by Mina Mesbahi
As solar plants grow in size and complexity, system operators are required to combine data gathered from multiple sources to make decisions to optimise the performance of their assets. In this webinar (of which the full recordings are now available), Rob Andrews, CEO of Heliolytics and Hugo Lapie, Director of Sales of QOS will discuss different cases of investigation using real-time data, aerial inspections, and prediction tools to identify issues, improve diagnosis and shorten resolution times.
1. Advanced data analytics
In terms of data exchange methods, there are not any standard practices. This leads to greater challenges when managing a higher number of assets since the data needs to be aggregated from multiple systems and then centralised. There are multiple factors to consider in the realm of advanced data analytics, the most important of which is data accuracy/quality. The accuracy of data can be improved using the QOS via:
Data cleansing: Despite implementing the best processes, the data received can be incoherent at times. According to Hugo Lapie of QOS Energy, in such instances, Machine Learning (ML) can be a robust and scalable tool to filter the inaccurate/ incoherent data.
Data filling: In addition to data cleansing, ML can be utilised to fill in any gaps of missing data (i.e. due to communication errors) using external data sources such as weather forecast.
All in all, detection of under-performances across a portfolio, or a specific plant, is feasible only when the accuracy of data is enhanced. Upon detection, a more in-depth investigation can be conducted using multiple sources of data including on-site inspection, weather forecast, aerial inspection, satellite radiation data, etc.
2. Aerial inspection for portfolio asset optimisation
As stated earlier, a number of data sources can be used in advanced data analytics to optimise plant performance. One of the newest data sources being introduced is aerial inspection. In this webinar, Rob Andrews, CEO of Heliolytics, discusses how aircraft PV inspections can result in an industry leading level automation in PV systems. According to Andrews, automation is valuable in terms of improving existing processes as well as enabling new processes.
2.1 Aircraft inspection process
The aircraft inspection process begins with creation of a digital as-builts, which is a digital asset registry, and includes the location of every module and component in the system. Secondly, aerial inspection is carried out by the aircraft. Heliolytics uses visible and infrared images to collect data, which allows for cross-correlation. The next step is very crucial, which is data analysis analytics. Heliolytics has developed an artificial intelligence toolset, which examines the data collected and identifies location classification faults. The last step is reporting, which can take different forms. One other critical point is the ability to distinguish between hotspots created by soiloing and hotspots due to inherent module defect.
2.1.1 Data capture
Heliolytics uses a manned aircraft to capture data, enabling them to place high-quality sensors onto the aircraft. This increases the quality of the data and the speed, at which data is gathered. Another important factor regarding data capture is that the system should be at a relative thermal equilibrium when the data is captured to avoid false positives in the analysis.
2.1.2 Data Analysis
The next step in the inspection process is the analytics side. Artificial Intelligence (AI) can be used to analyse the data capture and locate the defects. Overall, AI tools provide accurate loclalisation and classification of defects detected. Another important aspect of data analytics is the conversion from an analogue dataset to a digital one.
2.2 Aircraft inspection for asset optimisation
According to Andrews, The standard practice up until a few months ago has been manual DC preventive maintenance, which historically translates to manual I-V curve or electrical inspections. The cost structure associated with this practice is linear and therefore not a scalable proposition, in addition to imposing some health risks and safety hazards. The non-digital nature of the inspection calls for inconsistencies in accuracy. Heliolytics conducted a field study to compare manual electrical inspection with aircraft inspection and found out that manual electrical tests identified only 22% of faults detected by aircraft inspection. Therefore, aerial inspection is a more plausible method for asset optimisation.
2.2.1 Improve existing processes
Classification of faults: Digital aircraft inspection provides a unified view and a standardised classification system in identifying and disaggregating defect types.
Location of faults: A detailed site-mapping can be achieved when using aircraft inspection, which provides accurate site location and geolocation of system faults. This is particularly valuable in terms of remediation action planning and early identification of warranty defects.
2.2.2 Enabling new processes
Workflow integration: Having a digital product facilitates integration and exportation into existing workflows such as SCADA, asset management, case management systems.
Incorporation of disparate datasets: A serial number batch with a statistically higher level of module faults can be identified through correlation of module serial numbers. This also allows for spotting the early warning signs for warranty issues.
Fault evolution tracking: System evolution over time supplies crucial metrics on site or operations performance, which lays the groundwork for predictive analytics.
Portfolio level analytics: The standardised datasets across the entire portfolio allows a broader level portfolio analysis.