Neural Network MPPT with NASA/POWER Irradiance and Temperature Data
Collecting Data from NASA POWER
To start, we need historical data on irradiance and temperature for a specific location. Here’s how to collect this data:
Visit the NASA POWER Website:
Go to the NASA POWER website.
Navigate to Data Access and select the Power Data Access Viewer.
Select Location and Parameters:
Choose the location for which you want to collect data. For this example, we'll use Chennai.
Input the latitude and longitude of Chennai.
Select the time range (e.g., the past six months).
Select the parameters: Surface Downward Irradiance and Temperature.
Export Data:
Export the data in CSV format.
Preparing Data for Neural Network Training
Once you have collected the data, follow these steps to prepare it for training the neural network:
Load Data into MATLAB:
Open MATLAB and load the irradiance and temperature data into new variables.
Save these variables as a matrix named data.
Configure PV Panel Parameters:
Define the PV panel rating and other necessary parameters such as short circuit current, current at maximum power point, open circuit voltage, voltage at maximum power point, and temperature coefficients.
Load and Process Data:
Load the irradiance and temperature data into the MATLAB workspace.
Use the collected data to calculate the voltage at the maximum power point for the PV panel.
Training the Neural Network in MATLAB
With the data ready, we can now train the neural network:
Open the Neural Network Fitting Tool:
In MATLAB, open the Neural Network Fitting Tool (nftool).
Input Data:
Load the irradiance and temperature data as input.
Load the calculated voltage at maximum power point as output.
Train the Network:
Configure the training parameters and start the training process.
After training, validate the network by checking the regression plots and ensuring the R-value is close to 1.
Export the Trained Network:
Once the training is complete, export the neural network for use in Simulink.
Implementing MPPT in Simulink
Now, we will integrate the trained neural network into a Simulink model to perform MPPT for a solar PV system.
Create a Simulink Model:
Design a Simulink model for the PV system, including the solar panel, boost converter, and load.
Integrate the Neural Network:
Add the trained neural network block to the model.
Connect the irradiance and temperature inputs to the neural network.
Use the network’s output (voltage at maximum power point) to control the boost converter.
Simulate the Model:
Run simulations with varying irradiance and load conditions.
Observe the PV panel’s power output and ensure it matches the expected maximum power points.
Simulation Results and Discussion
After running the simulations, analyze the results:
Power Output:
Verify that the power output from the PV panel is maximized under different irradiance levels. Compare the simulated power output with the theoretical maximum power points.
Load Variation:
Test the system’s response to changing load conditions. Ensure the MPPT algorithm continues to extract maximum power despite load variations.
By following these steps, you can successfully implement and test a neural network-based MPPT algorithm for a solar PV system using data from NASA’s POWER project.
Conclusion
Using neural networks for MPPT in solar PV systems can significantly enhance the efficiency of power extraction. With historical data from reliable sources like NASA POWER, we can train accurate and robust neural networks to perform MPPT under varying environmental conditions.
Comments