AI-BASED WEATHER FORECASTING SYSTEM

ABSTRACT:
Weather forecasting plays a vital role in agriculture, disaster management, and daily planning. Traditional weather forecasting systems rely on large-scale meteorological infrastructure, which is costly and not suitable for localized prediction. This project presents an AI-based weather forecasting system using Raspberry Pi, designed to predict the probability of rainfall using real-time environmental data.
The system utilizes a DHT11 sensor to continuously measure temperature and humidity, which are critical parameters influencing rainfall. These real-time sensor readings are processed using a machine learning model trained on historical weather data. A Random Forest classifier is employed to estimate the probability of rainfall based on patterns learned from the dataset. The model outputs rainfall likelihood in percentage form, making the prediction more interpretable.
The system performs continuous real-time prediction of rainfall probability for the next one hour, which dynamically updates as new sensor data is collected. In addition, after collecting sufficient real sensor data, the system predicts the rainfall probability for the next 24 hours using machine learning–based analysis. The predictions are visualized through continuously updating graphs, where time is plotted against rainfall probability, and the results are also displayed on an LCD module for local monitoring.
This project demonstrates a low-cost, localized, and intelligent weather prediction solution using IoT and machine learning. The proposed system is suitable for applications in smart agriculture, environmental monitoring, and rural weather advisory systems, offering an efficient alternative to conventional forecasting methods.


OBJECTIVES:
The main objectives of the AI-Based Weather Forecasting System using Raspberry Pi are as follows:
1. To design and develop a low-cost weather forecasting system using Raspberry Pi and environmental sensors for localized weather monitoring.
2. To collect real-time temperature and humidity data using a DHT11 sensor and process the data continuously.
3. To implement a machine learning model trained on historical weather data to predict the probability of rainfall based on real-time sensor inputs.
4. To perform continuous short-term rainfall prediction by estimating the rainfall probability for the next one hour, dynamically updating the prediction with each new sensor reading.
5. To generate long-term rainfall forecasts by analyzing accumulated real sensor data and predicting the rainfall probability for the next 24 hours.
6. To visualize rainfall prediction results using time-based graphical plots that show rainfall probability variations over time.
7. To display real-time rainfall prediction output locally on an LCD module for easy monitoring without external devices.
8. To demonstrate the integration of IoT and Artificial Intelligence technologies for intelligent environmental monitoring and decision support.
9. To provide a scalable and extensible framework that can be enhanced with additional sensors or advanced machine learning models in the future.

Block Diagram