Predictive Maintenance for Wind & Photovoltaic plants

25th May 2020

Minimize risk with a proactive management

In a global scenario of decreasing electricity prices, minimizing lost production due to downtimes is key to deliver the targeted returns on renewable asset investments.
To help renewable asset owners to mitigate the risk of asset failure, Nispera has developed together with the Zurich University of Applied Sciences (ZHAW), a Predictive Maintenance module for wind and photovoltaic plants. Leveraging the readily available operational data generated by the assets, thus avoiding the installation of additional sensors, our Artificial Intelligence (AI)-based models can identify the failure patterns of the assets’ components and release alarms at an early stage. Thanks to its technologically innovative nature, the project has been selected for funding by Innosuisse, the Swiss Innovation Agency.


Learning from all the inputs received over a statistically relevant period, the AI-powered algorithm can replicate a reference behaviour of the components under different operational and ambient conditions. Deviations from the predicted reference are then analysed and ranked according to their likelihood to lead to a failure. In doing so, the AI-algorithm becomes more accurate the more data it can employ for learning. Real time notifications are triggered as soon as an asset’s component is showing an alarming trend. In addition, the Nispera web-platform offers different tools to support a deeper analysis of the suspect behaviour. In a few clicks, the user has all the information to plan corrective maintenance actions or to raise the awareness of the other actors involved in the asset management about the emerging issue. While the severity of some deviations signals an imminent failure and requires immediate actions, other malfunctions develop over a longer time and can be easily fixed by prioritizing inspections on the component during routine maintenance activities.

As former asset managers, we are aware that analysing data from the assets is a good starting point but it is not enough ” says Gianmarco Pizza, CEO of Nispera. “In our daily information overload, it is equally important to have tools cutting through the noise and able to provide immediate, clear suggestions on the next actions to take. We want our clients to spend less time on data crunching and free up resources to dedicate on initiatives that drive the actual improvement ”.


While the focus has always been on creating an intuitive and user-friendly solution, we also wanted to make sure the Users could rely on the insights offered by the most advanced analytics crunching the wealth of data available from the assets. As a tangible proof of our commitment, the results of our applied research activity on Predictive Maintenance together with ZHAW University have led to the preparation of several scientific papers, some already accepted for publication (selected to be presented at major scientific events in the Predictive Maintenance field*).

The alarming trends, clustered by asset and component, are reported in our cockpit in a clear and intuitive manner. 


A key input to our endeavour has come from our industry partner EKZ, the utility of Zurich canton, a leading player in the renewable energy sector owning and operating a large asset portfolio of wind and photovoltaic plants distributed across Europe. The involvement of EKZ from the beginning, was crucial to ensure the development of a tool able to offer a great user experience while living up to the requirements of the most advanced users. “We have followed the development and provided suggestions over the course of this exciting project, leading to an excellent solution which we have started using in our daily operations. The insights from Nispera’s Predictive Maintenance module have already helped us to detect real cases that otherwise would have gone unnoticed, avoiding the costs of component replacement and the related downtimes. Even if today most of the assets are still covered by full-scope O&M agreements, predictive tools are essential to minimize operational risks, and with the gradual shift to partial scope O&M agreements, such tools will become essential ” says Christian Hürlimann, Managing Director of EKZ Renewables.

If you would like to know more about how we can help to improve the performance of your renewable assets, get in touch with us to schedule a demo.

*Early Fault Detection Based on Wind Turbine SCADA Data Using CNNs, Eskil Jarlskog, Gianmarco Pizza, Jaakko Manninen, Lilach Goren Huber, Markus Ulmer, to be presented at the 33 rd International Conference on Condition Monitoring and Diagnostic Engineering Management (COMADEM 2020), 26-28 May 2020, Luleå, Sweden.

Early Fault Detection of Wind Turbines Based on Alarms and Warnings from the SCADA System, Antonio Notaristefano, Gianmarco Pizza, Gregory Fabbri, Lilach Goren Huber, to be presented at the 5 th European Conference of the Prognostic and Health Management Society (PHM 2020), 1-3 July 2020, Turin, Italy