top of page

Elipsa Documentation:
Fault Detection Page

The Fault Detection Page provides detailed insights into the status of various models and rules monitoring your equipment. It is designed to help you quickly identify and address issues that may be affecting system performance.

Navigation Pane

On the left side of the page is the Navigation Section, which includes:

​

  • Summary Link: At the top of the navigation, a Summary link directs you to the Summary FDD Charts for the selected equipment. This overview provides key insights into fault detection and diagnostics trends.

  • List of Models and Rules: Below the Summary link, you’ll find a comprehensive list of models and rules that the system is capable of running based on available data points:

    • Models: Predictive algorithms monitoring specific aspects of equipment performance.

    • Rules: Predefined conditions that check for deviations in expected behavior.

​

Status Indicators

​

Each model and rule is visually marked to indicate its current status:

​

  • No Border: No reported issues for this monitor. The system has not detected any anomalies or deviations.

  • Yellow Border: The monitor has identified a potential issue and has placed it on the Watch List. This indicates that the equipment is exhibiting signs of abnormal behavior that should be monitored but has not yet reached the fault threshold.

  • Red Border: The monitor has detected a significant issue that qualifies as a Fault. Immediate attention is recommended to diagnose and resolve the problem.

Details Pane

The Details Pane on the right side of the page provides in-depth information about the selected equipment:

​

  • Equipment Dropdown: At the top center, a gray dropdown allows users to switch between different pieces of equipment while remaining on the FDD page.

  • Admin Link to Equipment Edit: Next to the dropdown, administrators have access to an Equipment Edit link for updating equipment settings.

Summary View

Stats

The default display in the details pane is the Summary View, providing a high-level overview of detected issues:

​

  • Total Number of Models Reporting Issues: The count of models currently detecting performance deviations.

  • Average Confidence: The average confidence score across all reported issues, indicating the severity of detected faults.

  • Number of Rules Breached: Displays the total count of breached rules, highlighting specific conditions that have not been met.

Charts

  • Issues by Type: A breakdown of reported issues categorized by type, offering insights into the most common problems affecting equipment.

  • Issues by Day: A timeline chart showing the occurrence of issues by day, helping to identify patterns or trends over time.

FDD Details

The FDD Monitoring Details section explains the platform's real-time data analysis process:

​

  • As real-time data is received, the Elipsa platform evaluates it against the models and rules shown in the details pane.

  • Rules provide a binary result:

    • 0 indicates the rule's condition is not met (no issue detected).

    • 1 indicates the rule's condition is met (issue detected).

  • Models generate predictions with an associated confidence score, indicating the likelihood of an issue:

    • If the confidence score exceeds a certain threshold, the issue is placed on the Watch List.

    • If the confidence score indicates a significant deviation from normal operation, it is categorized as a Fault.

Models

Charts:

​

  1. Historical Data Chart:

    • Displays the relevant datapoints used by the model over time, providing context for the predictions.

  2. Predictions Chart:

    • Shows prediction results where:

      • 0 indicates normal operation.

      • 1 indicates a predicted issue.

      • -1 indicates no prediction run (e.g., insufficient data or a staging rule is in place, such as requiring equipment to be active).

  3. Confidence Chart:

    • Plots the confidence of predictions over time:

      • Confidence below 0 corresponds with normal predictions (0).

      • Confidence above 0 aligns with issue predictions (1).

      • Values range from -1 to 1, with higher values indicating more severe deviations from normal operation.

      • ​

This visualization helps users understand how closely current operations align with expected behavior and track the trend of issues over time.

FDD History

​

The FDD History section provides insights into the frequency and timing of detected issues:

​

  • A summary card displays:

    • The number of times an issue was predicted during the selected period.

    • The maximum confidence level recorded.

    • The average confidence of predictions.

  • Below the card, a detailed chart visualizes:

    • The timestamps of when issues occurred.

    • The confidence scores at each timestamp.

    • ​

This section helps users identify patterns and gain insights into when issues are most likely to occur, assisting in proactive maintenance and monitoring.

Diagnosis

​

  • The Diagnosis section features a blue button within the FDD History area.

  • When clicked, it generates an AI-driven response tailored to the specific issue being monitored.

  • The AI response provides a unique analysis, listing potential causes of the problem and offering recommendations for resolution.

Rules

In addition to model predictions, the Fault Detection Page includes a dedicated section for monitoring Rules. The key difference between models and rules lies in the results:

​

  • 0: The rule is satisfied, indicating no issue.

  • 1: The rule is not satisfied, indicating an issue has been detected.

  • ​

Charts Section

​

The Charts section for rules includes:

​

  1. Historical Data Chart: This chart provides a view of the historical data for the relevant datapoints, offering context on the performance over time.

  2. Rule Alerts Chart: This chart visualizes when rule alerts have occurred, showing instances where the rule was not satisfied (i.e., when a 1 was returned).​

​

Rules History

​

The Rules History section helps track rule breaches over time:

​

  • It displays the total number of faults detected during the selected period.

  • A table lists the exact timestamps of each fault occurrence, allowing users to see when specific rule violations happened.

bottom of page