Change your cover photo
Upload
Change your cover photo
Hello!! I'm Rithiga V. Currently pursuing BE. Electronics and Communication Engineering. I'm more passionate about Machine Learning and the Internet of Things.
This user account status is Approved

This user has not added any information to their profile yet.

AI-BASED AIR QUALITY MONITORING AND SUGGESTION SYSTEM
Maixduino Kit for AI+IoT

PROBLEM STATEMENT:
Currently, urban areas face significant challenges related to air pollution, with harmful gases such as particulate matter (PM), carbon monoxide (CO), sulfur dioxide (SO₂), nitrogen dioxide (NO₂), ozone (O₃), and volatile organic compounds (VOCs) adversely affecting public health. Traditional air quality monitoring systems often lack the granularity and real-time capabilities needed to provide timely and accurate information. Furthermore, they do not typically account for the varying susceptibilities of different age groups to air pollution, leading to generalized health advisories that may not adequately protect vulnerable populations such as children and the elderly. This gap in effective monitoring and targeted health recommendations underscores the need for a more sophisticated solution that leverages modern technologies such as AI and machine learning to deliver precise, actionable insights into air quality and its health impacts.

PROPOSED SOLUTION:
Our proposed solution involves developing a hardware device using the Sipeed Maixduino board equipped with sensors to read pollutant values. These readings are updated in real time to an Excel sheet. A pre-trained CNN or random forest regression model predicts the AQI index from these values, which is also logged in the Excel sheet. Based on the AQI range, the system provides health suggestions, displayed on the hardware's screen. This setup offers an integrated, portable, and user-friendly solution for real-time air quality monitoring and guidance, enhancing public awareness and promoting proactive health measures.

PROPOSED METHODOLOGY:
1) BOARD WE CHOOSE:
Sipeed Maixduino Board
Specifications:
1. SIPEED MaixDuino is an Arduino-compatible development board based on our M1 module (main controller: Kendryte K210).
2.Kendryte supports CNN and Random forest Regression.
3.The Sipeed MaixDuino can be powered via a USB Type-C connection (5V) or a single-cell Li-Po/Li-Ion battery (3.7V).
The power input range is 3.3V to 5V.

2). SENSORS INTEGRATION
*Nova PM Sensor SDS011: This high-precision laser PM2.5 air quality detection sensor measures PM2.5 and PM10 levels, with a range of 0.0-999.9 µg/m³. It operates on a 5V power supply and has a life span of 2-3 years.

*MQ131-H (Ozone Gas Sensor): The MQ131-H detects ozone in the range of 10-1000 ppm. It requires a loop voltage of 5±0.1V and has a life span of 5 years.

*MQ-7 (CO Sensor): This CO sensor operates at a DC voltage of 5V and has a detection range of 10 to 1000 ppm. Its life span is 2-3 years.

*ME-NO2: The ME-NO2 sensor detects nitrogen dioxide in the range of 0-20 ppm and operates within a temperature range of -20°C to 50°C. It has a life span of 2 years and requires a 5V operating voltage.

*AD 300 RO4A CIT (Sulphur Dioxide): This sulphur dioxide sensor measures SO2 levels from 0.1 ppm to 20 ppm. It functions in temperatures ranging from -40°C to 55°C and has a life span of 4 years, operating at 3V.

3). GOOGLE SHEET INTEGRATION
For an AQI calculator project using the Sipeed Maixduino board, we will be measuring the concentrations of six types of gases in the air. The Maixduino, equipped with sensors like MQ series or similar, will gather real-time air quality data. The Maixduino board features an ESP32 Wi-Fi module, which allows it to connect to the internet and facilitate data communication. To store this data in Google Sheets, you'll need to leverage the Google Sheets API. This involves setting up a Google Cloud project, enabling the Sheets API, and obtaining the necessary credentials for authentication.
In Maixduino firmware, we can write code to read sensor data and format it into an HTTP request. Using the Wi-Fi capability of the ESP32, we can send these HTTP requests to the Google Sheets API. Typically, this involves using the `HTTPSClient` library to manage secure connections. The HTTP request will include the sensor data in a JSON payload, specifying which cells in the spreadsheet to update. On the Google Sheets side, we can set up a script to handle the incoming data and populate the appropriate rows and columns.

4). INTEGRATION OF MACHINE LEARNING ALGORITHMS
The machine learning component leverages either a Convolutional Neural Network (CNN) or Random Forest Regression model, pre-trained on historical AQI data, to predict the current AQI based on real-time sensor inputs. These predictions are then recorded back into the Google Sheet. Based on the predicted AQI range, the system provides tailored suggestions displayed on the hardware, advising on outdoor activity levels and indicating which age groups are most at risk. This integration of hardware and ML models enables a comprehensive, real-time AQI monitoring solution that not only tracks air quality but also provides meaningful, user-specific advice, thereby promoting healthier lifestyle choices and ensuring public safety in varying air quality conditions.

5). DISPLAYING SUGGESTIONS
Integrating the display with the Maixduino involves connecting an OLED or LCD display using the appropriate communication protocol, such as I2C or SPI, and installing any necessary libraries for display control, like u8g2 for OLED displays. The next step is to retrieve AQI and suggestion data by using the Maixduino to send sensor data to a Flask server and receive the AQI value and suggestions, followed by parsing the response to extract AQI and suggestion strings. Finally, format the AQI value and suggestions for clear and readable display output.

NOVELTY
* Machine learning models excel at recognizing complex patterns in the data that traditional AQI calculators might miss, leading to more precise predictions. Additionally, the system can continuously improve its accuracy by learning from new data, refining its predictions, and adapting to changing conditions.
* It is user-friendly, with simple instructions and clear displays, ensuring that even individuals with limited education can easily set up and understand the air quality data.
* The novelty of this project lies in its development of a self-contained, hardware-based AQI monitoring system that does not rely on web-based platforms or servers. Unlike current systems that face issues like server downtime or limited accessibility, especially for uneducated individuals, our solution offers real-time monitoring and data analysis directly on the device.

NAGASRIVYSNAVI G
NIVETHITHA B
SUDHARSAN S
POOJA K R