This paper introduces a novel non-integral method for calculating AC capacitor voltage in shunt hybrid active power filters (SHAPFs). These filters commonly use passive power filters (PPFs) that can introduce voltage containing harmonics into the AC capacitor, leading to overvoltage situations that can be dangerous. To avoid accidents due to overvoltage, it is necessary to monitor the AC capacitor voltage continuously. The proposed method enables the calculation of the voltage across the AC capacitor, using only the sensor available in the SHAPF, without requiring additional sensors. By avoiding the use of integration to calculate the voltage, the method effectively mitigates errors due to DC offset inflows. The effectiveness of the proposed method was verified through simulations and actual experiments. Results demonstrate that the proposed voltage calculation method improves the voltage calculation accuracy by approximately 1.93[%] when compared to the general voltage calculation method. The proposed non-integral AC capacitor voltage calculation method represents a significant advancement for SHAPFs, providing an effective means for reducing overvoltage risks, without requiring additional sensors or costly equipment.

This study proposes a new maximum power point tracking (MPPT) method based on machine learning with improved power production efficiency for application to photovoltaic (PV) systems. Power loss occurs in the incremental conductance (InC) method, depending on the size of the voltage step used to track the maximum power point. Additionally, the size of the voltage step must be specified by the initial user; however, an appropriate size cannot be determined in a rapidly changing environment. To solve this problem, this study presents a gradient descent InC (GD InC) method that optimizes the size of the voltage step by applying an optimization method based on machine learning. The effectiveness of the GD InC method was verified and the optimized size of the voltage step was confirmed to produce the largest amount of power. When the size of the voltage step was optimized, a maximum difference of 4.53% was observed compared to case when the smallest amount of power was produced. The effectiveness of the GD InC method, which improved the efficiency of power production by optimizing the size of the voltage step, was verified. Power can be produced efficiently by applying the GD InC method to PV systems.