Model-based fault detection algorithm for photovoltaic system monitoring

Author(s):  
Fouzi Harrou ◽  
Ying Sun ◽  
Ahmed Saidi
2020 ◽  
Vol 17 (1) ◽  
pp. 1
Author(s):  
Nurmalessa Muhammad ◽  
Nor Zaini Ikrom Zakaria ◽  
Sulaiman Shaari ◽  
Ahmad Maliki Omar

The failure detection in a grid-connected photovoltaic (PV) system has become an important aspect of solving the issue of the reduced energy output in the PV system. One of the methods in detecting failure is by using the threshold-based method to compute the ratio of actual and predicted DC array current and DC string voltage value. This value will be applied in the failure detection algorithm by using power loss analysis and may reduce the time, cost and labour needed to measure the quality of the energy output of the PV system. This study presented the threshold value of DC array current and DC string voltage to be implemented in the algorithm of fault detection in grid-connected photovoltaic (PV) system under the Malaysian climate. Data from the PV system located at Green Energy Research Center (GERC) was recorded in 12 months interval using the monocrystalline PV modules. The actual data was recorded using five minutes interval for 30 consecutive days. The prediction of the data was calculated using the mathematical method. The threshold value was determined from the ratio between actual and predicted data. The results show that the DC array current threshold value, σ is 0.9816. While, DC string voltage threshold value, λ is 0.9261. The proposed value may be beneficial for the determination of threshold value for regions with the tropical climate.


Author(s):  
A Nikranjbar ◽  
M Ebrahimi ◽  
A S Wood

Much research works address model-free or signal processing and spectral analysis-based fault detection schemes for rotor eccentricity fault in induction motors. Nevertheless, despite existing reliable fault-embedded eccentricity mathematical models such as the winding function method an integrated model-based fault detection algorithm for detecting this fault yet has not been fully explored. This article presents model-based mixed-eccentricity fault detection and diagnosis for induction motors. The proposed algorithm can successfully detect faults and their severity using stator currents. To determine the values of the fault-related parameters, an adaptive synchronization-based parameter estimation algorithm is introduced using particle swarm optimization. Simulation and experiments demonstrate the ability of the algorithm to detect and diagnose these faults. The proposed algorithm can be employed to estimate the parameters, in addition to slowly time varying and abruptly changing parameters.


2021 ◽  
Vol 3 (5) ◽  
Author(s):  
Ujjval B Vyas ◽  
Varsha A Shah ◽  
Srivani S G

AbstractMultilevel Inverters (MLI) are a viable option for a filter-less and transformerless photovoltaic system for direct grid integration, reducing losses, space and cost provided the issues of control and reliability are resolved. A modified cascaded half-bridge MLI with polarity changer is proposed with a reduced number of switches, thereby reducing control complexity. The proposed converter is designed for both seven-level and nine-level topologies. Selective harmonic elimination has been adopted to switch converter and the transcendental equations are solved by the gravitational search algorithm. The THD for nine-level configuration is 7.94% and 5.86% with MPPT and DC source inputs, respectively. The analysis on the open-circuit fault of switches confirms the presence of only five critical switches irrespective of the number of increase in levels, thereby requiring only five redundant switches at the Polarity Changer. The output voltage waveform is subjected to multiresolution analysis for feature extraction of voltages under various irradiance and temperature conditions. A conditions based fault detection algorithm is developed based on the observations of energies of signals to detect the open circuit fault in switches. Based on the comparative analysis, the proposed converter has fewer controlled switches for nine levels or higher configurations than other topologies. Apart from this, the high reliability due to inherent fault isolation capability restricted to half-bridge compared to different MLI topologies shows its superiority. For a nine-level MLI proposed topology reduces the cost considering redundancies for FDI in range of 67–30% depending on the topology for comparision. The system is simulated using MATLAB –Simulink and further validated by experimental results.


Sign in / Sign up

Export Citation Format

Share Document