. When you have finished making changes, deploy the analytic instance.

You can export a SmartSignal analytic from APM to a file format that is compatible for importing back to your SmartSignal Classic product.

1. Access a SmartSignal analytic.
2. In the Actions menu, select Import & Export > Export.
   An .xfr file is created and downloaded based on your browser download settings. This file contains the data for the selected analytic.

In the SmartSignal Maintenance module, you can access the constants associated with a deployment.

1. Access SmartSignal Maintenance, and then select a deployment.
2. In the Actions box, select Settings > View Constants.
   The View Constants window appears.

In the SmartSignal Maintenance module, you can validate that the Asset Tags that have been associated with the analytic instance are without issue.

1. Access SmartSignal Maintenance, and then select a deployment.
2. In the Actions box, select Health > Validate with Asset Tags.
   If there are no issues found with the current tag configuration, the Validate with Asset Tags window appears.
   
   

   If one or more of the following conditions are identified:

   • A time series link in analytic data source is different from the link defined in the associated APM Asset tag.
   • A tag alias was not found in the input or output mapping of the analytic.
   • A tag alias was not found in the associated APM Asset tag list.
   • The UOM defined in the SmartSignal Analytic Instance was not compatible with the UOM defined in the associated APM Asset tag.
   • The UOM defined in the APM Asset tag is not support by the SmartSignal Analytic engine.

   The Validate with Asset Tags – Issues Found window appears.

   
   
   

   The table displayed in the image are described in the following table.

   Column Name	Description
   Tag Alias	Analytic instance tag alias. This name is defined in the analytic template.
   Source Tag	Analytic instance source tag name.
   Standard Units	Analytic instance standard units.
   Standard Units Category	Analytic instance Unit of Measure category
   APM Asset Tag Name	APM Asset Tag’s name in the I/O map
   APM Asset Tag Timeseries Link	APM Asset Tag’s Timeseries link
   APM Asset Tag UOM	APM Asset Tag’s Unit of Measure
   APM Asset Tag UOM Category	APM Asset’s Tag Unit of Measure category
   Auto Fix	True: The issue will be fixed when instance is deployed. False: The issue cannot be fixed automatically. User action required.

3. Select Close once you are done with the review.