Development of smart online partial discharge monitoring system for medium voltage power cable

Received Apr 1, 2019 Revised Jun 30, 2019 Accepted Jul 11, 2019 This paper presents, the development of smart online partial discharge (PD) monitoring system for medium voltage (MV) underground power cable. PD monitoring is the highly efficient tool to monitor insulation degradation for high voltage (HV) equipment in order to avoid failures or breakdown. Selection of improved technology and performance of PD detection sensor, effective measurement technique and smart user friendly of graphical user interface (GUI) system are contributed towards the development of efficient monitoring system. This paper addresses three main aspects which are needed in completing the monitoring system. They are, the use of Rogowski coil (RC) as detection sensor, processing unit using Alterra board and integrated with GUI PD monitoring system for underground cable using LabVIEW. The monitoring system is compared to the conventional method with the PD signal used is measured from the real on-site measurement in order to analyse its performance. The analysis is performed in MATLAB and LabVIEW software’s environment and the maximum peak of PD signal is enabled to view using GUI which complete with the location information of PD source. Furthermore, this paper has contributed to solve the problem in selection the simplified and practical approach for PD sensor and monitoring system. In the perspective of automated condition monitoring, this smart online PD monitoring system provides a complete solution towards latest industrial revolutions


INTRODUCTION
Condition monitoring technology plays a major role to keep the electrical supply network in reliable operation. The electrical distribution over the world are operated at 24 hours and 7 days a week simultaneously. Therefore, various electrical power components faced with environmental stresses [1]. Condition monitoring in power system requires best infrastructure of components and equipment which is reliable when effective maintenance applied [2]. Early information for potential breakdown in on-line mode especially for PD in HV equipment is consider crucial [3]. Therefore, the utility providers keep on striving determine solutions to improve the reliability in continuity of the power supply [4].
MV underground power cable is among the key component in distribution networks. Normally, this component is undergoing stress condition includes, electrical, thermal, mechanical and ambient, stresses, that significantly toward to the degradation of its insulation [5,6]. The degradation insulation in MV underground power cable is highly proven originated from PD activities. PD is a localize discharge that only partially bridges the insulation between the conductor and which may or may not occur adjacent to a conductor [7]. The high accuracy of PD monitoring system is needed to capture PD activity since it occurs in short duration of current pulses, normally for only a few nanoseconds [8]. Referring to the data from [9], PD usually occurs before breakdown happened indicate that PD monitoring can be a good system to installed. The data collected through PD monitoring system also can give a real time of insulation system healthy. Thus, the reliability, efficiency, equipment safety and cost can be improved. Traditional method is using off-line measurement where it will interrupt the normal operation. There will be more economical, if the equipment can be tested without interrupt the whole system. Thus, online PD monitoring is the best way to ensure the performance of PD analysis efficiently [10]. The real-time PD monitoring and activity using simulation LabVIEW software is used for further processed [11]. The biggest advantage of a GUI is that it works without prior knowledge and it is the most important aspect for any application [12][13][14].
Currently, conventional PD sensor are saturated, big size, bulky, low sensitivity and high cost such as High Frequency Current Transformer (HFCT) and Coupling Capacitor (CC) compared to Rogowski coil (RC) [15]. PD activity will be measured thru current pulse when RC is placed around the cable with supply make it convenient to use [16]. While during on-line measurement also have limitation such as need manually collecting data and require a lot of time and human force [17][18][19][20]. Therefore, in this paper, an attempt is made to develop a smart online partial discharge monitoring system for high voltage underground cable complete with hardware and graphical user interface (GUI). The peak detector and thresholding technique were a used and indicate good result compare to others,

RESEARCH METHODS
This monitoring system is developed based on the combination of hardware, software and PD measurement setup at on-site. The RC is used as a measuring sensor for the measurement. Figure 1 shows the complete diagram for PD monitoring system at MV distribution network system. The system is comprised with three RC sensors, labelled as RC set A, RC set B and RC set C, looped along 5 km power cable. Each RC is integrated with nano5 which is able to transmit data to the server or cloud. Then, monitoring system using GUI which is under LabVIEW interface will always work in real time at maintenance office thru Wi-Fi connection to receive and process the data from measuring sensors. The PD activity usually start with small condition and expand rapidly if there are no prevention action taken and will end up with fault or breakdown in the whole system. Therefore, monitoring procedure must be performed in real time in order to determine the actual condition of power cable.

Sensors for Partial Detection (PD) Detection
Table1 represents the comparison between different type of measuring sensors for PD detection. It shows that the electromagnetic detection using modified RC address many advantages in terms of price, weight, performance and installation method compared to the HFCT and CC sensors [21,22].

Partial Discharge (PD) Detection Technique
A fault detection technique was developed in order to improve accuracy and at lower cost. The peak detection flow under processing unit is as shown in Figure 2. The processing flow is started with detection counter in term of peak value setting versus time in Alterra board. The PD signal will be declared occur when the peak signal is more than threshold value set in the processing unit. The process is also confirmed by the control unit [23,24]. This detection applied the concept of eliminating low voltage signal as shown in Figure  3. Finally, the data is sent to the cloud through Wi-Fi. Hence, the ADC topology is a novel by implementing in wireless RC sensor development based on PD detection technique and become the preferred solution [19].  (1) (2) (3)and (4) was used to simulate the exact signal to fulfil PD characteristic and then use for PD localization analysis [25]. The different time arrivals between sensor A and B is represented by tab, and between sensor B and sensor C is represented by tcb respectively.
( 1 ) where A is 0.01magnitude coefficient a1= 1Ms -1 , a2= 10Ms -1 , φ tan w a , wd = 2 , fd = 1 MHz Figure 4 shows the internal block diagram of PD monitoring system under LabVIEW software environment. Several types of graphical functions were used, such as formula waveform vi, subtraction and multiplication, exponential, spectral measurement vi to complete the circuit connection. Besides, the specific input from measurement of sensor A,B,C will analyse and perform using algorithm to produce signal that give PD signal with the highest peak.  Figure 5(a) shows the complete of PD monitoring system hardware for 11 kV power cable. The measuring sensor developed can be used for the power cable size up to 180 mm 2 . While Figure 5(b) represents the schematic diagram for the monitoring system developed. RC sensor is the main component for the PD detection circuit. and the Altera DE2-115 board system for the processing unit. The nano5 unit is integrated with each RC sensor in order to transmit signal through Wi-Fi to the cloud. The algorithm is integrated on Alterra board which complete with PD peak detector, analogue to digital conversion and Wi-Fi receiver.

Graphical User Interface (GUI) for Partial Discharge Monitoring System
(a) (b) Figure 5. (a) Smart online PD monitoring system hardware, (b) Schematic diagram. Figure 6 shows the actual PD signal waveform measured using RC at site. Meanwhile Figure 7 shows the peak voltage of the highest PD signal measured from the measurement. This waveform is captured by RC and trimmed by the processing unit in order to identify the PD signal. The processing unit helps to identify the highest PD signal because the peak signal is set to triggered at certain level, system make easier to identify the highest PD signal. The selection of N-bit resolution value result is 0.029 V. Then, Figure 8 shows the conversion from analogue signal to digital waveform. Although, the higher resolutions have a better in minimum digital voltage, but it is difficult to identify the greatest peak voltage due to the value is really close, compared with 16-bit voltage resolutions. The analog signal and digital signal is compared in term of the maximum peak signal by following (4) where, the output of Altera DE0-nano board is 3.3 V. Based on the results presented, it is conclude that low resolutions has higher number of samples which more accurate in term of the sampling signal.  Figure 9 represents, the complete GUI for smart online PD monitoring system where it capable to shows the detected online PD signal in real time condition depend on three located RC sensors and the distance of PD signal detected was automated calculate and indicate in this platform. Result indicates that the peak amplitude of PD signal within 4.3 metre detected along 5 km 11 kV power cable. Although the result indicate several different in term of peak amplitude, but the figure of peak and value still significant to be used. Three exact location of the PD is displayed individually depend on sensor A, B and C located at power cable. Comparison between three sensor location indicated that the nearest PD signal is exist at 1.3-meter form sensor B. Next, the maintenance unit can easily perform their task to solve this problem. Therefore, the feasibility and validity of this monitoring system were verified through real on-site measurement. Figure 9. The complete GUI of online PD monitoring system

CONCLUSION
This paper presents the development of smart online PD monitoring system for medium voltage power cable. The RC sensor's developed function as a powerful component for PD detection was employed with wireless system using raspberry pi and monitored by smart interface using LabVIEW making the system valuable for on-line of a power cable. It shows that the PD signal in power cable can be detected by the RC sensor and notify the user or utility wirelessly by the monitoring system. Finally, the application of smart online PD monitoring can be applied and will help to improve the capability of PD detection in the medium voltage power cable.