Development of condition monitoring and fault detection techniques for photovoltaic systems

Abstract

Due to the increasing demand for renewable energy, the photovoltaic (PV) system as one of the most popular renewable systems is developing significantly. However, the safety, reliability, and efficiency of a PV system are undermined by the potential defects and faults. In the component level of a PV system, due to the high energy density and low cost of an aluminum electrolytic capacitor (AEC), it is extensively used for energy storage and filtering applications in the power electronic converters. However, the lifetime of an AEC is limited due to the electrolyte vaporization. The reliability and efficiency of the power electronic converter such as a boost converter in a PV system are jeopardized by the degradation of AECs. Thus, it is necessary to develop a reliable condition-monitoring technique for AECs in PV systems. In this research, the ripple method calculating the equivalent series resistance (ESR) by dividing the voltage ripple by the current ripple is further developed for condition monitoring of AECs in the boost converter. The effects of the load resistance, duty cycle, inductance, ESR, and capacitance (C) on the ESR estimation in a boost converter are investigated. The estimation of C can be achieved by considering the voltage drops on C. An online AEC-monitoring scheme based on the proposed ESR-estimation and C-estimation methods and magnetic-field sensing by magnetic sensors is developed for PV systems. The capability of the technique was validated with simulation and experimental results under various working conditions. In the system level, the electric faults can cause malfunctions and even fire hazards in a PV system. The series arc fault establishes a current path in the air that may be caused by an unintended discontinuity of current-carrying conductors. However, the randomness and complexity of series arc faults make it is difficult to define their characteristics for arc fault detection. In this research, the pink noise equation is applied to analyze the characteristics of arc faults under various conditions. The magnetic sensor is used to capture the arc induced noises to detect the arc faults non-invasively. To avoid other environmental and power electronic noises, the empirical mode decomposition analysis is adopted for arc-signature extraction. The effectiveness of the developed technique was experimentally validated in a PV system. Apart from the arc fault, the line-line (LL) fault is also challenging the normal operation of a PV system. The LL fault which establishes an unintentional current path between two points of different potentials cannot be terminated by the conventional protection devices. In this research, the effects of the maximum power point tracking controller, partial shading, and blocking diodes on string current under LL fault conditions are studied. A LL fault detection technique based on the fault string-current behavior and current sensing is developed. The simulation results verified that the proposed technique can detect the LL faults effectively. Based on the proposed techniques, the magnetic sensors can be placed on target locations to measure the current for the condition monitoring and fault detection of PV systems.