Résumé: The kWh price of renewable energy power plants is still very costly, so any malfunction or weak yield is prejudicial to guarantee the investment payback. Therefore, in this case, sustainability re-assessment of the whole system is required. In general, faults are difficult to detect, and very quickly they evolve rapidly and exponentially. Keeping the faulty part in operation has a negative influence on the other healthy parts, through additional electrical constraints, which increasingly weaken them (risk of cascading faults). So, it is important to detect faults early in order to eliminate or correct them. Early detection implies in-depth diagnosis, for this purpose, several detection processes require the planned disconnection of photovoltaic power plant production. This technique has negative economic repercussions. Our proposal is based on continuous monitoring which automatically launches the detection process as soon as a tiny anomaly appears, without stopping production, unlike other techniques. It is also a real time faults parameters observation. The sensitivity of the developed algorithm is based on the interaction between the electrical characteristics of the panels. The behavior of the photovoltaic panel under different levels of degradation is highlighted and discussed. Therefore, a new diagnosis process is developed and achieved successfully based on specified points gathered from new proposed characteristics very sensitive to any variation from healthy case. This sensitivity allows to an earlier fault detection.

Résumé: The reliability of any electrical power system must be evaluated in order to increase durability. In accordance with the IEEE standard 1366, 2012, reliability is modeled by quality of service and profitability indices. This reliability highly depends on the applied grid code. An investigation was conducted on data recorded during these last five years since 2018 from OUED KEBRIT PV power plant located in Algeria. It was highlighted in this paper that the inverter power controller can decrease significantly the reliability. In many cases electrical protections of PV power plant react by disconnecting it, whereas the fault is outside. Data mining shows that many PV power plant breakdowns are not necessary. A modified active and reactive power control strategy was applied to the PV power plant and compared to the existing one. The proposal is based on software and hardware management. Thanks to the proposed strategy control, the PV power plant decoupled control is more reliable and was highly improved in this paper.