Skip to content

Introduction to Falkonry Time Series Intelligence

Welcome to Falkonry Time Series Intelligence (TSI), an intelligent decision support system designed to help operations teams excel at asset maintenance and operational efficiency. Falkonry TSI aims to simplify the analysis of production data with time series intelligence tools.

In this user guide we will cover the key capabilities of TSI

Key Capabilities

Falkonry TSI provides four primary capabilities that enable the early identification and comprehension of invisible, complex, and concurrent operational challenges:

1. Calculations

It supports generating new time series signals from one or more existing signals using Python-based logic. It works on real-time and historical data, supports downsampling for efficiency, and applies consistently across signals. Outputs can be used in Patterns, Insights, and Rules for deeper analysis.

2. Rules

These are used to convert analog quantities like raw sensor values, anomaly scores, and condition labels into discrete events based on defined thresholds. Rules enable condition-based actions based on specific behaviors derived from any combination of signals, all through a no-code interface. They can be simple, multi-signal, or compound/nested.

3. Insights

This is an AI-based anomaly detection method for operational time series and metrics data. Insights automatically discover and highlight periods and signals of unusual behavior with a heatmap. It helps engineers expedite root cause analysis of anomalous and abnormal behaviors by revealing unexpected trends, waveforms, shapes, noise, levels, and duration.

4. Patterns

Falkonry Patterns focuses on multivariate detection. It extracts real-time understanding of conditions from patterns in time series data. Patterns supports semi-supervised learning, allowing for pattern discovery with few or no labeled examples. It provides confidence assessments for detected patterns and explanation scores for signal contribution.

In addition to these core AI capabilities, TSI offers:

Intuitive Visualization

Provides a fluid, accurate, and responsive view of high-resolution time series data, including flags, measurements, and alarms. Users can review data across multiple parameters and zoom into specific timeframes.

Reports

Enables users to create various charts for analysis and documentation without AI/ML modeling. Reports help users compare signals and time ranges to understand data behavior and capture expert knowledge. They can be organized into personal and group folders with unlimited nesting.

Signal Management

Allows users to organize large volumes of time series data for ease of use and machine learning. Signals can be structured and tagged using multiple flexible hierarchies, providing intuitive navigation and a "bird's eye view" for live monitoring.

Getting Started with Falkonry TSI

Access to Falkonry TSI is by invitation only. Once an account is created, the account owner receives an email to sign up. The initial steps typically involve:

  1. Setting Up Your Account: Accessing your account via the invitation link.
  2. Connecting a Data Source: Creating an inbound connection to bring your time series data into Falkonry.
  3. Managing Signals: Reviewing newly discovered "draft" signals, approving them, and adding metadata for context.
  4. Organizing Signals: Structuring your signals into signal trees for effective navigation and analysis.
  5. Applying AI: Utilizing Rules, Insights, or Patterns to start detecting anomalies, patterns, and generating alerts.

Note

Falkonry recommends starting with a small sub-system to develop an understanding of how the system works and build confidence before scaling up

Warning

Throughout this user guide, you will encounter screenshots of the Falkonry Time Series Intelligence (TSI) user interface (UI). While these visuals are highly representative, please note that the Falkonry TSI UI is continuously updated and improved. Therefore, the appearance of some elements in the live application may differ slightly from what is shown in these screenshots. They are primarily used here to convey concepts and give the reader a rough sense of what the UI may look like.