Cool Edge

temperature and CO2 level sensors kit
fleet management solution
mobile app for train manual control

Pitch Deck Link

https://docs.google.com/presentation/d/1HgKsuCO0TIvrAvmmfvHxJ9g411ua0w9tNwF_5QX-f5g/edit?usp=sharing

Summary

Heating, ventilation, and air conditioning systems, or HVAC has some limitations. It is costly, unintelligent, and can be dangerous in some scenarios. For examples, during summer it costs £31 / carriage day, it heats or cools empty carriages the same as full ones, and after a power cut, HVAC depletes train battery in 1 hour.

Our project leverages VeeaHub edge network nodes to manage on-carriage heating, ventilation and air conditioning systems, or HVAC, more efficiently.

Temperature and CO2 levels are gathered in realtime from the Tinkerforge sensors and passed into the streaming data analysis system we've created. This system monitors changes in carriage passenger counts through CO2 concentration, and applies a physical model to command the most efficient use of the available HVAC systems through a wiring unit we're developing with Angel Trains.

sensors kit

As the train's data connection comes and goes, the edge network uploads thinned data to our cloud-based fleet management solution at the operating company's head office. This shows the performance of all running trains, and highlights anomalies where preventative maintenance may be required.

fleet management dashboard

System Architecture

The edge-deployed system is written as a cluster of distributed python processes, using a shared event bus based on SQLite3 for persistence.

Sensor data is pushed into the system by a TinkerForge collection process, which communicates with the proprietary TinkerForge daemon and fetches new readings as required.

An HTTP server also provides a mobile web app allowing crew manual control over HVAC services in emergency situations. Data from this mobile app is pushed onto the same event bus.

mobile app for train manual control

An HVAC management process monitors all events passing the bus and decides how to control the HVAC unit. It attempts gradient temperature descent augmented by predictions of increasing heat load from high CO2 rate increase. It accepts emergency override messages from the mobile app.

These events can thinned and pushed to arbitrary clients, including the fleet management system.

One of the clients that listens to the events is the fleet management system, as shown above. It is a single page application written with React JS and is deployed as a static file to S3 bucket. This system can communicate with the sensors event by listening to the event data pushed by the edge system to the same or another bucket.