Dust InSMS: Intelligent soiling measurement system for dust detection on solar mirrors using computer vision methods

https://doi.org/10.1016/j.eswa.2022.118646Get rights and content

Highlights

  • A new method for soiling quantification is proposed based on CNN approach.

  • Experimental measurements of a Fresnel solar field are conducted to collect data.

  • The innovation, software and hardware of Dust InSMS sensor are described.

  • A good agreement is proven during outdoor validation and tests of the Dust InSMS.

  • An optimal cleaning scenario is developed based on genetic algorithms.

Abstract

The dust accumulation strongly impacts the optical efficiency of solar concentrators, in particular the reflectivity of solar mirrors. Therefore, reducing the impact of reflectivity losses due to soiling and optimizing cleaning strategy are key factors. In this paper, the impact of dust accumulation on the reflectivity parameter of Fresnel mirrors is studied at the GEP research platform during the dry period. Based on the collected data, a new system for dust detection is proposed based on the classification approach using the convolutional neural networks and image processing algorithms in which no similar work is presented in the literature that uses the same approach to quantify the soiling phenomenon on CSP mirrors. The test loss and accuracy obtained by the proposed model are respectively 0.28 and 0.96. The outdoor validation results obtained so far suggest that the Dust InSMS concept and method could be a promising efficient and low-cost sensor. As the proposed system performs the GPS coordinates for each measurement, an optimal cleaning scenario is developed based on genetic algorithms to optimize the cleaning scenario and to come up with the shortest cleaning path.

Introduction

The majority of solar programs have been setting up in the sun-abundant regions of North Africa, Middle East, United States, India, and China with hundreds of gigawatts of capacity. These geographic sites are characterized by harsh climate conditions, such as dust storms, heavy pollution, and high airborne and settled particulate environments. Therefore, the majority of sites suitable for CSP plants are located in arid or semi-arid zones, where they suffer not only from these harsh climate conditions, but also from the water resource scarcity that limits the cleaning and restoration methodologies. The soiling or dust accumulation has an impact on the optical efficiency of solar concentrators and in particular on the reflectivity of solar mirrors. The dust particles accumulated on the surface of the mirrors deviate and affect the collected incoming direct solar irradiance, where a portion is absorbed by these particles and the rest reflected but not on the right focus (not on the receiver). Consequently, the reflectivity losses due to the soling phenomenon impact strongly the optical efficiency and reduce the entire productivity of the solar power plant.

Several researchers have studied the impact of soiling on the optical efficiency and on the overall productivity of the solar plant. Azouzoute et al. evaluated the impact of soling on both PV and CSP technologies considering the reflectance and the transmittance drop of mirror and glass samples exposed at GEP research platform during the dry period (Azouzoute et al., 2020). The soiling drop of mirror samples reached up to 35 %, while it reached up to 12 % for glass samples. They concluded that CSP mirrors are three times more impacted by soiling than PV glass samples. Furthermore, Bouaddi et al. investigated the impact of soiling on the mirror and aluminum samples exposed at two sites in southwest Morocco (Agadir and Tantan) (Bouaddi et al., 2017). For 12 weeks of exposure, the mirror samples lost up to 73 % of their initial cleanliness. In addition, the authors concluded that mirror samples are more affected by soiling than the aluminum during the dry season without rain. Merrouni et al. also studied the effect of soiling on mirror and aluminum samples, but this time considering different directions and four different angles of exposure (Merrouni et al., 2015). The effect of angles has also been investigated by (Heimsath et al., 2016). Bellmann et al. also compare simultaneously the effect of soiling under the same conditions for both PV and CSP (Bellmann et al., 2020). They exposed solar glass and mirror samples in southern Portugal and found an 8 to 14 factor of soiling rates between CSP and PV. Moreover, Raillani et al. proposed a techno-economic study of the impact of optical soiling losses on Fresnel and solar tower power plants based on experimental and numerical investigations (Raillani et al., 2022). Costa et al. provided an updated state of the art of the solar energy dust and the impact of soiling on both PV and CSP technologies by reviewing and classifying the majority of recent publications (Costa et al., 2016, Costa et al., 2018). Recently, Conceiçao et al. published a detailed description, comprehensive, and contextualization of the research evolution of soiling impact from 1942 to 2019 (Conceição et al., 2022).

The cleaning process is the best solution to overcome the soiling phenomenon by removing dust particles and reducing the reflectivity loss. Cleaning activities are frequent and represent a significant portion of the total O&M costs. Therefore, optimizing cleaning frequency and strategy reduces the cleaning and O&M costs. An optimal cleaning strategy or schedule guarantees the balance between productivity losses and cleaning costs. In order to define the optimal cleaning schedule, the O&M team should have an in-depth knowledge of soiling mechanisms and ensure frequent soiling measurements of the solar field. For this reason, few sensors have been proposed to measure and quantify the soiling loss (will be detailed in the next section). On the other hand, some researchers and institutes have chosen to rely on modeling the soiling phenomenon based on historical meteorological data and modeling methods to select the optimal cleaning frequency (Bouaddi et al., 2015, Bouaddi et al., 2018, Conceição et al., 2018, Zitouni et al., 2020). These models are site-specific by introducing a general cleaning schedule of the solar plant and do not consider the soiling distribution throughout the same solar field. Therefore, direct measurements using soiling sensors remain very accurate and can introduce a more detailed cleaning schedule considering the soiling distribution on the solar field since the solar collectors are not impacted by the same amount of dust.

Measuring the reflectivity of solar mirrors and quantifying the soiling issue using existing sensors have many limitations related either to the time required or to the cost of the sensor. Three types of sensors can be found so far. The spectrometers with high accuracy but are limited to laboratory uses and cannot be used in real working conditions. The reflectometers are suitable for on-site measurement and are portable, yet these sensors are time consuming and require several measurements to quantify soiling on a mirror, which makes the assessment of several mirrors very difficult task and time consuming. The TraCS sensor is a non-portable device and could continuously measure the soiling while installing several sensors is very expensive. This work introduces an easier and faster method for direct measurement of soiling or reflectivity loss of CSP mirrors. The main advantages of this method are the rapidity of measurements and the simplicity of arrangements using RGB photo to estimate the reflectivity of mirrors rather than relying on several measurement points and then considering their average. In addition, the proposed sensor also uses the GPS coordinates of each measurement to perform the shortest cleaning path and understand the soiling distribution on the solar field. In this respect, this paper will be organized as follows: in the second section, the existing sensors for measuring the reflectivity and quantifying the soiling phenomenon of solar mirrors will be reviewed and classified. Following, the methodology will be introduced in the third section, including the system description and the measurement methods. In the fourth section, the main results and performance evaluation of the proposed system will be discussed. An application based on the system measurements will be presented in order to select the shortest cleaning path in the fifth section. Finally, the main conclusions and perspectives will be presented and discussed.

Section snippets

Background and related work

Specular reflectivity of solar mirrors is the key element in CSP plants represented by the conversion of solar irradiation into heat and then into electricity. In addition, specular reflectivity is used to monitor the soiling phenomenon in CSP plants over time, which plays an important role in resource assessment studies and cleaning cycle optimization. For these reasons, reflectivity measurements are essential and should be with high accuracy and with less arrangement. As reflectivity

System description

Dust InSMS is proposed to rapidly and precisely calculate the mirrors reflectivity only in one shot using an RGB image. The system is developed and validated at the Green Energy Park research facility. The Fresnel mirrors are employed with the Reflectometer “D&S R15” to establish this work. The system can easily be adapted to other mirrors, such as parabolic trough technology. This solution is proposed to overcome the difficulties encountered by the O&M team at the GEP. Traditionally, before

Classification and model training

After gathering a database of mirror images along with their parameters, such as reflectivity, dust density and cleaning cycle, each class is labeled in a folder to train a classification model based on the deep learning algorithms, particularly the Convolutional Neural Network (CNN) Architecture. Image classification defined as categorizing a set of images into one or several predefined classes. The difficulty of classifying images differs from one application to another depending on the

Application

The main purpose of this subsection is to provide a very advantageous application of the Dust InSMS system. As already mentioned, the proposed system aims to rapidly and precisely determine the reflectivity of the mirrors and then to investigate the soiling phenomenon. One of the important usefulness of the dust InSMS system is to consider the GPS coordinates of each measurement. Based on these GPS coordinates and the soiling values of each measurement, a soiling map of the entire solar field

Conclusions and perspectives

In this paper a new system to evaluate the dust accumulation effect on the solar mirrors was presented. The Dust InSMS system can be classified into the direct measurement sensors of soiling quantification by calculating the reflectivity of the solar mirrors. In addition to the reflectivity parameter, this system can also estimate the dust density accumulated on the mirror surface, as well as the cleaning cycle which indicates how much the mirror is exposed on the solar field after the last

CRediT authorship contribution statement

Massaab El Ydrissi: Conceptualization, Methodology, Software, Writing – original draft. Hicham Ghennioui: Validation, Supervision, Writing – review & editing. El Ghali Bennouna: Visualization, Investigation, Validation. Azouzoute Alae: Visualization, Investigation, Validation. Mounir Abraim: Data curation, Software. Ibrahim Taabane: Data curation, Software. Abdi Farid: Supervision, Validation.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

We appreciatively acknowledge the research platform Green Energy Park (GEP), Benguerir Morocco, for its support and reception.

References (59)

  • S.C. Costa et al.

    Solar energy dust and soiling R&D progress: Literature review update for 2016

    Renewable and Sustainable Energy Reviews

    (2018)
  • A. Fernández-García et al.

    Equipment and methods for measuring reflectance of concentrating solar reflector materials

    Solar Energy Materials and Solar Cells

    (2017)
  • H. Kim et al.

    A study on path optimization method of an unmanned surface vehicle under environmental loads using genetic algorithm

    Ocean Engineering

    (2017)
  • A.A. Merrouni et al.

    Investigation of soiling effect on different solar mirror materials under Moroccan climate

    Energy Procedia

    (2015)
  • A. Mouaky et al.

    Simulation and experimental validation of a parabolic trough plant for solar thermal applications under the semi-arid climate conditions

    Solar Energy

    (2019)
  • F. Wolfertstetter et al.

    Monitoring of mirror and sensor soiling with TraCS for improved quality of ground based irradiance measurements

    Energy Procedia

    (2014)
  • H. Zhang et al.

    Path planning for the mobile robot: A review

    Symmetry

    (2018)
  • Al-Saffar, A. A. M., Tao, H., & Talab, M. A. (2017). Review of deep convolution neural network in image classification....
  • S.M. Anwar et al.

    Medical image analysis using convolutional neural networks: A review

    Journal of Medical Systems

    (2018)
  • Cardinale-Villalobos, L., Meza, C., & Murillo-Soto, L. D. (2020). Experimental comparison of visual inspection and...
  • T. Carneiro et al.

    Performance analysis of google colaboratory as a tool for accelerating deep learning applications

    IEEE Access

    (2018)
  • Devices and Services. (2009). Portable Specular Reflectometer Model 15R-USB. Operation and Maintenance Manual. Dallas:...
  • P. Elmer

    High Performance Lambda Spectrophotometers

    (2007)
  • Freiburg: PSE & AG; (2016). PFlex. A new portable soiling/cleanliness handheld...
  • Google Colaboratory. (n.d.). Retrieved 20 September 2020, from...
  • He, K., Zhang, X., Ren, S., & Sun, J. (2016a). Deep residual learning for image recognition. 770–778....
  • He, K., Zhang, X., Ren, S., & Sun, J. (2016b). Identity mappings in deep residual networks. 630–645....
  • Heimsath, A., Lindner, P., Klimm, E., Schmid, T., Moreno, K. O., Elon, Y., Am-Shallem, M., & Nitz, P. (2016). Specular...
  • Hinton, G. E., Srivastava, N., Krizhevsky, A., Sutskever, I., & Salakhutdinov, R. R. (2012). Improving neural networks...
  • Cited by (7)

    • Cloud-edge collaborated dust deposition degree monitoring for distributed photovoltaic systems

      2023, International Journal of Electrical Power and Energy Systems
    View all citing articles on Scopus
    View full text