Lingen Luo

A Novel Partial Discharge Localization Method in Substation Based on a Wireless UHF Sensor Array

Effective Partial Discharge (PD) localization can detect the insulation problems of the power equipment in a substation and improve the reliability of power systems. Typical Ultra-High Frequency (UHF) PD localization methods are mainly based on time difference information, which need a high sampling rate system. This paper proposes a novel PD localization method based on a received signal strength indicator (RSSI) fingerprint to quickly locate the power equipment with potential insulation defects. The proposed method consists of two stages. In the offline stage, the RSSI fingerprint data of the detection area is measured by a wireless UHF sensor array and processed by a clustering algorithm to reduce the PD interference and abnormal RSSI values. In the online stage, when PD happens, the RSSI fingerprint of PD is measured via the input of pattern recognition for PD localization. To achieve an accurate localization, the pattern recognition process is divided into two steps: a preliminary localization is implemented by cluster recognition to reduce the localization region, and the compressed sensing algorithm is used for accurate PD localization. A field test in a substation indicates that the mean localization error of the proposed method is 1.25 m, and 89.6% localization errors are less than 3 m.

Partial Discharge Detection and Recognition in Random Matrix Theory Paradigm

The detection and recognition of partial discharge (PD) is an important topic in insulation tests and diagnoses. Take advantage of the affluent results from random matrix theory (RMT), such as eigenvalue analysis, M-P law, the ring law, and so on, a novel methodology in RMT paradigm is proposed for fast PD pulse detection in this paper. Furthermore, a scheme of time series modeling as random matrix is also proposed to extend RMT for applications with non-Gaussian noise context. Based on that, the eigenvalue distribution property is used for PD pattern recognition, which is completely new compared with traditional phase resolved PD and time-resolved PD methods. The simulation and experimental results show that the proposed methods are efficient, reliable, and feasible for PD detection and recognition especially for online applications.

Framework of Random Matrix Theory for Power System Data Mining in a Non-Gaussian Environment

A novel empirical data analysis methodology based on the random matrix theory (RMT) and time series analysis is proposed for the power systems. Among the ongoing research studies of big data in the power system applications, there is a strong necessity for new mathematical tools that describe and analyze big data. This paper used RMT to model the empirical data which also treated as a time series. The proposed method extends traditional RMT for applications in a non-Gaussian distribution environment. Three case studies, i.e., power equipment condition monitoring, voltage stability analysis and low-frequency oscillation detection, illustrate the potential application value of our proposed method for multi-source heterogeneous data analysis, sensitive spot awareness and fast signal detection under an unknown noise pattern. The results showed that the empirical data from a power system modeled following RMT and in a time series have high sensitivity to dynamically characterized system states as well as observability and efficiency in system analysis compared with conventional equation-based methods.

Data Mining and Principal Component Analysis Based Power Equipment Status Evaluation

The steady and stable operation of power equipment has a direct influence on the safety and stability of the electric system. This paper thoroughly analysed the mass data that reflect the status of power equipment by principal component analysis (PCA) on the basis of data mining technique, and established a principal component system that integrates key parameters of power equipment status together. It also established a comprehensive evaluation model of power equipment operating condition and achieved dynamic evaluation of power equipment health and rapid detection of abnormal condition. Transformer insulation status evaluation has been used as an example to prove that the method proposed in the paper may make up for the flaws of traditional status evaluation method and this method does offer certain reliability and practicability.

A Novel Error Correction Method for UHF Partial Discharge Localization

The error of partial discharge (PD) localization by UHF method will become gradually larger with the increase of source distance due to the UHF signal caused by PD is smaller or even undetectable. To improve the localization accuracy efficiently, this paper proposed a novel error correction method based on BP neural network. The measurement error dataset was trained by our designed algorithm and a compensation curved surface was drawn. The mean square root error of field test localization data was corrected by the compensation surface and the results proved that our proposed method could be used to improve the accuracy of UHF based PD localization for power equipment of power station.

RSSI fingerprinting-based UHF partial discharge localization technology

This paper proposes a partial discharge (PD) localization method based on received signal strength indication (RSSI) fingerprint technology which considers its both low hardware cost and good environmental adaptability features. The RSSI fingerprint library is built from PD measurements obtained by UHF sensors considering complex spatial characteristics of radio environment. a BP neural network based algorithm is proposed for accurate PD localization. The field test results show that the average localization error of proposed algorithm is 0.484 meter and 87% localization errors are less than 1 meter which prove the accuracy and stability of our proposed algorithm.

Partial discharge pattern recognition using random matrix theory

A new methodology of partial discharge (PD) pattern recognition under random matrix theory (RMT) paradigm is proposed in this paper. RMT has exhibited great potential values in the ongoing researches of big data in power systems applications. This paper extends RMT from Gaussian noise to non-Gaussian pattern by modeling time series as random matrix. Based on that, the characteristic parameters of PD signal are obtained from the eigenvalues distribution of assembled random matrix and corresponding covariance matrix. The BP neural network is used to perform the pattern recognition of six partial discharge models. The results show that the recognition rate of proposed method is above 85% and very robust to the pulse interference.