Investigating the impact of data quality on the energy yield forecast using data mining techniques

In this paper, we analysed the impact of using optimum combination of input variables and low dimensional subspace on Photovoltaic (PV) production forecasting accuracy. We worked in collaboration with Prof. Mussetta from Politecnico di Milano.

The main contribution presented in the paper is divided in two parts:

1. Optimum combination of input meteorological features using feature extraction technique
2. Low dimensional subspace using dimensional reduction technique

We assess and compare two cases when forecasting models are fed with all the features with the case when low subspace of dataset is used as an input to the models.

The simulation results reveal that depending on the location under study and the regression methods, using less variables as input to the forecasting models are enough to generate nearly similar results without affecting the performance. However, it is necessary to conduct the tests under different climatic conditions so as to ensure the reliability of the results. 

The figures below show the results obtained applying Pearsons correlation and principal component analysis. Figure 1 represents strength of association between two variables. Figure 2 is the biplot representation of the input features contributing variance on principal components PC1 and PC2.

Fig.1 : Pearson correlation map

Fig.2 : Biplot representation 

ISGT Europe 2020 was held virtually and we recorded the presentation for the same. The presentation is available at this link.

 

To support reproducibility and validating the results we have released the dataset utilized in the work along with the codes. 

Github repository with used dataset and evaluation code

For more information please see the paper:

Ekanki Sharma, Marco Mussetta, and Wilfried Elmenreich. Investigating the impact of data quality on the energy yield forecast using data mining techniques. In Proceedings of the IEEE PES Innovative Smart Grid Technologies Europe (ISGT-Europe). IEEE, October 2020.

Enable GingerCannot connect to Ginger Check your internet connection
or reload the browserDisable in this text fieldEditEdit in GingerEdit in Ginger×

Photovoltaics Energy Payback Time

Photovoltaic systems are great in producing clean energy without CO2 emissions. A question that remains however, is the amount of energy invested into production and transport of the materials, cells and panels. To answer this question, we had a look into the annual Photovoltaics Report of the Fraunhofer Society (German: Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e. V.).

To give a short summary: The energy payback time for current modules is around one to two years. Considering a lifetime of 20 years, this means that photovoltaic systems are quite effective in providing clean energy. The energy payback time depends mainly on three parameters: (i) the  material usage for the system, (ii) the effiency of the cells and panel, and (iii) the irradiation striking onto the panel. Regarding the first aspect, the report shows that material usage for silicon cells went down by a factor of 2.5 over the last ten years due to increased
efficiencies and thinner wafers. Efficiency is improving slower, currently the best multicrystalline modules provide an efficiency of 18.5%, panels with monocrystalline cells a 22.9% and upcoming thin film technologies have a range between 10.9% and 17.5%.

The largest influence for the energy payback time currently is the place where you put your module: In regions with an annual irradiation of around 1000 kWh/m² - this is basically the value for PV panels installed in Germany, they energy payback time is 2 years, while in sunny areas, the annual irradiation can be double or more, leading to an energy payback time of around 1 year.

Energy payback time for typical PV systems in different regions of Europe

Call for Papers on Optimization of Photovoltaic Power Systems

The Hindawi Jounral of Engineering is publishing a special issue on optimization of photovoltaic power systems. Journal of Engineering is a peer-reviewed, open access journal that publishes original research articles as well as review articles in several areas of engineering.

PV generation system is one of the most popular uses of direct solar energy and its installation is rapidly growing because it is considered as a clean and environmentally friendly source of energy. The primary obstacle to increased use of PV systems is their high initial cost. Currently, many research works are carried out focusing on optimization of PV systems in order to reduce the capital cost of the PV system without affecting its reliability. The optimization of a PV system means that the system parameters such as number of PV modules, capacity of storage battery, capacity of inverter, and PV array tilt angle must be selected optimally. In addition, diesel generator and wind turbine capacities must be optimized in case of hybrid PV systems. Moreover, the optimization term includes the electronic features that maximize the yield of these systems such as sun trackers, MPPT, and smart inverters.

This special issue aims to discuss the recent developed and contribution of photovoltaic system optimization science.

Potential topics include, but are not limited to:

• Modeling and characterization of photovoltaic systems
• Optimal sizing and installation of standalone photovoltaic system
• Optimal sizing of hybrid photovoltaic systems
• Optimal sizing of grid connected photovoltaic systems
• Optimal placement and management of photovoltaic systems in power system
• Power electronics for photovoltaic system
• Sun trackers
• MPPTs
• Solar inverters
• Solar chargers
• Photovoltaic field performance assessment
• Modeling of solar radiation

Authors can submit their manuscripts via the Manuscript Tracking System
Manuscript Due:    Friday, 29 January 2016
First Round of Reviews:    Friday, 22 April 2016
Publication Date:    Friday, 17 June 2016

Lead Guest Editor

Tamer Khatib, An-Najah National University, Nablus, State of Palestine

Guest Editors

Wilfried Elmenreich, Alpen-Adria-Universität Klagenfurt, Klagenfurt, Austria
Azah Mohamed, National University of Malaysia, Bangi, Malaysia
Suvash Saha, Queensland University of Technology, Brisbane, Australia
Hussein Kazem, Sohar University, Sohar, Oman

Modeling Solar Radiation

In our tutorial at the 2014 IEEE Innovative Smart Grid Technologies - Asia conference, Tamer Khatib and I covered the application of machine learning techniques for energy applications, in particular for modeling solar radiation. In the first part we explored meta-heuristic search algorithms and envisioned their application for designing distributed, self-organizing control systems using evolutionary algorithms. We provide an open-source software tool, FREVO, to conveniently apply this approach of finding the proper configuration of a local agent.

In the second part, we targeted the problem of solar radiation modeling. After stepping through different classical modeling approaches, we presented the possibility of using artificial neural networks to learn the correlation of input parameters such as latitude, longitude, temperature, humidity, month, day, hour to predict global and diffuse solar radiation. Experiments show that this method can achieve a high accuracy compared to existing models.

Links:

• Download FREVO (open source, platform-independent)
• Video: 6 minute introduction to FREVO
• A. Sobe, I. Fehérvári, and W. Elmenreich. FREVO: A tool for evolving and evaluating self-organizing systems. In Proceedings of the 1st International Workshop on Evaluation for Self-Adaptive and Self-Organizing Systems, Lyon, France, September 2012.
• I. Fehervari and W. Elmenreich. Evolution as a tool to design self-organizing systems. In Self-Organizing Systems, volume LNCS 8221, pages 139–144. Springer Verlag, 2014.
• T. Khatib, A Mohamed, K Sopian. A review of solar energy modeling techniques. J. of Renewable & Sustainable Energy Reviews. 2012.16(5): 2864-2869.
• T. Khatib, A. Mohamed, K. Sopian, M. Mahmoud. Assessment of Artificial Neural Networks for Hourly Solar Radiation Prediction. J. of Photoenergy. 2012. 2012(ID 946890):1-7.
• IEEE Innovative Smart Grid Technologies - Asia conference site

Energiewende - the Game

"Energiewende" is a computer game about the transition of our energy system from fossil fuels to renewable and sustainable energy sources. The game was developed by Manuel Herold, Matija Kucko, Andrea Monacchi, John N. A. Brown, and Wilfried Elmenreich as contribution to the CROSMOS GameJam. At this event teams had to develop a computer game to a topic just revealed at the contest in just two days - 2014 the topic was related to the services of Stadtwerke Klagenfurt. 

The game Energiewende let's you try out the strategic and tactical aspects of an electrical energy system that is mostly powered by renewable energy sources. Unlike a coal power plant, photovoltaic or wind power plants cannot save fuel in times of low load for using it later. Therefore, it is necessary to plan the distribution and placement of power plants well according to the expected power demands of their users - in the game they are modeled as houses with an energy consumption behavior of typical households. After placing power plants and transmission lines, the game features a real-time mode, where your system is simulated throughout three phases of a day (night/morning, daytime, evening). In case there is not enough energy in one phase, you must prevent a blackout (otherwise you loose the game) by balancing the grid. This is done by turning off devices in the houses. But be aware - user might not like this, especially if you turn off a device the user was about to use right now. A "complain-o-meter" is showing the aggregated dislikings of the users - once the complain-o-meter runs over, the game is lost.
Thus, one learns also about the typical consumption of devices and their potential for balancing the grid. The method shown in the game called demand response is currently a frequently discussed method for the future smart grid.

Got everything? Try out the game by clicking the image above. Have fun!

New Photovoltaic Power Plant

The smart micro grid lab has the major aim to provide the capability to pursue research and to teach students. To meet these requirements the lab is designed to establish a virtual common house with renewable energy sources and the ability to perform energy management tasks.

PV modules on the roof of the
 Lakeside Science and Technology Park, Klagenfurt, Austria

In detail, the lab finished the first construction phase and is now equipped with a photovoltaic power (PV) plant built up by the company Energetica. The PV has a capacity of 4.8kW and has the ability to act in the grid-connected mode, where energy can be produced, consumed and be fed into the grid and the island-mode, where the lab can be disconnected from the grid and be powered by the PV power plant and a battery buffering systems. To be able to simulate the lab in the grid-connected and the island-mode, the lab has the ability to switch from one mode to the next mode by demand. The concrete technical equipment is based on a Backup System of SMA, which delivers all necessary hardware to switch from island to grid-connected mode and to operate the PV power plant with the PV modules and the batteries.

PV performance over 3 days for cloudy and sunny days

In future, the smart micro grid lab will also be equipped with common household appliances which can be controlled and measured according to the consumed power.  With the ability to store and to generate energy, to connect the lab to and off the grid and to measure and to control appliances, the lab will have the ability to test and to develop algorithms and techniques to improve energy management systems of the future.

Links:

Energetica - energetica Energietechnik GmbH

SMA - SMA Solar Technology AG

Smart Microgrid Laboratory

Keeping an eye on my Photovoltaic System

Guest Post by Benjamin Steinwender

At my home place we have recently installed two photovoltaic systems with a total peak power production of 15 kW. From the first moment my dad confronted me with this idea, I wanted to record time series data of any measurement signal I could obtain from the system. Most promising was the fact of an included web server on the power converter.

Storage

Storing time series data over long periods can lead to high disk usage. Fortunately, I came across RRDtool (round-robin-database tool) - http://oss.oetiker.ch/rrdtool/ when I installed a monitoring solution for a FEM simulation server cluster and its UPS some time ago.

RRDtool stores data in a circular buffer so that its file size remains constant over time. However, older data will be overwritten after some time. The idea is to store the data in so-called round-robin archives with a consolidation function applied in order to keep the interesting properties. In my special application, the following archive sizes are used:

•     10080 x 1 minute samples for 7 days
•     11760 x 5 minute maximum & average values for 40 days, 20 hours
•     4704 x 30 minute maximum & average values for 98 days
•     2604 x 2 hour maximum & average values for ~ 7.2 months
•     2678 x 1 day maximum & average values for ~ 7.3 years

This gives plenty of storage with a reasonable file size of roughly 420 kB per data series.

Presentation

To ease the data presentation and acquisition, cacti http://www.cacti.net/ is used. This web-based application (favorable a Linux web server + PHP and MySQL required) is based on the RRDtool and provides Data Input Methods and Graph Templates as well as a nice graphical user interface to view the recorded data. Cacti primarily supports data acquisition via SNMP queries, but the power converter does not provide such interface. However, custom scripts can also be used.

Acquisition

Therefore, a PHP script has been written to fetch the current web page of the power converter and a simple HTML DOM (document-object-model) parser http://sourceforge.net/projects/simplehtmldom was used to extract the required field values. The tools run on a 24/7 NAS server where a cron job invokes the cacti poller (and thus the script) once every minute.

Results

Data acquisition is now successfully running since about 1 month. The total power production from last week can be seen in the following figure. The light-blue area represents the power output of the smaller (5 kW) system and the stacked dark-blue area is the power output of the larger (9.8 kW) system. From the data series we can even infer on the weather – the figure indicates that is was sunny at the weekend, cloudy on Monday, Tuesday and Friday and rainy on Wednesday and Thursday.