This guide provides step-by-step instructions for using the FilantropiaSolar v1.0 production application.

# Navigate to project directory
cd FilantropiaSolar

# Activate virtual environment (if using one)
source venv/bin/activate  # On macOS/Linux
# or
venv\Scripts\activate  # On Windows

# Run the application
python main.py

When you run the application for the first time:

1. Data Loading: The system will automatically load:

   • Historical PV data from PV Plants Datasets.xlsx
   • Weather data from weather_files/Lisbon_weather.csv
   • This may take 1-2 minutes depending on data size

2. Model Training: Machine learning models will be trained:

   • Random Forest, Gradient Boosting, and Linear Regression models
   • Training happens automatically for each Lisbon installation
   • Progress is shown in the status label

3. Ready State: Once complete, the status will show "Data loaded successfully"

• Options: Lisbon_1, Lisbon_2, Lisbon_3, Lisbon_4
• Purpose: Each represents a different PV installation in Lisbon
• Recommendation: Start with Lisbon_1 for testing

• Format: YYYY-MM-DD (e.g., 2024-03-15)
• Options:
  • Past dates: Analyze historical performance
  • Today: Current day predictions using real-time weather
  • Future dates: Forecast based on weather predictions
• Tip: Click "Today" button to quickly set current date

• Default: 10.0 kWp
• Range: Any positive number
• Purpose: Scales energy production to your specific installation size
• Example: If you have a 5 kWp system, enter 5.0

The results display provides comprehensive information:

• Total Energy Production: Sum of all hourly predictions for the day
• Average Specific Energy: Energy per kWp installed capacity
• Average Ranking: Overall quality rating for the day

• Average Temperature: Daily temperature average
• Average Cloud Cover: Percentage cloud coverage
• Average Solar Radiation: Key parameter affecting solar production

• 24-hour view: Energy production for each hour
• Ranking: Quality rating (1-5) for each hour
• Format: "Hour XX: YY.YY kWh (Rank Z)"

• ±7 days: Shows 15 days total (7 before + target day + 7 after)
• Target marker: "← TARGET" indicates your selected date
• Comparison: Compare target day with surrounding dates

• Blue bars: Daily energy production
• Colored edges: Ranking quality (see legend)
• Red line: Temperature overlay
• Green dashed line: Target date marker
• "TARGET" label: Highlights selected date

• Red (Rank 1): Poor conditions (0.1-0.2 kWh/kWp)
• Orange (Rank 2): Fair conditions (0.2-0.4 kWh/kWp)
• Gold (Rank 3): Good conditions (0.4-0.6 kWh/kWp)
• Green (Rank 4): Very good conditions (0.6-0.8 kWh/kWp)
• Dark Green (Rank 5): Excellent conditions (≥0.8 kWh/kWp)

1. Select Installation: Choose "Lisbon_1"
2. Set Date: Click "Today" button
3. Set Capacity: Enter your system size (e.g., 5.0 for 5kWp)
4. Generate: Click "Generate Prediction"
5. Review Results:
   • Check daily total energy production
   • Note the average ranking for planning energy usage
   • Look at hourly breakdown to identify peak hours

1. Select Installation: Choose your preferred installation
2. Set Date: Enter tomorrow's date (YYYY-MM-DD)
3. Set Capacity: Enter your system capacity
4. Generate: Click "Generate Prediction"
5. Plan Activities:
   • Rank 4-5 hours: Schedule energy-intensive tasks
   • Rank 1-2 hours: Minimize electrical usage
   • Use weekly trend to compare with other days

1. Select Installation: Choose installation to analyze
2. Set Date: Enter a past date (e.g., 2023-06-15)
3. Set Capacity: Enter system capacity
4. Generate: Click "Generate Prediction"
5. Compare: Use the plot to see how the selected day compares with surrounding dates

1. Select Installation: Choose your installation
2. Set Date: Pick any date of interest
3. Generate: Click "Generate Prediction"
4. Analyze Trend: Look at the ±7 days view in both text and chart
5. Identify Patterns:
   • Find best days for high energy consumption
   • Schedule maintenance on low-production days
   • Plan battery charging/discharging cycles

• Rank 5 hours: Run dishwashers, washing machines, electric cars charging
• Rank 4 hours: Most appliances, heat pumps, air conditioning
• Rank 3 hours: Normal usage, computers, lights, TV
• Rank 2 hours: Light usage, avoid heating/cooling systems
• Rank 1 hours: Essential devices only, rely on stored energy

• Rank 5 hours: Schedule energy-intensive manufacturing
• Rank 4 hours: Normal operations, office equipment
• Rank 3 hours: Reduced load operations
• Rank 2 hours: Essential services only
• Rank 1 hours: Switch to backup power/grid

• Check: Ensure Excel files are in the project directory
• Solution: Verify PV Plants Datasets.xlsx and weather files exist
• Location: Files should be in the main project folder

• Cause: Internet connection or API issues
• Solution: Check internet connection
• Fallback: System uses default weather values when API fails

• Cause: Model training failed or insufficient data
• Solution: Ensure historical data contains enough records
• Minimum: At least 50 data points needed for training

• Format: Must use YYYY-MM-DD format
• Examples: 2024-03-15, 2023-12-01
• Avoid: MM/DD/YYYY, DD/MM/YYYY formats

• Solution: Close and restart the application
• Check: Ensure Python tkinter is properly installed
• Alternative: Try running on a different machine

1. Use recent dates: Newer weather data provides better accuracy
2. Correct capacity: Enter accurate kWp rating for your system
3. Stable internet: Ensures reliable weather data retrieval
4. Regular updates: Restart application periodically to refresh models

1. Close unused applications: Free up system memory
2. Stable power: Avoid interruptions during model training
3. SSD storage: Faster file access improves loading times

1. Generate Prediction: Create the forecast you want to save
2. Click Export: Use "Export Results" button in output panel
3. File Location: CSV file saved in project directory
4. Filename Format: predictions_{installation}_{date}.csv
5. Content: Complete hourly data with all weather and energy parameters

• Models: Trained models saved in models/ directory
• Logs: Application logs in logs/application.log
• Exports: Exported data in project root directory

• Multiple Installations: Compare different installations by switching between them
• Date Ranges: Analyze seasonal patterns by testing different months
• Capacity Scaling: Test different system sizes to plan upgrades

• Seasonal Comparison: Compare same dates across different years
• Weather Impact: Observe how weather changes affect rankings
• System Planning: Use predictions for sizing battery systems or grid connections

1. Morning Check: Analyze today's forecast to plan daily activities
2. Weekly Planning: Check next 7 days for energy-intensive task scheduling
3. Regular Monitoring: Compare actual vs predicted performance

1. Peak Hour Identification: Use hourly breakdown to find best production times
2. Load Shifting: Move flexible tasks to high-ranking hours
3. Storage Planning: Charge batteries during peak production periods

1. Trends Over Time: Look for patterns in weekly views
2. Weather Correlation: Notice how weather affects production
3. Seasonal Variations: Understand yearly cycles for long-term planning

Need Help? Check the main README.md for detailed technical information or create an issue on the project repository.