Generative Adversarial Model-Guided Deep Active Learning for Voltage Dip Labelling

Document identifier: oai:DiVA.org:ltu-76910
Access full text here:10.1109/PTC.2019.8810499
Keyword: Engineering and Technology, Electrical Engineering, Electronic Engineering, Information Engineering, Other Electrical Engineering, Electronic Engineering, Information Engineering, Teknik och teknologier, Elektroteknik och elektronik, Annan elektroteknik och elektronik, Automatic Labelling, Deep Active Learning, Deep Learning, Generative-Discriminative Model, Semi-supervised Training, Voltage Dip, Electric Power Engineering, Elkraftteknik
Publication year: 2019
Relevant Sustainable Development Goals (SDGs):

The SDG label(s) above have been assigned by OSDG.ai

Abstract:

In many real applications, the ground truths of class labels from voltage dip sequences used for training a voltage dip classification system are unknown, and require manual labelling by human experts. This paper proposes a novel deep active learning method for automatic labelling of voltage dip sequences used for the training process. We propose a novel deep active learning method, guided by a generative adversarial network (GAN), where the generator is formed by modelling data with a Gaussian mixture model and provides the estimated probability distribution function (pdf) where the query criterion of the deep active learning method is built upon. Furthermore, the discriminator is formed by a support vector machine (SVM). The proposed method has been tested on a voltage dip dataset (containing 916 dips) measured in a European country. The experiments have resulted in good performance (classification rate 83% and false alarm 3.2%), which have demonstrated the effectiveness of the proposed method.

Authors

Azam Bagheri

Luleå tekniska universitet; Energivetenskap
Other publications >>

Irene Y.H. Gu

Department of Electrical Engineering Chalmers University of Technology Gothenburg, Sweden
Other publications >>

Math Bollen

Luleå tekniska universitet; Energivetenskap
Other publications >>

Record metadata

Click to view metadata

	
    header: 
        identifier: oai:DiVA.org:ltu-76910
        datestamp: 2021-05-12T23:06:02Z
        setSpec: SwePub-ltu
    metadata: 
        mods: 
            @attributes: 
                version: 3.7
            recordInfo: 
                recordContentSource: ltu
                recordCreationDate: 2019-11-28
            identifier: 
                http://urn.kb.se/resolve?urn=urn:nbn:se:ltu:diva-76910
                10.1109/PTC.2019.8810499
                2-s2.0-85072337776
            titleInfo: 
                @attributes: 
                    lang: eng
                title: Generative Adversarial Model-Guided Deep Active Learning for Voltage Dip Labelling
            abstract: 
In many real applications the ground truths of class labels from voltage dip sequences used for training a voltage dip classification system are unknown and require manual labelling by human experts. This paper proposes a novel deep active learning method for automatic labelling of voltage dip sequences used for the training process. We propose a novel deep active learning method guided by a generative adversarial network (GAN) where the generator is formed by modelling data with a Gaussian mixture model and provides the estimated probability distribution function (pdf) where the query criterion of the deep active learning method is built upon. Furthermore the discriminator is formed by a support vector machine (SVM). The proposed method has been tested on a voltage dip dataset (containing 916 dips) measured in a European country. The experiments have resulted in good performance (classification rate 83% and false alarm 3.2%) which have demonstrated the effectiveness of the proposed method.

            subject: 
                    @attributes: 
                        lang: eng
                        authority: uka.se
                    topic: 
                        Engineering and Technology
                        Electrical Engineering Electronic Engineering Information Engineering
                        Other Electrical Engineering Electronic Engineering Information Engineering
                    @attributes: 
                        lang: swe
                        authority: uka.se
                    topic: 
                        Teknik och teknologier
                        Elektroteknik och elektronik
                        Annan elektroteknik och elektronik
                    @attributes: 
                        lang: eng
                    topic: Automatic Labelling
                    @attributes: 
                        lang: eng
                    topic: Deep Active Learning
                    @attributes: 
                        lang: eng
                    topic: Deep Learning
                    @attributes: 
                        lang: eng
                    topic: Generative-Discriminative Model
                    @attributes: 
                        lang: eng
                    topic: Semi-supervised Training
                    @attributes: 
                        lang: eng
                    topic: Voltage Dip
                    @attributes: 
                        lang: eng
                        authority: ltu
                    topic: Electric Power Engineering
                    genre: Research subject
                    @attributes: 
                        lang: swe
                        authority: ltu
                    topic: Elkraftteknik
                    genre: Research subject
            language: 
                languageTerm: eng
            genre: 
                conference/other
                ref
            note: 
                Published
                3
                
ISBN för värdpublikation: 978-1-5386-4722-6

            name: 
                    @attributes: 
                        type: personal
                        authority: ltu
                    namePart: 
                        Bagheri
                        Azam
                    role: 
                        roleTerm: aut
                    affiliation: 
                        Luleå tekniska universitet
                        Energivetenskap
                    nameIdentifier: 
                        azabag
                        0000-0001-8504-494x
                    @attributes: 
                        type: personal
                    namePart: 
                        Gu
                        Irene Y.H.
                    role: 
                        roleTerm: aut
                    affiliation: Department of Electrical Engineering Chalmers University of Technology Gothenburg Sweden
                    @attributes: 
                        type: personal
                        authority: ltu
                    namePart: 
                        Bollen
                        Math
                    role: 
                        roleTerm: aut
                    affiliation: 
                        Luleå tekniska universitet
                        Energivetenskap
                    nameIdentifier: 
                        matbol
                        0000-0003-4074-9529
            relatedItem: 
                    @attributes: 
                        type: host
                    genre: grantAgreement
                    name: 
                        @attributes: 
                            type: corporate
                        namePart: Energimyndigheten
                        role: 
                            roleTerm: fnd
                    @attributes: 
                        type: host
                    titleInfo: 
                        title: 2019 IEEE Milan PowerTech
            originInfo: 
                dateIssued: 2019
                publisher: IEEE
            physicalDescription: 
                form: print
            typeOfResource: text