Systematic cross-validation of photovoltaic energy yield models for dynamic environmental conditions

Anagnostos, Dimitrios; Goverde, Hans; Catthoor, Francky; Soudris, D.; Poortmans, Jef

doi:10.1016/j.solener.2017.07.011

Cited by 7 publications

(8 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The major differences in energy yield predictions occur only when the time step is raised to 600 s, likely due to the loss of information on the dynamics of the system. This result is in agreement with findings in [51] and [52], which highlight the inability of temporal resolutions above 600 s to capture the (BI)PV dynamics.…”

Section: Shading Effectssupporting

confidence: 91%

Simulating building integrated photovoltaic facades: Comparison to experimental data and evaluation of modelling complexity

2021

View full text Add to dashboard Cite

show abstract

Section: Shading Effectssupporting

confidence: 91%

Simulating building integrated photovoltaic facades: Comparison to experimental data and evaluation of modelling complexity

2021

View full text Add to dashboard Cite

show abstract

“…All the previous methods are optimized for the prediction of "power" and not "energy". As exhibited in (Anagnostos et al, 2017, Goverde et al, 2015, the thermal state of the PV system cannot ignored for time steps below 5min, as it augments inaccuracies in the final estimation. Another important piece of information that is rarely used is the type of weather itself.…”

Section: Related Workmentioning

confidence: 99%

“…One source of this offset has been recognized in (Schmidt, 2017) , where the bias of the forecasted GHI increases with the horizon in a similar trend. Another contribution to the bias has been described also in (Anagnostos et al, 2017), which is the inability of the parametric yield models to capture the thermal behavior of a PV system for high time resolutions. This inability usually appears as an offset on the power production estimation, due to the calculated operating temperature and can only be mitigated by a dynamic model, like the proposed one.…”

Section: B Comparison To Power Forecastmentioning

confidence: 99%

“…This is usually not an issue as most forecasts average data in time and space, therefore mitigating the issue, but for local control applications, more detailed time series are required. The developed model tackles the issue by using recursive networks, as suggested in (Anagnostos et al, 2017) and ensuring the correct estimation and retention of the thermal state of the PV system.…”

Section: B Comparison To Power Forecastmentioning

confidence: 99%

“…The proposed model utilizes a low cost sky-imager in order to produce intermediate inputs. These are then passed to a specialized cluster of neural networks that make use of previously published methods (Anagnostos et al, 2017, Goverde et al, 2016 in order to capture the full dynamic behavior of the installation, including its thermal state. As a result, a highly accurate, highly detailed forecast is produced (<20% nRMSE for the whole horizon), which can then be used for applications such as local grid control, energy storage management, energy pricing, data center scheduling etc.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A method for detailed, short-term energy yield forecasting of photovoltaic installations

Anagnostos

Schmidt²,

Cavadias

et al. 2019

Renewable Energy

Self Cite

View full text Add to dashboard Cite

The global shift towards renewable energy production suggests a promising future on reducing carbon emissions and avoiding the detrimental effects of global warming, but also creates additional challenges on all levels of energy production and distribution. The expected penetration of electric cars, increasing energy usage of cloud computing centers and the transformation of the electricity grid itself towards the "Smart Grid" requires novel solutions on all levels of energy production and management. Forecasting of energy production especially will become a major component for design and operation in all temporal and spatial scales, creating opportunities for optimized control of energy storage, local energy exchange etc. To this end, a method for the creation of detailed and accurate energy yield forecasts for PV installations is presented. Based on sky-imager information and using tailored neural networks, highly detailed energy yield forecasts are produced for a monitored test installation, for horizons up to 15 minutes and with a resolution of 1 second. The described method manages to outperform state of the art models by up to 39% in forecast skill, while at the same time retaining temporal resolutions that enable control schemes and energy exchange in a local scale.

show abstract

Neural Network Based Approaches for Fault Diagnosis of Photovoltaic Systems

Van Gompel,

Spina,

Develder

2024

Learning and Analytics in Intelligent Systems

View full text Add to dashboard Cite

Faults in photovoltaic (PV) systems due to manufacturing defects and normal wear and tear are practically unavoidable. The effects thereof range from minor energy losses to risk of fire and electrical shock. Thus, several PV fault diagnosis techniques have been developed, usually based on dedicated on-site sensors or high-frequency current and voltage measurements. Yet, implementing them is not economically viable for common small-scale residential systems. Hence, we focus on cost-effective techniques that enable introducing fault diagnosis without incurring costs for on-site sensor systems. In this chapter, we will present in particular two machine-learning-based approaches, built on recent neural network models. The first technique relies on recurrent neural networks (RNNs) using satellite weather data and low-frequency inverter measurements for accurate fault detection, including severity estimation (i.e., the power loss caused by the fault, usually not quantified in state-of-the-art methods in literature). The second technique is based on graph neural networks (GNNs), which we use to monitor a group of PV systems by comparing their current and voltage production over the last 24 h. By comparing outputs from multiple (geographically nearby) PV installations, we avoid any need for additional sensor data. Moreover, our results suggest that the GNN-based model can generalize to PV systems it was not trained on (as long as nearby sites are available) and retains high accuracy when multiple PV systems are simultaneously affected by faults.

show abstract

Systematic cross-validation of photovoltaic energy yield models for dynamic environmental conditions

Cited by 7 publications

References 13 publications

Simulating building integrated photovoltaic facades: Comparison to experimental data and evaluation of modelling complexity

Simulating building integrated photovoltaic facades: Comparison to experimental data and evaluation of modelling complexity

A method for detailed, short-term energy yield forecasting of photovoltaic installations

Neural Network Based Approaches for Fault Diagnosis of Photovoltaic Systems

Contact Info

Product

Resources

About