Easy to Use

No expertise in machine learning or software development is required. Subject matter experts can build condition recognition models in hours for the systems they operate and maintain.

Engineered for Real-World Time Series Data

Works in the presence of irregular and missing data. Supports sampled data and event data as well as batched and sliding windows. Advanced features include adaptive window sizing and frequency analysis based compression.

Quickly Move from Experimentation to Live Operation

Easily develop, deploy, and evolve the condition recognition models within the same environment. A simple browser-based UI supports the full lifecycle of use.

Powerful Embedded Machine Learning

Falkonry’s machine learning engine supports automated feature extraction, clustering and classification deployed in a highly scalable parallel processing environment. Unsupervised, supervised, and semi-supervised modes of learning are supported.

Deployment, Integration and Operation

Flexible and Easy Deployment

The Falkonry Server is delivered as self-contained SaaS application that can be easily deployed on premise or in public cloud environments from providers like Microsoft Azure, Oracle Cloud, Google Compute Engine, or Amazon Web Services.

Works with Existing Data Infrastructure and Applications

In a typical deployment, Falkonry connects to a time series repository like a process historian and augments it with real-time condition assessments based on existing data streams. A pre-built integration agent is available for OSIsoft PI, and other integrations are supported by an easy to use SDK and APIs.

Advanced Falkonry System Configurations

The Falkonry System is typically deployed via a single Falkonry Server that is responsible for learning models and applying those models to recognize patterns in live streaming data. There are some circumstances where it is advantageous to divide learning and live recognition responsibilities across a set of Falkonry Server instances. In such a configuration, models can be learned in one server and then be delivered to other servers for purposes of live recognition.

Multiple Deployment Choices

On-Premise Deployment

Bare Metal Servers

Virtual Servers

Cloud Deployment

Deploy in any cloud