A New Method to Improve the Sensitivity of Leak Detection in Self-Contained Fluid- Filled Cables L. Hao 1, P. L. Lewin 1, S. G. Swingler 1 and C. Bradley.

Slides:



Advertisements
Similar presentations
An Improved TCP for transaction communications on Sensor Networks Tao Yu Tsinghua University 2/8/
Advertisements

Moisture effect on the dielectric response and space charge behavior of mineral oil impregnated paper insulation Jian Hao1, 2, George Chen2, Ruijin.
Thermal performance of high voltage power cables James Pilgrim 19 January 2011.
Influence of Ageing and Temperature on the Polarization/Depolarization Current Behaviour of Paper Immersed in Natural Ester Jian Hao 1, 2*, Ruijin Liao.
Smart Materials as Intelligent Insulation A. F. Holt, A. C. Topley, R. C. D. Brown, P. L. Lewin, A. S. Vaughan, P. Lang * University of Southampton, Southampton,
The Discrete Wavelet Transform The discrete wavelet transform is formed by passing the signal through a series of filters. The signal is passed through.
MEP201 Mechanical Engineering Drawing 1st semester
Running a model's adjoint to obtain derivatives, while more efficient and accurate than other methods, such as the finite difference method, is a computationally.
ECG Signal processing (2)
Classification / Regression Support Vector Machines
Managing Director, HVPD Ltd
Concept Design Review (CoDR) Shore Station DC Breaker Cable model Transient Analysis Components.
Maximum Covariance Analysis Canonical Correlation Analysis.
Smart Sensor Technology For Rail Network health Hawk Measurement Systems.
High Sensitivity External Leak Detection for Liquid Fuel Pipelines David Parman P.Eng. Ken McCoy.
Smart Sensor Technology for Gas Pipeline Monitoring
Partial discharge analysis of defective three-phase cables An overview of an ongoing project involving the generation and analysis of PD within distribution.
High Throughput Computing and Protein Structure Stephen E. Hamby.
Nexans experience in deployment of sensors in subsea cable systems
CHAPTER 9 Electrical Design of Overhead Lines
Electrical Systems Electrical Engineering in Railways –Rolling Stock –Communications & Control –Signalling –LV Installations in Stations and Buildings.
Cross Strait Quad-Regional Radio Science and Wireless Technology Conference, Vol. 2, p.p. 980 – 984, July 2011 Cross Strait Quad-Regional Radio Science.
Lecture 3: Bridge Circuits
Attention Deficit Hyperactivity Disorder (ADHD) Student Classification Using Genetic Algorithm and Artificial Neural Network S. Yenaeng 1, S. Saelee 2.
Features: ERTL TESTED  Displays distance to leak location in meters.  Indicates Cable break condition.  Modbus Output for BMS Monitoring.  Operates.
Author: Sotetsu Koyamada, Yumi Shikauchi, et al. (Kyoto University)
Prediction model building and feature selection with SVM in breast cancer diagnosis Cheng-Lung Huang, Hung-Chang Liao, Mu- Chen Chen Expert Systems with.
Levels of high voltage: World over the levels are classified as: LOW MEDIUM HIGH EXTRA and ULTRA HIGH Voltages However, the exact magnitude of these levels.
 For AC ramp breakdown testing a Phenix AC Dielectric Test Set, Type 600C was used with a custom built test cell.  The test cell used mushroom electrodes.
23 rd World Gas Conference 5 – 9 June 2006 Amsterdam Full Scale Qualification Trials and Certification of the AMPLITUDE-LNG LOADING SYSTEM Phillip COX.
SMART SENSORS FOR A SMARTER GRID BY KAILASH.K AND LAKSHMI NARAYAN.N.
GA-Based Feature Selection and Parameter Optimization for Support Vector Machine Cheng-Lung Huang, Chieh-Jen Wang Expert Systems with Applications, Volume.
Dynamic Thermal Ratings for Overhead Lines Philip Taylor, Irina Makhkamova, Andrea Michiorri Energy Group, School of Engineering Durham University.
Kernel Methods A B M Shawkat Ali 1 2 Data Mining ¤ DM or KDD (Knowledge Discovery in Databases) Extracting previously unknown, valid, and actionable.
Influence of Copper on the By-products and Dielectric Properties of Different Oil-paper Insulations Jian Hao 1, 2, Ruijin Liao 1, George Chen 2 and Chao.
Space Charge Behaviour of Mineral Oil/Kraft Paper Insulation with Different Ageing Conditions Jian Hao 1, 2*, G Chen 2, Ruijin Liao 1 and Wei Li 1 1 University.
Review of Current Conditions Report and Work Plan for Area 1 Presented by The Great Plains/Rocky Mountain Technical Outreach Services for Communities.
Project logo / LP logo EUROPEAN UNION GOVERNMENT OF ROMANIA SERBIAN GOVERNMENT Structural Funds Common borders. Common solutions. Romania – Republic.
Numerical Modelling of Needle-Grid Electrodes for Negative Surface Corona Charging System Y. Zhuang*, G. Chen and M. Rotaru University of Southampton,
I. M. DMYTRAKH and V. V. PANASYUK Karpenko Physico-Mechanical Institute, National Academy of Sciences of Ukraine 5 Naukova Street, Lviv, 79601, UKRAINE.
Reservoir Uncertainty Assessment Using Machine Learning Techniques Authors: Jincong He Department of Energy Resources Engineering AbstractIntroduction.
Infra-red Technique for Damage Tolerant Sandwich Structures W.Wang 1 J.M.Dulieu-Barton 1, R.K.Fruehmann 1 and C.Berggreen 2 1 Faculty.
GENDER AND AGE RECOGNITION FOR VIDEO ANALYTICS SOLUTION PRESENTED BY: SUBHASH REDDY JOLAPURAM.
Lecture 3: Bridge Circuits
SEMINAR PRESENTATION ON 220 KV GSS UDAIPUR
A confocal Raman microprobe analysis of partial discharge activity in gaseous voids N A Freebody 1*, A SVaughan 1, G C Montanari 2 and L Wang 2 1 University.
SVMs in a Nutshell.
4/28/2017 Stress Corrosion Cracking Assessment in Pipeline Mohammed Abu Four October 11, 2010.
Ying Yi PhD Lab 3 Resistance and Ohm’s Law 1 PHYS HCC.
Improvement of SSR Redundancy Identification by Machine Learning Approach Using Dataset from Cotton Marker Database Pengfei Xuan 1,2, Feng Luo 2, Albert.
High resolution product by SVM. L’Aquila experience and prospects for the validation site R. Anniballe DIET- Sapienza University of Rome.
BIOINFORMATION A new taxonomy-based protein fold recognition approach based on autocross-covariance transformation - - 王红刚 14S
HNC/D Engineering Science
Measurement Of Resistance
TRANSDUCERS PRESENTATION BY: Dr.Mohammed Abdulrazzaq
Transmission Terminations and Joints
Bridging Domains Using World Wide Knowledge for Transfer Learning
AHMEDABAD INSTITUTE OF TECHNOLOGY
HVDC LIGHT:- NEW TECHNOLOGY FOR A BETTER ENVIRONMENT
Design of the thermosiphon Test Facilities 2nd Thermosiphon Workshop
Leakage control in transmission pipelines
How SCADA Systems Work?.
Hao Zhai, Hao Yi, Zhirong Zeng, Zhenxiong Wang, Feng Wang, Fang Zhuo
Subsea 100 Global Services Products Presentation
The Quench Detection-Wire-Feedthrough Plug-In of W7-X
Overview Key diagram of 220 KV Sub station, Ablowal Cables
Use of Lightning Data for Electricity Transmission Operations
GAUSSIAN PROCESS REGRESSION WITHIN AN ACTIVE LEARNING SCHEME
Shih-Wei Lin, Kuo-Ching Ying, Shih-Chieh Chen, Zne-Jung Lee
Model generalization Brief summary of methods
Presentation transcript:

A New Method to Improve the Sensitivity of Leak Detection in Self-Contained Fluid- Filled Cables L. Hao 1, P. L. Lewin 1, S. G. Swingler 1 and C. Bradley 2 1 University of Southampton, Southampton, UK 2 National Grid, UK Cable Monitoring System (Drallim) Raw Data and Data Pre-processing Introduction Cable and Cable Route Results Conclusions Fluid-filled cables are the most widely used type of transmission cable in power networks due to their outstanding performance and long service history. As a fluid filled system, the cable circuit may have the potential to leak due to damage caused by unforeseen circumstances such as environmental effects, manual intrusions, installation problems or manufacturing defects. Leakage from a fluid-filled cable may have great impact on the reliability of network operation and cause environment contamination. Therefore, detection and location of leaks along the cable route is of interest to system operators. Research to date has considered the detection of leakage from high pressure fluid-filled (pipe type) cables. Many methods have been investigated. However, due to the sensitivity or feasibility of these proposed methods, they have not been widely applied in the field. System operators need to be convinced that reliable detection or location of leak is readily achievable. Moreover, there is little published research on detection of leaks for low pressure (self- contained) fluid-filled cables, due to the difficulties in measurement compared with pipe type cables. Proposed methods to date require rearrangement of the cable circuit and additional equipment whilst taking the circuit out of service. A method of real-time detection of leaks for self-contained fluid-filled cables without taking them out of service has been assessed and a novel machine learning technique has been employed. This approach is based on the analysis of the measureable physical parameters of a 400 kV oil-filled cable system, in terms of pressure, temperature and load current, obtained from sensors of the existing condition monitoring system. A regression analysis based on the use of the Support Vector Machine technique is employed to predict future oil pressure trends in the cable system. University of Southampton, Highfield, Southampton, SO17 1BJ, UK Contact details : Figure 2 Schematic diagram of Drallim cable monitoring system 400 kV 2000 mm 2 copper conductor Polypropylene paper laminate (PPL) insulation Corrugated seamless aluminium (CSA) sheath PVC anti-corrosion over sheath Improved sensitivity compared to the existing pressure falling and low alarm system The use of DTS and RTTR system may provide improved detection sensitivity and feasibility of locating leaks 5.7 km total length Double circuits Two cables per phase 10 straights and 1 stop joint per cable Figure 1 Schematic diagram of the arrangement of the cable circuits Pressure Remote Digital Transducers (RDT) 4 pressure RDTs/ cable × 12 cables = 48 pressure RDTs Temperature RDTs 4 oil tank temperature RDTs 2 ambient temperature RDTs 1 ground temperature RDT Current RDTs 1 current RDTs/cable × 12 cables = 12 current RDTs Figure 3 Load current in cables of circuit 1 group A Figure 4 Pressure in cables of circuit 1 at north compound Figure 5 Temperature of cables of circuit 1 at north compound Load current 1 A CT resolution Pressure 0.1 kPa pressure RDT sensitivity Temperatuer 0.1 °C (K) temperature sensitivity Sampling interval 2 hours (nominal) Support Vector Machine Regression for Data Analysis The Support Vector Machine (SVM) is a method for finding functions from a set of labelled training data. The function can be either a classification function or a regression function. This learning machine uses a central concept of linear function (classification SVC and regression SVR) and kernel mapping for a number of learning tasks. SVR Training Gaussian Radial Basis Function: K(x i, x j ) = exp(-γ||x i -x j || 2 ) Cross-Validation: 5-fold cross-validation Grid-search : γ and C Training data: 4 days Figure 6 SVM regression Figure 7 Training accuracies for cable 1_1_R_A_NC Figure 8 Predicted and measured pressure for cable 1_1_R_A_NC (4 days training data) Figure 9 Prediction error rates for cables 1_1_RYB_AB_NC Figure 10 Adjusted prediction error rates for cables 1_1_RYB_AB_NC Error rate is calculated: P p – predicted pressureP m – measured pressureP F – pressure falling alarm Adjusted error rate is calculated: