Use this URL to cite or link to this record in EThOS:
Title: Advanced statistical process control methods for PMU-based loss-of-mains detection
Author: Guo, Yuanjun
ISNI:       0000 0004 5988 914X
Awarding Body: Queen's University Belfast
Current Institution: Queen's University Belfast
Date of Award: 2015
Availability of Full Text:
Full text unavailable from EThOS.
Please contact the current institution’s library for further details.
High penetration levels of distributed renewable generation has brought considerable issues and challenges to the operation and control of power systems for countries and regions where there are abundant, but geographically distributed, renewable energy resources. One of such challenging issue is the Loss-of-mains (LOM) and islanding detection. This thesis aims at developing a variety of monitoring methods to tackle this problem. Principal Component Analysis (PCA) method is a linear method which assumes that data follows a normal Gaussian distribution. Therefore, for a Gaussian distributed frequency variable, a PCA model can be calculated with corresponding confidence limits, which is then used for detecting abnormal transients and identifying the islanding sites occurring in the UK utility network. In order to monitor the dynamic characteristics of power systems, a dynamic PCA (DPCA) approach is proposed by incorporating time lags at the modelling stage, thus the auto- and cross-correlations can be extracted. Moreover, power system processes are time-varying, thus a recursive PCA (RPCA) monitoring scheme is also proposed as a reliable extension of PCA to achieve adaptive updating of training data and confidence limits. A novel radial basis function neural network (RBFNN) model-based PCA method is then proposed to monitor non-Gaussian variables in power systems, combing with the newly developed Teaching-Learning-Based-Optimisation (TLBO) method to tune the non linear parameters of RBF neurons. This research has shown that statistical process control methods such as PCA and its extensions can be successfully applied to power system loss-of-mains detection, so that large volumes of PMU data can be efficiently processed and an adequate monitoring model can be built for a given confidence limit. They further provide a wide-area view and early accurate warning of the network for system operators, with the length of time to identify faults and the risks of damaging connected equipment being reduced.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available