Neural Network control of grid-connected PV system
Introduction
The neural network boost control approach is designed to improve power quality in grid-connected solar PV systems. This method utilizes a neural network to generate reference currents for controlling a voltage source inverter (VSI), thereby optimizing power delivery from the PV system to the grid.
Simulink Model Explanation
Overview of the Model
The Simulink model includes several key components:
Solar PV Panel: Each panel has a rating of 213 watts.
PV Array: The array consists of 20 series modules and 18 strings, generating a total power of approximately 95.9 kilowatts under standard test conditions.
MPPT Algorithm: Incremental conductance method is used for MPPT, generating reference voltage based on PV panel voltage and current.
Voltage Source Inverter (VSI): Connected directly to the PV panels via a DC capacitor and an inductor, and further connected to a 3-phase grid and a nonlinear load.
Control Logic
The control logic for the VSI involves generating reference currents to be compared with the actual VSI currents for impulse generation. This process involves several steps:
Measure PV Panel Voltage and Current: These values are used as inputs for the MPPT algorithm.
Generate Reference Voltage: The incremental conductance method generates the reference voltage.
Current Reference Generation: Reference current is generated based on the PV power and load current, processed through a neural network.
Neural Network Control: The neural network is trained with various load conditions (nonlinear, unbalanced, normal) to generate accurate reference currents.
Voltage and Current Measurement: Terminal voltage and current are measured and used to control the VSI, ensuring reactive power injection and power quality improvement.
Neural Network Control Details
Training the Neural Network
The neural network is trained using load current data from different scenarios:
Nonlinear Load
Unbalanced Load
Normal Load
The network receives three inputs (load currents in each phase) and generates reference currents based on the trained data. These reference currents are used to control the VSI, maintaining sinusoidal source currents and improving power quality.
Control Process
Reference Current Calculation: Based on PV power and load conditions.
Voltage Measurement: Terminal voltage is calculated using measured voltages (Vsa, Vsb, Vsc).
Current Comparison: Reference currents are compared with actual VSI currents for impulse generation.
VSI Control: The VSI is controlled to inject reactive power and mitigate harmonic problems, ensuring sinusoidal source currents.
Simulation Results and Discussion
PV System Performance
Irradiance Change: The simulation demonstrates the system's response to changes in irradiance from 1000 W/m² to 500 W/m², showing corresponding changes in PV voltage, current, and power.
MPPT Efficiency: The incremental conductance method effectively tracks the maximum power point, even under varying irradiance conditions.
Power Quality Improvement
Nonlinear Load Conditions
Load Current: Non-sinusoidal load current is observed.
Grid Current: Despite the nonlinear load, the grid current remains sinusoidal due to the neural network control.
Unbalanced Load Conditions
Load Condition: Unbalanced load condition is created by disconnecting one phase at 5 seconds.
System Response: The system maintains sinusoidal grid and inverter currents, with the PV voltage and power stabilizing after a brief oscillation period.
Dynamic Performance
PV Voltage and Power: Stable PV voltage and power are observed after initial transients due to load changes.
Grid Current: Sinusoidal grid current is maintained throughout the simulation, demonstrating the effectiveness of the neural network control in improving power quality.
Conclusion
The neural network-based control approach for grid-connected PV systems significantly enhances power quality by maintaining sinusoidal source currents and mitigating harmonic problems, even under varying load conditions. This method ensures efficient and reliable power delivery from solar PV systems to the grid.
ความคิดเห็น