Building Robust Prediction Models for Defective Sensor Data Using Artificial Neural Networks

Claudio Frederico Pinho Rebelo de Sá, Arvind Kumar, Hugo Ferreira, Carlos Soares

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    30 Downloads (Pure)

    Abstract

    Sensors are susceptible to failure when exposed to extreme conditions over long periods of time. Besides they can be affected by noise or electrical interference. Models (Machine Learning or others) obtained from these faulty and noisy sensors may be less reliable. In this paper, we propose a data augmentation approach for making neural networks more robust to missing and faulty sensor data. This approach is shown to be effective in a real life industrial application that uses data of various sensors to predict the wear of an automotive fuel-system component. Empirical results show that the proposed approach leads to more robust neural network in this particular application than existing methods.
    Original languageEnglish
    Title of host publication14th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019)
    EditorsHéctor Quintián, José António Sáez Muñoz, Emilio Corchado, Francisco Martínez Álvarez, Alicia Troncoso Lora
    Pages142-153
    Number of pages12
    Volume950
    ISBN (Electronic)978-3-030-20055-8
    DOIs
    Publication statusE-pub ahead of print/First online - 1 May 2019
    Event14th International Conference on Soft Computing Models in Industrial and Environmental Applications 2019 - Seville, Spain
    Duration: 13 May 201915 May 2019
    Conference number: 14
    http://2019.sococonference.eu/

    Publication series

    NameAdvances in Intelligent Systems and Computing
    Volume950
    ISSN (Print)2194-5357
    ISSN (Electronic)2194-5365

    Conference

    Conference14th International Conference on Soft Computing Models in Industrial and Environmental Applications 2019
    Abbreviated titleSOCO 2019
    CountrySpain
    CitySeville
    Period13/05/1915/05/19
    Internet address

    Fingerprint

    Neural networks
    Sensors
    Automotive fuels
    Fuel systems
    Industrial applications
    Learning systems
    Wear of materials

    Cite this

    Pinho Rebelo de Sá, C. F., Kumar, A., Ferreira, H., & Soares, C. (2019). Building Robust Prediction Models for Defective Sensor Data Using Artificial Neural Networks. In H. Quintián, J. A. Sáez Muñoz, E. Corchado, F. Martínez Álvarez, & A. Troncoso Lora (Eds.), 14th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019) (Vol. 950, pp. 142-153). (Advances in Intelligent Systems and Computing; Vol. 950). https://doi.org/10.1007/978-3-030-20055-8_14
    Pinho Rebelo de Sá, Claudio Frederico ; Kumar, Arvind ; Ferreira, Hugo ; Soares, Carlos. / Building Robust Prediction Models for Defective Sensor Data Using Artificial Neural Networks. 14th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019). editor / Héctor Quintián ; José António Sáez Muñoz ; Emilio Corchado ; Francisco Martínez Álvarez ; Alicia Troncoso Lora. Vol. 950 2019. pp. 142-153 (Advances in Intelligent Systems and Computing).
    @inproceedings{681c0563854c4d16bcfdaddbbe276dff,
    title = "Building Robust Prediction Models for Defective Sensor Data Using Artificial Neural Networks",
    abstract = "Sensors are susceptible to failure when exposed to extreme conditions over long periods of time. Besides they can be affected by noise or electrical interference. Models (Machine Learning or others) obtained from these faulty and noisy sensors may be less reliable. In this paper, we propose a data augmentation approach for making neural networks more robust to missing and faulty sensor data. This approach is shown to be effective in a real life industrial application that uses data of various sensors to predict the wear of an automotive fuel-system component. Empirical results show that the proposed approach leads to more robust neural network in this particular application than existing methods.",
    author = "{Pinho Rebelo de S{\'a}}, {Claudio Frederico} and Arvind Kumar and Hugo Ferreira and Carlos Soares",
    year = "2019",
    month = "5",
    day = "1",
    doi = "10.1007/978-3-030-20055-8_14",
    language = "English",
    isbn = "978-3-030-20054-1",
    volume = "950",
    series = "Advances in Intelligent Systems and Computing",
    pages = "142--153",
    editor = "H{\'e}ctor Quinti{\'a}n and {S{\'a}ez Mu{\~n}oz}, {Jos{\'e} Ant{\'o}nio} and Emilio Corchado and {Mart{\'i}nez {\'A}lvarez}, Francisco and {Troncoso Lora}, Alicia",
    booktitle = "14th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019)",

    }

    Pinho Rebelo de Sá, CF, Kumar, A, Ferreira, H & Soares, C 2019, Building Robust Prediction Models for Defective Sensor Data Using Artificial Neural Networks. in H Quintián, JA Sáez Muñoz, E Corchado, F Martínez Álvarez & A Troncoso Lora (eds), 14th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019). vol. 950, Advances in Intelligent Systems and Computing, vol. 950, pp. 142-153, 14th International Conference on Soft Computing Models in Industrial and Environmental Applications 2019, Seville, Spain, 13/05/19. https://doi.org/10.1007/978-3-030-20055-8_14

    Building Robust Prediction Models for Defective Sensor Data Using Artificial Neural Networks. / Pinho Rebelo de Sá, Claudio Frederico; Kumar, Arvind; Ferreira, Hugo; Soares, Carlos.

    14th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019). ed. / Héctor Quintián; José António Sáez Muñoz; Emilio Corchado; Francisco Martínez Álvarez; Alicia Troncoso Lora. Vol. 950 2019. p. 142-153 (Advances in Intelligent Systems and Computing; Vol. 950).

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    TY - GEN

    T1 - Building Robust Prediction Models for Defective Sensor Data Using Artificial Neural Networks

    AU - Pinho Rebelo de Sá, Claudio Frederico

    AU - Kumar, Arvind

    AU - Ferreira, Hugo

    AU - Soares, Carlos

    PY - 2019/5/1

    Y1 - 2019/5/1

    N2 - Sensors are susceptible to failure when exposed to extreme conditions over long periods of time. Besides they can be affected by noise or electrical interference. Models (Machine Learning or others) obtained from these faulty and noisy sensors may be less reliable. In this paper, we propose a data augmentation approach for making neural networks more robust to missing and faulty sensor data. This approach is shown to be effective in a real life industrial application that uses data of various sensors to predict the wear of an automotive fuel-system component. Empirical results show that the proposed approach leads to more robust neural network in this particular application than existing methods.

    AB - Sensors are susceptible to failure when exposed to extreme conditions over long periods of time. Besides they can be affected by noise or electrical interference. Models (Machine Learning or others) obtained from these faulty and noisy sensors may be less reliable. In this paper, we propose a data augmentation approach for making neural networks more robust to missing and faulty sensor data. This approach is shown to be effective in a real life industrial application that uses data of various sensors to predict the wear of an automotive fuel-system component. Empirical results show that the proposed approach leads to more robust neural network in this particular application than existing methods.

    U2 - 10.1007/978-3-030-20055-8_14

    DO - 10.1007/978-3-030-20055-8_14

    M3 - Conference contribution

    SN - 978-3-030-20054-1

    VL - 950

    T3 - Advances in Intelligent Systems and Computing

    SP - 142

    EP - 153

    BT - 14th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019)

    A2 - Quintián, Héctor

    A2 - Sáez Muñoz, José António

    A2 - Corchado, Emilio

    A2 - Martínez Álvarez, Francisco

    A2 - Troncoso Lora, Alicia

    ER -

    Pinho Rebelo de Sá CF, Kumar A, Ferreira H, Soares C. Building Robust Prediction Models for Defective Sensor Data Using Artificial Neural Networks. In Quintián H, Sáez Muñoz JA, Corchado E, Martínez Álvarez F, Troncoso Lora A, editors, 14th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019). Vol. 950. 2019. p. 142-153. (Advances in Intelligent Systems and Computing). https://doi.org/10.1007/978-3-030-20055-8_14