Artificial intelligence for solar plants | Iqony Solar Energy Solutions

Smart maintenance: AI for solar plants

How artificial intelligence improves your solar park’s performance


Anyone operating a solar park expects the highest yield over the entire lifespan of the plant. Avoiding downtime while keeping maintenance and operating costs low is just as important. Luckily there are AI-based monitoring systems, which independently monitor the system data, detect deviations at an early stage and enable predictive maintenance before yield loss occurs. Read on to learn more about the advantages and functionality of artificial intelligence in solar monitoring.

PV ground-mounted systems are generally considered to be less vulnerable to faults. The longer they are in use, however, the more often they might show signs of wear, resulting in more maintenance and a decreasing energy yield. Pollution can also cause a plant to deliver a lower yield, which means good monitoring is essential for optimal energy production. A Swiss study* confirms that a PV system’s performance is strongly influenced by the quality of maintenance.

Artificial intelligence (AI) is the simulation of human intelligence processes by machines, especially computer systems, meaning that AI deals with methods that enable a computer to imitate human-like abilities such as logical thinking, learning, planning and creativity.

Classic remote maintenance systems often react too late

Solar plant operators usually commission external specialists to monitor their plant who analyse the transmitted data via remote monitoring. If there are deviations or something wrong with the strings or inverters, they send a technician on site to rectify the fault. Though commonplace, this kind of maintenance carries the risk that operating errors are only detected when yield loss or even major damage has already occurred. This is where AI-based methods for monitoring solar parks come in, as they can predict impending plant problems before they occur, allowing you as the operator to intervene preventively – hence the term "predictive maintenance".

Artificial intelligence enables early fault detection

A practical example shows the potential of AI-based solar park monitoring:

The central inverter of a SENS client’s PV system repeatedly switches off for a short time, the reason being that the insulation resistance is too low. However, their current monitoring system is not detecting these failures. The several times the inverter switches off are instead recorded as small deviations within the 15-minute measurements; the software only sounds the alarm when the inverter fails for longer than a minute and the deviations increase.

That the real error is only recognised at a late stage is partly due to the rigid limit values of older systems, since they are deliberately set high to prevent harmless deviations from causing a flood of error messages. But as you can see, this system is not monitoring the insulation value, which dropped a long time before any error messages were registered.

AI-based monitoring systems, such as Sensaia from SENS and Iqony, can continuously analyse a wide range of values. They register even the smallest deviations within tolerance range via algorithms that evaluate and compare them with reference values and historical data. "That way, the plant operator can also recognise faults that are starting to creep up", explains SENS Data Analyst Jonas Hergenhan, who is in charge of the Sensaia software solution. In this particular instance, Jonas is sure that "our software would have detected the error three months earlier".

Predictive maintenance increases yield

Early fault detection is crucial for the efficient operation of solar parks, allowing you to plan maintenance in advance and have the replacement or repair of a defective component carried out at the same time as other work. If solar modules are dirty, the monitoring system uses weather forecasts to calculate how significant of a yield loss there could be in the coming days. It also shows whether an upcoming rainfall, for example, may already solve the problem. This type of predictive maintenance can therefore significantly reduce operating costs and make cleaning much easier to plan.

Fewer false alarms mean fewer unnecessary site visits

Another factor that saves the service team time and the system owner money is the pre-filtering of alarm messages. If the system detects a deviation, it not only generates a message but also analyses possible correlations with other messages based on AI grouping them accordingly. Staff can therefore get to the bottom of the anomalies, and what caused them, more quickly. The AI-supported quality check also reduces the number of false alarms, minimising unnecessary on-site operations.

Artificial intelligence provides valuable support to the monitoring staff but is not meant to replace them; a staff member is the last line of defence that checks each error message and makes the final decision on whether to issue a work order for a service technician.

Solid data for making decisions

AI-based monitoring systems provide you with a solid basis for your operational decisions, but a good data base is the central success factor here, as Jonas Hergenhan explains: "The more data we have, the more accurate the system can be, so it’s important to give feedback to the system on the work done and whether the system’s prediction was correct. The system ‘learns’ from this so that fault detection and forecasts become more and more reliable over time", with AI-based systems also analysing error data from third-party systems in addition to the data from the connected photovoltaic system.

The operator can access all this data at any time via the Sensaia platform and thus obtain real-time insight into the plant’s current status. The owner, too, can view the live status at any time and monitor the performance of the entire solar plant as well as any maintenance work.

As these intelligent maintenance concepts have already proven their worth in ensuring maximum yield while minimising operating costs in conventional power plants and the wind sector, their widespread use in PV systems is not far behind.

Would you like to know more about Sensaia? Sensaia looks beyond standard monitoring and alarm management, picking up where others leave off. Predictive maintenance is based on experience gained from using proven algorithms for over ten years. Valid alarm recognition and intelligent management combined in one interface sets the course for the future, adding a new level of efficiency and planning security.


Bild: Sensaia Software von Iqony und SENS

* Vontobel, Thomas (2019): Performance von PV-Anlagen unter der Lupe. hg. von: Bulletin SEV/VSE.

Go to top