Anomaly Fault Detection and Diagnostics

Get Started

Setup:

Create .env file

# postgres
POSTGRES_CONNECTION_STRING=<>

# neo4j
NEO4J_URL=

You can use docker to start postgres and neo4j services locally

This also starts a service to insert sample data into the databases.

docker compose up

How to run

poetry install
poetry run afdd

Creating Rules

A rule defines when certain conditions happen, like if sensors detect high CO2 levels. It uses Brick Schema to standardize and track these conditions, as well as COBie and BACnet.

Add rules by editing the rules.json

Structure of a rule

• rule_id: A unique identifier for each rule.
• name: Name of the rule.
• component_type: The Brick class that represents the equipment type
• sensor_types: List of Brick classes for the points that are needed for this rule
• description: A brief explanation of when the rule will trigger.
• condition:
  • equation: The mathematical expression for the rule condition, indicating thresholds for sensor readings.
  • metric: "average", "max", "min"
  • duration: The time period over which the sensor readings are averaged, in seconds.
  • sleep_time: The time period before the rule can be re-triggered, in seconds.

Example Rule

{
  "rule_id": 1,
  "name": "High CO2",
  "component_type": "IAQ_Sensor_Equipment",
  "sensor_types": ["CO2_Sensor"],
  "description": "Triggers when average CO2 level is between 1000 and 1500 ppm for 10 minutes",
  "condition": {
    "equation": "1000 < CO2_Sensor < 1500",
    "metric": "average",
    "duration": 600,
    "sleep_time": 1800,
    "severity": "high"
  }
}