Monitoring of PV plants : enhanced methods and yield forecasting

Abstract

In the past 20 Years Photovoltaic has evolved to be widely known as a reliable renewable energy source. The fact that a PV-plant can be realized without any moving parts, thus being very easy to maintain, makes this resource even more interesting for anyone interested in constant trouble free energy production. But - as in every power electric process - the conversion from light to power is still constantly under the threat of malfunction, due to various external and internal reasons. To minimise possible downtimes and to have the ability to exactly locate possible errors in order to fix them as soon as possible, automated monitoring systems have become more and more important in existing and newly erected plants. This is true not only for big installations but recently has become available also for small household installations. The framework of this Master thesis is a detailed analysis of all possible errors that can occur in the PV power conversion. The first part describes the possibility of fast and exact detection of an error within a PV plant. Detection methods are discussed and evaluated by describing character and appearance of the error. The second part of the work is a detailed overview over the monitored parameters and the methods that can be used to evaluate the measured data. Diagrams and statistics are presented and discussed in order to understand their importance as a source for error localisation. The third chapter describes enhanced surveillance methods based on the capabilities of methods presented in the previous findings. Finally a possible method of yield forecasting is described and discussed. Based on meteorological data provided by ZAMG a calculation is made to evaluate the capability of such a forecasting method. One outcome of this demo calculation is that it can be said that an exact yield forecast is nearly impossible because of the unpredictability of clouds. On the other hand it can be shown that under stable weather conditions the accuracy of prediction can reach more than 90%.