Adaptive Neural Network-Based Control of Hybrid AC/DC Microgrid Notes

Introduction to Hybrid AC/DC Microgrid

  • Investigation of grid-connected hybrid AC/DC microgrids integrating diverse renewable energy sources (RES) and energy storage systems.
    • Sources include photovoltaics (PV), wind turbines (WT), solid oxide fuel cells (SOFC), and battery energy storage systems (BESS).
  • Main contribution: Design and validation of an adaptive neural network-based control system.
    • Online-trained artificial neural networks (ANN) used for:
    • Tracking the maximum power point (MPP) of renewable energy generators.
    • Controlling power exchange between front-end converters (FEC) and the electrical grid.
    • Proposal of a fuzzy logic-based power management system (PMS) to minimize grid energy purchases.
    • Simulation in MATLAB/Simulink shows the system's effectiveness and robustness.

Key Terminologies

  • FEC (Front-End Converter): Interfaces the microgrid with the electric grid.
  • PMS (Power Management System): Optimizes power flow and energy exchanges in the microgrid.
  • MPPT (Maximum Power Point Tracking): Algorithms to harvest maximum energy from RESs, considering their intermittent nature.

System Configuration and Modeling

  • The microgrid model includes:
    • 21kWp PV generator
    • 10kW WT generator
    • 10kW SOFC
    • 20Ah Lithium-Ion BESS
    • AC/DC loads
  • Bidirectional buck-boost converter interfaces BESS with the DC-link.
  • Control scheme defines various roles of each energy source and storage facility in maintaining stable operations and meeting demands.

Control Structures

  • Maximum Power Point Tracking (MPPT):
    • The PV system's power output depends on unique operating points governed by the variable climatic conditions.
    • Incremental Conductance (IncCond) algorithm commonly used, but improved by using an SN-RBFN (Single Neuron Radial Basis Function Network) for better dynamics.
  • Predictive Torque Control for PMSG:
    • Applied to the PMSG connected to the WT to enhance dynamic performance in mechanical torque control.
  • ADALINE-based Controllers: Adaptive controllers for adjusting operational parameters under varying conditions to ensure optimal energy extraction from RESs.

Simulation Results

  • Tests were conducted under different conditions to validate performance, such as varying irradiance levels and wind speeds:
    • Demonstrated fast tracking of MPP with SN-RBFN.
    • Effectiveness in following dynamic load changes and responding to grid conditions.
  • Under variable load demands: The system adapts effectively without needing prior knowledge of the external conditions.
  • Performance in grid disturbances: The design exhibited robustness during three-phase voltage dips, ensuring continuity in power supply.

Conclusion

  • The proposed control methods showcase the ability to manage a complex hybrid microgrid in real-time, utilizing adaptive techniques to deal with uncertainties in RES outputs and fluctuating demands.
  • Forthcoming applications will involve implementing these artificial intelligence-based controllers on FPGA platforms for real-world testing.