AI Solar Prediction Model

This project provides a web-based interface to predict solar power generation using advanced AI techniques. The model uses weather data as input and provides hourly solar generation predictions.

---

Datsets likns:

Solar power-generation data (DATA.GO.KR) https://www.data.go.kr/data/15099650/fileData.do Period: 2022-04-01 ~ 2024-06-30 (39,360 ea, over 2 years) (Datetime) Year, Month, Date, hour, minutes, second (Location) Plant name, lat, lng (Generation) Power-generation value(Wh)

Weather data (KMA) https://data.kma.go.kr/data/grnd/selectAsosRltmList.do?pgmNo=36 Period: 2022-04-01 ~ 2024-06-30 (Temp) Temperature, Ground temperature (Solar) Irradiance(hr), Radiation(MJ/m2) (Etc) Humidity, Amount of cloud, Wind direction, Wind speed

Web App Details

Features

• Interactive Web Interface: Enter weather data, select prediction duration, and view results.
• Hourly Solar Power Forecasting: Realistic hourly generation patterns based on diurnal variations.
• Dynamic Graphs: Line plots visualizing predictions for durations ranging from 1 day to 2 months.
• Export Options: Download predictions as a CSV file.

Input Parameters

• Weather Inputs:
  • Temperature
  • Wind Speed
  • Wind Direction
  • Humidity
  • Air Pressure
  • Cloud Cover
  • Ground Temperature
• Prediction Duration:
  • 1 Day
  • 2 Days
  • 1 Week
  • 1 Month
  • Custom durations up to 2 months

Output

• Graph: Visualizes hourly solar power generation predictions.
• Downloadable CSV: Predictions available for export.

---

Plugin Details

Python 3.8 or later Flask PyTorch Numpy Pandas Scikit-learn Chart.js (for frontend visualization)

Features

• Seamless Integration: Easily integrate the AI model into other applications via API endpoints.
• Realistic Patterns: Hourly predictions account for sunrise, sunset, and peak daylight hours.

API Endpoints

1. Home: GET /
   Displays the home page.

2. Input Page: GET /input
   Form to input weather parameters and select prediction duration.

3. Prediction: POST /predict_input
   Returns hourly solar generation predictions as JSON or renders a graph on the results page.

---

Running the App

1. Clone the repository:

   git clone https://github.com/your-username/solar-prediction-model.git
   cd solar-prediction-model