Fault Diagnosis Based Machine Learning and Fault Tolerant Control of Multicellular Converter Used in

Currently, providing water in developing countries, especially in dry and hot rural areas, is a significant challenge. However, creating new electric grids is often expensive. Therefore, the use of low-cost photovoltaic (PV) panels in water pumping systems, without chemical energy storage, based on high-performance and more efficient power converters with increased time life and lower maintenance interventions is needed. In this study, a photovoltaic water pumping system with two power converters, the first is used to extract the maximum power using the maximum power point tracking (MPPT) algorithm, and the second is a three-cell multicellular power converter used to control the DC motor with a submerged pump. Meanwhile, the serial connection and redundant topology of multicellular converters render the system more vulnerable to failure. fault diagnosis-based machine learning approach and fault tolerant control (FTC) are proposed for multicellular power converters. Simulation results with MATLAB show the effectiveness and practicability of the proposed structure and control to isolate the faulty capacitor, increase the sustainability of the system, assure the supply of water under faulty conditions, minimize the mechanical vibrations in electric DC motors, and avoid PV system shutdown.