Shading pattern detection using electrical characteristics of photovoltaic strings

Partial shading conditions are inevitable especially in building-integrated photovoltaic (PV) systems. Detection of partial shading is vital for monitoring and supervising purposes as well as invoking global maximum power point tracking (MPPT) algorithms. Normally, a sudden big change in output power is used as a partial shading occurrence indicator. However, it cannot ensure detection accuracy. Utilizing the electrical characteristics of PV strings, this paper proposes a shading pattern detection method which applies multiple-output support vector machine (M-SVM) to estimate the shading rate and shading factor. A non-dominated sorting genetic algorithm-II is used to select hyper parameters of M-SVM. Simulations in Matalb and PSIM validate the effectiveness of the proposed method in the face of various shading patterns.