Inferential statistics for monitoring and fault forecasting of PV plants

This paper proposes a procedure, based on both descriptive and inferential statistics for diagnosis of PV plants. This study aims to developing an algorithm able to recognize accurately among a degradation status and a system abnormality before a fault occurs. The statistical approach, based on the ANOVA and Kruskal-Wallis tests, is effective in locating abnormal operating conditions even in the presence of a reduced availability of energy measures. The proposed algorithm has been applied to a case study and advantages and limitations are presented.