^{1}

^{1}

A new methodology for the detection and identification of insulator arc faults for the smart grid environment based on phasor angle measurements is presented in this study and the real time phase angle data are collected using Phasor Measurement Units (PMU). Detection of insulator arcing faults is based on feature extraction and frequency component analysis. The proposed methodology pertains to the identification of various stages of insulator arcing faults in transmission lines network based on leakage current, frequency characteristics and synchronous phasor measurements of voltage. The methodology is evaluated for IEEE 14 standard bus system by modeling the PMU and insulator arc faults using MATLAB/Simulink. The classification of insulator arcs is done using Support Vector Machine (SVM) technique to avoid empirical risk. The proposed methodology using phasor angle measurements employing PMU is used for fault detection/classification of insulator arcing which further helps in efficient protection of the system and its stable operation. In addition, the methodology is suitable for wide area condition monitoring of smart grid rather than end to end transmission lines.

Smart Grid is considered as a future of power grid which is able to cater the production, transmission and distribution of generated electricity by modern technology to resolve many issues in modern power grid. The permanent and temporary transmission line faults can have huge impact on operational stability of Smart Grid such that these faults lead to the islanding of faulted system [

Majority of investigations focuses on detection/classification of permanent faults which include symmetric and asymmetric faults [

Considering these facts, the paper work deals with a new methodology for the detection and classification of insulator arcing faults for the smart grid environment based on phasor angle measurements. The research work includes the following aspects: Section 2 describes about insulator arc modeling at dynamic conditions. Phasor Measurement Technology and the real time phasor angle estimation using Phasor Measurement Units (PMU) are dealt in Section 3. In Section 4, the detection and classification of insulator arc faults in smart grid network are based on feature extraction of frequency components using IEEE 14 bus; and in Section 5, SVM technique used for classification of insulator arc faults is explained. Finally verification of simulation results for insulator arc faults for smart grid IEEE 14 bus using SVM is done using MATLAB/Simulink.

A theory which converts flashover mechanism to measurable parameters accounting the instantaneous changes of an arc suitable for flashover prediction is developed in this section. The leakage current during an arc is pre-do- minantly random in nature in any insulator and their dynamic properties are evaluated in terms of this leakage current. Surface discharges in the form of intermittent arcs, sustained arcs, etc. precede contamination flashover and it is mandatory to develop a computer model of the arc which could aid in understanding of the flashover process. For a given insulator profile (supply voltage pollution severity and arc length), the model estimates the voltage drop induced by the arc, arc resistance, etc.

A dynamic arc model devised from the external electric circuit which has an arc resistance in series with a pollution resistance, is supplied by a sinusoidal voltage. The arc is a single dominant sinusoidal voltage and the pollution resistance represents the pollution layer of the unbridged portion of the insulator. The assumptions made are such that source impedance is neglected, arc appears only along the surface of the insulator and the pollution layer is uniform [

The arc resistance per unit length is obtained dynamically from Mayr’s equation. The developed arc model with analytically calculated parameters given in

ESDD (mg/cm^{2}) | σ_{s} (μS) | R_{po}_{ln} (Ω/cm) | L_{arc} (cm) |
---|---|---|---|

0.05 | 5 | 27,240 | 12.7 |

0.1 | 10 | 13,620 | 9.655 |

0.15 | 15 | 8840 | 6.52 |

The arc conductance and Equivalent Salt Deposition Density (ESDD) can be given by (3)

where σ_{s} is the arc conductance in Siemens [

A pure sinusoidal waveform can be represented by a complex number which is known as a phasor. Consider a sinusoidal signal

The magnitude is the rms value of the sinusoid

However for any computational purpose, a unified reference signal of considerable accuracy is required from which the relative phase angles can be determined. This is because to make the phasor quantities to be made com- parable with other quantities measured at various locations around the network. Phasor estimation is initiated with sampling of original input signal at a rate of N-samples per cycle which constitutes a sample window. In order to achieve the phasor (magnitude and angle) of the input signal, Discrete Fourier Transform (DFT) is applied on the

above sample window.

The dynamics of equivalent power factor angle (EPFA) at the time of any disturbance can be achieved only by rotating phasor, which is not the case of commercial PMUs [

At nominal frequency of the grid, the mapping of three phase synchronous voltage and current measurements to a rotating reference frame is done to estimate three phase equivalent voltage and current phasors. Using Park’s Transformation, These three phase quantities are mapped to two phase orthogonal quantities which can be given by (6)-(9), where V_{d}_{0} and V_{q}_{0} stand for direct and quadrature axes voltage, I_{d}_{0} and I_{q}_{0} stand for current in rotating reference frame.

During normal operating conditions, V_{RN}, V_{YN} and V_{BN}, the peak values of phase voltage are equal to each other whereas I_{RN}, I_{YN} and I_{BN} are peak values of three-phase line current respectively. The power factor angle between line voltage and current is termed as ϕ. The frequency of three-phase voltage and rotating reference frame are termed as ω and

The equivalent voltage phasor (V_{0}) and current phasor (I_{0}) form angles γ and η, respectively, with rotating axes d and q, which can be calculated as

The equivalent phase angle waveforms are then analyzed although FFT over a time span of (1/f) to estimate the frequency coefficients. The frequency coefficients corresponding to 0 Hz of equivalent voltage and current phase angle variations are represented by α_{γ} and α_{η} respectively which is shown in (12) and (13) as

For achieving the dynamics of these angles, voltage and current phasor have to be estimated for every sample window that consists of N number of samples of the original waveform. In MATLAB tool, the tangent operation can be performed by using trigonometric functions block. Major advantage of this transformation for classification problem is that only two-phase data are required to process for analyzing three-phase system. In addition, for identification of temporary faults such arcs, insulator discharge, etc. this transformation provides most promising results and the same can be observed in the simulation results.

Formulation of the proposed insulator arc identification and classification methodology based on phase angle oscillations of equivalent voltage and current phasor values is presented here.

PMU installed at generator bus estimates the voltage and current phasors from measurements from capacitor voltage transformer (CVT) and current transformer (CT) whose schematic diagram is shown in

the arc is initiated at various instants, the sinusoidal supply voltage V(t) is injected into the system bus. The equivalent voltage and current phasors are measured by modeled PMU. The frequency coefficients are extracted by FFT analysis. It is observed from extensive case studies that FFT spectrum consists of frequency coefficients corresponding to 0 Hz only and all other frequency coefficients remain at zero during normal operating conditions.

However, during fault conditions frequency coefficients corresponding to 0 and 100 Hz have severe dependency on the type of insulator arc. So from FFT analysis, the percentage magnitude of harmonics at various frequencies are noted and analysed.

The above methodology is implemented in the IEEE 14 bus system with inducing the modeled arc in the trans- mission line between 4 and 5. The analysis is done for four different cases

1) System under normal condition

Under normal conditions, the leakage current across the insulators are almost zero and the system operates under stable condition.

2) System under sustained arc condition

The sustained arc is characterized by a leakage current magnitude of about 100 mA as shown in

3) System under insulator flashover condition

The insulator flashover is characterized by a leakage current magnitude of about 200 mA as shown in

SVM is one of the predominant machine intelligence techniques for fault classification in power systems which have less empirical risk [

with following constraints (15)

The Kernel function used is radial basis function which is given as

σ is a positive integer.

The training data (x_{i}, y_{i}) have been categorized with sustained arc, flashover or normal condition labels. The set of equivalent phase angle coefficients (α_{γ} and α_{η}) and normalised three-phase voltage coefficients and total harmonic index forms the input data set for SVM classifier. The labeled fault type data corresponding to each set of values in input data set constitutes the output data set for SVM classifier. SVM classifier is trained with 1000 data sets to frame boundary for arc fault classification. The input data sets are generated for voltage and current phasor angle at 0 Hz, voltage and current phasor angle at 100 Hz and total harmonic index for synchronous voltage.

The proposed fault classification methodology has been implemented in MATLAB/SIMULINK environment for SVM training and classification. Radial basis function has been utilized as the kernel function which is defined as (16) [

The results of dq0 voltage, equivalent voltage phasor angle and FFT analysis for normal operation of the system, sustained arc and during insulator flashover are shown. The frequency analysis characteristics are tabulated at the end of the detection section. In addition, the SVM classification results of insulator arcs based on their frequency coefficients are discussed in this section.

The synchronous dq0 voltage, equivalent voltage phasor angle is computed using Park’s transformation. The input data for this computation is produced from the modeled PMU installed at generator bus 2 in IEEE 14 bus system. From the synchronous voltage, FFT analysis tool in MATLAB/Simulink is used for estimating total harmonic index. The modeled insulator arcs such as sustained arc and insulator flashover is induced in the transmission line insulators between bus 4 and bus 5 in the IEEE 14 bus system.

Figures 9(a)-(c) shows the synchronous dq0 voltage with respect to reference axes, estimated equivalent voltage phasor angle and FFT analysis results for the system under normal condition. The total harmonic index is 3.83% which is very well less than 5%. The results give a clear idea that for the stable operation of the system.

During sustained arc condition, the insulator arc remains from 0.18 to 0.42 seconds, whose dq0 voltage is shown in

The insulator flashover occurs at 0.26 seconds and thus the dq0 voltage under this condition gets increased from 1 pu to 1.2 pu as shown in

the range of 29% as in

The results for first order, third order and DC component values for all the discussed cases are tabulated in

The synchronous voltage and current phasor angle along with THD values of synchronous voltage has higher impact on the arc characteristics. So these values are chosen as input data for SVM training and classification. The synchronous phasor angle ranges from 0.36% to 0.5% for sustained arc condition whereas for normal condition, the variation is around 0.25% which is shown from the vector machines in

Cases Considered | %Mag of 1^{st} Order Harmonics | %Mag of 3^{rd} Order Harmonics | DC Component | THD in % |
---|---|---|---|---|

Normal Condition | >0.25 | >0.2 | Yes | 3.83 |

Sustained Arc | 0.1 | 0.04 | Yes | 70.76 |

Insulator Flashover | 0.08 | 0.02 | Yes | 29.46 |

The synchronous phasor value for flashover condition ranges from 0.2% to 0.3% which is evident from support vectors shown in

This paper proposes a real time insulator arc identification and detection methodology which paves a better way for condition monitoring of transmission line insulators of smart grid system. The major contribution of this work is PMUs installed in the smart grid for the measurement purposes could also be optimally used for monitoring the transmission line insulators and special devices or sensors are hardly required for the insulator condition monitoring. In addition, the methodology can be utilized for the purpose of scheduling preventive maintenance of insulators before any approaching insulator damage. Hence the significance of the work is to assist the wide area protection scheme of the Smart Grid systems with synchro phasor measurements done by PMU installed at the generator bus. Here the methodology to identify the location of insulator arc fault is not included which can be considered for future work.

Kailasam Saranya,Chinnusamy Muniraj, (2016) A SVM Based Condition Monitoring of Transmission Line Insulators Using PMU for Smart Grid Environment. Journal of Power and Energy Engineering,04,47-60. doi: 10.4236/jpee.2016.43005