Design of Parallel Hybrid Active Power Filter with Adaptive DC-Link Voltage Control Based on Artificial Neural Network

Design of Adaptive Hysteresis Current Control Based on Artificial Neural Network for a Single-Phase Bidirectional Voltage Source Inverter

The rapid growth of electric vehicles and energy storage systems requires efficient two-way power conversion systems, such as bidirectional VSIs operating in Grid-to-Vehicle (G2V) and Vehicle-to-Grid (V2G) modes. Unfortunately, conventional Hysteresis Current Control (HCC) methods lead to unstable switching frequencies, degrading the system's power quality and performance. This study proposes a single-phase two-way Voltage Source Inverter (VSI) with a full bridge topology controlled by an Artificial Neural Network (ANN) based Adaptive Hysteresis Current Control (AHCC) scheme. ANN is used to define hysteresis bands adaptively to keep the switching frequency stable. The system consists of a two-way VSI connected to the grid and a two-way DC-DC converter that connects the battery via DC-Link. The simulation results showed that the proposed method produced a narrower instantaneous frequency switching range of 5.263 kHz-25 kHz compared to Fixed-HCC of 11,111kHz-50 kHz, so that AHCC-ANN produced an average frequency switching value of 13.37 kHz, close to the desired frequency switching of 15 kHz, while Fixed-HCC was 20,25 kHz. On the other hand, the% of THD for AHCC-ANN (2.72%) is lower than that for Fixed-HCC (3.2%). On the other hand, the DC-link voltage can also be maintained at 400 V during charging and discharging. These results show that AHCC with ANN can stabilize switching frequencies and DC-link voltages and support effective bidirectional power flow.

An Adaptive BOA-Tuned PI Controller for DC-Link Voltage Regulation and Bidirectional Power Flow Control in EV Charging Applications

This paper presents an advanced grid-connected electric vehicle (EV) charging system incorporating bidirectional battery energy storage and an Adaptive Butterfly Optimization Algorithm (ABOA)-tuned PI controller for DClink voltage regulation and bidirectional power-flow control. The proposed architecture consists of a PWM rectifier for AC/DC conversion with near-unity power factor and reduced input current harmonics. A highfrequency inverter and isolation transformer provide compact and efficient isolated power transfer for EV charging applications. The major contribution of this work is the implementation of an Adaptive BOA-tuned PI controller that dynamically optimizes PI gains to maintain stable DC-link voltage under varying load conditions and battery operating states. The adaptive optimization process minimizes voltage error, overshoot, settling time, and ripple, thereby improving transient response and converter performance. MATLAB/Simulink simulation results demonstrate that the proposed controller achieves superior performance compared with conventional PI, PID, FLC, ANN, PSO, GA, SMC, and MPC-based control methods, achieving a settling time of 0.089 s, conversion efficiency of 98.5%, and THD of 0.6%. Hardware implementation using dsPIC30F4011 digital controllers validates the practical feasibility of the proposed system, showing close agreement between simulation and experimental results in terms of DC-link stability, bidirectional energy management, and EV charging performance. Overall, the proposed Adaptive BOA-tuned PI controller provides a computationally efficient and cost-effective solution for next-generation smart EV charging infrastructures with improved dynamic stability, power quality, and energy efficiency.

FUZZY-BASED ADAPTIVE STEP-SIZE FINITE VOLTAGE SET MODEL PREDICTIVE CONTROL FOR GRID-CONNECTED PHOTOVOLTAIC SYSTEMS

ANFIS-Based Duty-Cycle Control for a Non-Isolated Interleaved Bidirectional DC–DC Converter in Hybrid PV–Battery Systems

In this paper, an Adaptive Neuro-Fuzzy Inference System (ANFIS)-based duty-cycle correction method is proposed for a non-isolated interleaved bidirectional DC–DC converter used in a hybrid photovoltaic (PV)–battery system. The ANFIS controller was developed using training data generated from an optimized conventional fuzzy-logic controller operating in both buck and boost modes. The main objective of the proposed control strategy is to produce a smoother duty-cycle response, improve transient behavior, and maintain better output-voltage stability than the conventional fuzzy-logic approach. The converter and control system were modeled and tested in MATLAB/Simulink under constant-voltage and constant-current operating conditions. The simulation results indicate that the proposed ANFIS-based controller improves the converter's dynamic response and provides smoother duty-cycle adjustment. In buck mode, the output-voltage ripple is reduced from 0.1181 V to 0.1051 V, indicating a modest improvement. A more significant improvement is observed in boost mode, where the voltage ripple decreases from 2.963 V to 1.348 V. The results indicate that the proposed ANFIS controller works effectively in boost mode, particularly given its greater sensitivity to duty-cycle changes, switching dynamics, and transient disturbances.

Design of an Adaptive Estimator‐Based Sliding Mode Controller With an Initial Voltage Error‐Independent Reaching Time for Voltage Control of a DC Microgrid With Constant Power Loads

DC microgrids face major operational challenges because constant power loads maintain their power consumption at fixed levels. The network experiences its main distortion source from these elements, which produce oscillations and lead to system breakdown through their negative incremental impedance effects. The authors of this study present a fast‐SMC system that regulates DC microgrid voltage levels while operating with constant power loads. The system enables fast bus reference voltage tracking, together with better tracking performance and reduced system oscillations. To achieve these capabilities, a new reaching law is introduced that is independent of initial conditions and provides fixed‐time convergence. It also helps reduce chattering by reducing the sign function coefficient. In addition, a new adaptive estimator is presented that is able to calculate the network power instantaneously and makes the control law robust to power variations. The research team conducted experimental testing to assess how well the control system performed. The research findings reveal that the recommended scheme achieves better results than all other tested methods.

Enhanced Model Predictive Control Using Modified Space Vector Modulation for Grid-Connected 3L-NPC Inverters

This paper presents a power control strategy for a grid-connected three-level neutral-point-clamped inverter. The proposed approach integrates Model Predictive Control (MPC) with a modified Space Vector Modulation (SVM) scheme. The control structure comprises an inner current control loop and an outer DC-link voltage control loop to regulate both active and reactive power. To address the neutral-point voltage imbalance issue, a modified SVM method is developed by adaptively adjusting the dwell times of voltage vectors. This approach effectively balances the capacitor voltages, mitigates voltage fluctuations, and reduces current harmonic distortion. Simulation results in Matlab/Simulink verify that the active and reactive powers accurately track their reference values, achieving minimal steady-state error and fast dynamic response. In addition, the DC-link voltage remains stable, and the capacitor voltages are well balanced even under the application of medium voltage vectors. Furthermore, compared with the conventional finite control set MPC (FCS-MPC), the proposed method achieves superior performance in terms of lower total harmonic distortion of the grid current, owing to its fixed switching frequency, as well as reduced neutral-point voltage ripple.