Original papers
Peach growth cycle monitoring using an electronic nose

https://doi.org/10.1016/j.compag.2019.104858Get rights and content

Highlights

  • New device for in-loco application (orchard) for maturation phase analysis.

  • Comparison of the results of the methods used (KNN, SVM, RF, ELM).

  • Pearson's Chi-square test to remove inconsistent and redundant attributes (13–7).

  • Dimensionality reduction with PCA and LDA.

  • Improved accuracy of classifications with proposed methods.

Abstract

In regions with the predominance of agriculture, an inspection of the quality and fruit maturity index in the orchard is usually analyzed by the farmer’s experience, which can be subject to errors and generate a greater cost of time and money. Thus, monitoring equipment that generates a rapid and accurate response to the growth cycle of the peaches in the crop is desirable, together with a low marketing cost. For this purpose, electronic noses prove to be the most suitable equipment, since it allows online monitoring of the VOCs (Volatile Organic Compounds) generated by the crop. In this context, a prototype was developed to perform the classification of the fruit growth cycle (pre-harvest and post-harvest). Models with the 13 gas sensors made with a metal oxide semiconductor (MOS) and the reduction to 7 sensors were studied with the aid of the Pearson’s Chi-square test, for comparison. Samples with 4 growth stages were used for the training and construction of the model. The accuracy of 99.23% in the validation step and 98.08% in the sample test step using the Random Forest method with linear discriminant analysis for the reduced data set for 7 sensors shows that the device is promising for monitoring of areas with an intense emission of VOCs.

Introduction

Fruits produce different volatile organic compounds (VOCs) and their quality measured by aroma, taste and color changes constantly throughout the growth and maturation phase, which occurs in the pre-harvest period (Baietto and Wilson, 2015). When measuring these properties, some instrumental methods used are manual and destructive. Non-destructive measurement of internal fruit quality is becoming important for industry and for consumers (Rajkumar and Wang, 2012). The development of sensor technology allowed the electronic noses to be presented as simple devices with high detection accuracy, and these devices are increasingly being used as an alternative to traditional methods (Wu et al., 2017, Zhu et al., 2017, Lin and Zhang, 2016, Jiang and Wang, 2016, Men et al., 2018). Moreover, electronic noses can also be applied to the monitoring of air quality (Deshmukh et al., 2015, Abraham and Pandian, 2013, Bagula et al., 2012, Peterson et al., 2017, Kim and Hwangbo, 2018, Blanco-Novoa et al., 2018, Laref et al., 2018), gases emitted by the soil (Bieganowski et al., 2018, Sudarmaji and Kitagawa, 2016, Dorji et al., 2017, Pineda and Pérez, 2017), food quality (Aleixandre et al., 2015, Wojnowski et al., 2017, Chen et al., 2018, Srivastava et al., 2019, Mishra et al., 2018), beverage quality (Aleixandre et al., 2018, Wei et al., 2017, Santos and Lozano, 2015, Ab Mutalib and Jaswir, 2013, Nurul et al., 2017, Ragazzo-Sanchez et al., 2008), in the medical field (Huang et al., 2018, Voss et al., 2012, Li et al., 2017, Lorwongtragool et al., 2014), among others (see Fig. 1).

Electronic noses are made up of an array of chemical sensors based on metal oxide materials (MOS). In the same way as human noses, these devices are not able to identify the substances separately in each sample. It may make an analogy of the data obtained by these sensors with the fingerprint since it is very difficult to find two different substances with the same pattern, that is, it is possible to classify substances according to their patterns (Pineda and Pérez, 2017). These devices consist of three parts: sensor array, signal processing unit and pattern recognition. These three parts simulate, respectively, the acquisition of information by sensory neurons of the human olfactory receptor, the encoding of the olfactory nerve (bulb), brain memory and information processing by the human olfactory system (Huang et al., 2018).

The biological mechanism is a common indication of fruit ripeness. In places where agriculture predominates the inspection of the quality and maturation of the fruits is given by the experience of the farmer, which can be subject to errors and carelessness (Chen et al., 2018). In addition, if a specific period for fruit harvesting is considered, there may be situations where the fruits are of different qualities at different points in the planted area, which is part of precision agriculture. In view of this, it is desirable to have equipment that monitors the different degrees of maturation at the cultivation points and provided this can be identified by VOCs, the electronic noses are a robust and low-cost alternative for this purpose, besides allowing the monitoring of culture.

Based on gas chromatography and mass spectrometry (GC–MS) (Horvat et al., 1990, Baraldi et al., 1999) recognized the main VOCs exhaled by peach during and after the full bloom phase. Table 1 shows the main gases generated as by-products in this process.

Currently, the peach crop has expanded, with the planting of more than 500 thousand seedlings/year in southern Brazil (Dutra et al., 2002). Considering the physicochemical characteristics of the peach fruit growth, for this work was carried out the monitoring of the cultivar Eragil in the city of Ponta Grossa, Paraná. Thus, non-destructive equipment was used, applied near the crop and capable of detecting the gases exhaled by peach, to identify changes in the different cycles, from growth, maturation, and post-harvest.

Section snippets

State-of-the-art

There are many papers in the literature applying electronic noses to qualitatively discriminate peaches. Brezmes et al. (2000) observed the ripening process of peach fruits. Based on an arrangement of chemical tin oxide sensors and pattern recognition techniques based on neural networks, the designed olfactory system can classify fruit samples into three different maturation states. Measures made with peaches show a success rate above 93% (green, ripe and super mature). An additional feature of

Materials and methods

The methodology performed in this work is basically divided into four stages: equipment construction, execution of the experiment protocol, pre-processing and data analysis.

Results and discussion

Fig. 7(a) illustrates typical sensor responses for an experiment measured on the peach-tree in stage 1. Remembering that the first 600 s comprise the preheating time and the transient state period and were removed from the data analysis. The figure shows the response obtained by the first set of data.

Fig. 7(b)–(d) illustrates the typical responses of the sensors to an experiment measured on the peach in stage 2, stage 3, and stage 4, respectively. In the same way, the initial 600 s comprise the

Conclusions

To develop equipment capable of classifying peach growth phases, the equipment developed with the 13 MOS sensors and the ventilation system with ambient air was successful in discriminating these phases. Samples with the four growth stages that were used for the training and construction of the model prove to be an adequate methodology in the application of a device for this purpose. Thus, farmers have the possibility of obtaining a robust and low-cost device capable of determining the

Acknowledgments

The authors thank the Brazilian financial agency (CNPq) for the scholarships, grant number [312104/2015-4].

References (53)

  • S. Srivastava et al.

    Fuzzy controller based E-nose classification of Sitophilus oryzae infestation in stored rice grain

    Food Chem.

    (2019)
  • C. Visai et al.

    Volatile compound production during growth and ripening of peaches and nectarines

    Sci. Horticult.

    (1997)
  • H. Zhang et al.

    Application of electronic nose and statistical analysis to predict quality indices of peach

    Food Bioprocess Technol.

    (2012)
  • J. Zhu et al.

    Evaluation of the synergism among volatile compounds in Oolong tea infusion by odour threshold with sensory analysis and E-nose

    Food Chem.

    (2017)
  • Ab Mutalib, N.A., Jaswir, I., Akmeliawati, R., 2013. IIUM-fabricated portable electronic nose for halal authentication...
  • Abraham, K., Pandian, S., 2013. A low-cost mobile urban environmental monitoring system. In: 2013 4th International...
  • M. Aleixandre et al.

    A wireless and portable electronic nose to differentiate musts of different ripeness degree and grape varieties

    Sensors

    (2015)
  • M. Aleixandre et al.

    Quantification of wine mixtures with an electronic nose and a human panel

    Front. Bioeng. Biotechnol.

    (2018)
  • A. Bagula et al.

    Ubiquitous sensor networking for development (usn4d): an application to pollution monitoring

    Sensors

    (2012)
  • M. Baietto et al.

    Electronic-nose applications for fruit identification, ripeness and quality grading

    Sensors

    (2015)
  • W. Barbosa et al.

    Desenvolvimento dos frutos e das sementes de pêssegos subtropicais de diferentes ciclos de maturação

    Pesquisa Agropecuária Brasileira

    (1993)
  • W. Barbosa et al.

    Ecofisiologia do desenvolvimento vegetativo e reprodutivo do pessegueiro em região subtropical

    (1990)
  • A. Bieganowski et al.

    Evaluation of hydrocarbon soil pollution using E-nose

    Sensors

    (2018)
  • O. Blanco-Novoa et al.

    A cost-effective IoT system for monitoring Indoor radon gas concentration

    Sensors

    (2018)
  • L.-Y. Chen et al.

    Development of a dual MOS Electronic nose/camera system for improving fruit ripeness classification

    Sensors

    (2018)
  • Couto, M., 2006. Efeito da temperatura durante a diferenciação de gemas, floração, crescimento e desenvolvimento de...
  • Cited by (39)

    View all citing articles on Scopus
    View full text