Operational Risk Evaluation Of Cable Plugs Using An Automated Multisensor Classification System

Konferenz: VDE-Hochspannungstechnik 2018 - ETG-Fachtagung
12.11.2018 - 14.11.2018 in Berlin, Deutschland

Tagungsband: ETG-Fb. 157: VDE-Hochspannungstechnik

Seiten: 6Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Boettcher, Bjoern; Sinai, Ali; Menge, Matthias; Graef, Thomas; Huecker, Thomas (Hochschule für Technik und Wirtschaft, Berlin, Germany)
Plath, Ronald (Fachgebiet Hochspannungstechnik, Technische Universität Berlin, Germany)

The integrity of high-voltage (HV) insulation can be evaluated by measuring partial discharge (PD) activity. PD detection and monitoring is usually performed based on a single physical sensor and is susceptible for noise interference. In order to increase the evaluation reliability, an automated multi sensor expert system for PDs at cable terminations of medium-voltage gas-insulated switchgears is presented. The objective of this work is to combine data from different sensors that capture various physical quantities, i.e. sound and electrical wave to improve reliability. The introduced system includes an intelligent classification algorithm, which classifies the detected signals from an acoustic sensor and a high frequency current transformer (HFCT). The system utilizes three redundant classification algorithms, which work independently and thus make the evaluation result even more reliable. For the classification features the Fourier transformations of the PD apparent charge Q, the voltage gradient DeltaU/Deltaphi and the voltage change DeltaU between pulses are used. An L2 distance classifier examines the data of corresponding PD defect classes in respect to certain probabilities.