With an aim to help decrease air pollution, Kutluhan Aktar, a self-taught full-stack developer, has built a custom weather station that can predict the air quality index (AQI) of a region based on the current weather conditions. The developer used various low-cost equipment to build the weather station that can calculate the ground-level ozone concentration to predict air pollution in the environment. Let’s take a look at the details.
Aktar, who recently pursued research papers on air quality and pollution, realized that current weather forecasting systems do not take ground-level ozone concentration into account when monitoring the weather of an environment. As a result, he went on to build a DIY weather station using relevant sensors and a Raspberry Pi 4 that can monitor the ground-level ozone concentration in an environment and predict the air quality index. You can watch it working on Aktar’s recent YouTube video.
Due to these issues, Aktar built a homemade weather station to monitor and predict air pollution. He used an Arduino Nano 33 board to collect a data set from his balcony. He then sent the data to a Raspberry Pi 4 that was set up indoors over Bluetooth. Combining the data with the local air quality index, Aktar trained a TensorFlow Kera H5-based artificial neural network to predict outside air quality based on the current weather conditions and the ozone data.
The neural network data was converted to a C array using a custom-built application that runs on the Arduino Nano board, which sits inside a 3D-printed weather station alongside the ozone sensor and anemometer.
With the help of the aforementioned, the weather station can show the data on a screen that is being collected by the sensors. These include ozone concentration, wind speed, temperature, atmospheric pressure, and altitude. Other than this, the weather station can also generate a graphical representation, showing its prediction of the air quality. It can display the AQI prediction as “Good,” “Moderate,” or “Unhealthy.”
