- October 22, 2019
October 22, 2019 -- FogHorn, a developer of edge computing software for commercial and industrial Internet of Things (IoT) solutions,announced the availability of Lightning Edge AI platform features, including tools and enhancements for operations technology (OT) professionals. The drag-and-drop analytic programming capabilities and visualization dashboards enable OT staff to derive insights from real-time data without the need for assistance from data science teams.
The Lightning edge computing platform brings intelligence to the edge, at or near the point where data originates, and facilitates analysis with the lowest latencies to improve operational outcomes. Artificial intelligence (AI) is enabled through built-in closed-loop edge-to-cloud machine learning, where FogHorn Lightning can detect drifts in model accuracies and automatically trigger cloud-based retraining with Google Cloud Platform (GCP) and now, Microsoft Azure IoT, and republish new models to the edge in an iterative fashion until the expected accuracy is reached.
This latest release of Lightning Edge AI makes meaningful improvements for the productivity of OT teams, including:
A visual programming tool, VEL Studio, creates analytic expressions that derive actionable insights from streaming control & sensor data. A drag-and-drop library of over 100 built-in code blocks lets OT professionals perform traditional data science tasks without the need for any programming skills. This functionality allows users to drag blocks to the workspace, fill in required parameters and connect the code blocks. These code blocks perform analytic functions, including; data cleansing and filtering, data collection and type conversion, event/pattern detection, signal processing and mathematical and statistical analysis. FogHorn also released OT centric blocks for manufacturing-specific use cases to create analytics including anomaly and failure condition detection.
VIZ Dashboards allows OT teams to visualize real-time data streams and monitor the efficiency and health of their environments. Dashboards are a user-defined canvas of widgets that visualize results of analytic expressions, display output of machine learning algorithms, validate sensors, and troubleshoot diagnostics of input sources. Based on how the user needs to employ each dashboard, widgets can be drawn to any size and include data visualizations, such as line graphs, bar charts, gauges, last state cards, maps, video feeds, images, and containers for nested dashboards.
Additional features of note:
- Closed-Loop ML: Closed loop machine learning from edge to cloud in order to maintain the health of deployed models. Now supported on Azure, along with Google Cloud Platform.
- EDA: An exploratory data analysis tool to help teams determine the value of data before investing in analytics and machine learning initiatives.
- Addition of ingestion agents such as OPC-DA.