Deep learning based control of DSTATCOM for harmonic mitigation in hybrid distributed generation system with non-linear loads

The rising energy demand from the growing population has been historically met by conventional sources, but their availability is diminishing. To counter this, renewable energy sources are emerging as global alternatives to fossil fuels, with solar and wind power leading the way. Solar and wind energy, known for their reliability, are combined to enhance energy system stability and reduce generation costs. Implementing such hybrid systems enhances reliability and cost efficiency for load demands, especially for power electronics-based loads introducing harmonics. Various mitigation strategies have been developed, including active power filters like Distribution Static Compensators (DSTATCOM) that regulate power quality. However, conventional methods such as Synchronous Reference Frame (SRF) & Instantaneous Reactive Power (IRP) have limitations with non-linear loads and filter parameter estimation. To address these challenges, the paper proposes a Deep Learning (DL) control algorithm, using backpropagation (BP) for hybrid solar-wind systems, providing consistent and cost-effective energy solutions. The proposed technique's effectiveness is demonstrated through MATLAB/Simulink and real-time hardware setups.