KR20220036126A - System and Method for monitoring partial discharge using sensor data - Google Patents

Abstract

A partial discharge detection system utilizing sensor data that can simultaneously analyze multiple online data to compensate for the shortcomings of offline analysis and the inherent limitations of single online sensor data is disclosed. The sensor data utilization partial discharge detection system is a sensor system that generates a plurality of sensor data by sensing a distribution line switch, converts the plurality of sensor data obtained from the sensor system into characteristic data, and uses the characteristic data to detect the sensor data. It is characterized by comprising a terminal that determines whether a partial discharge or noise has occurred in the distribution line switch, and a network that communicates and connects the sensor system and the terminal.

Description

{System and Method for monitoring partial discharge using sensor data}

The present invention relates to partial discharge detection technology, and more specifically, to a partial discharge detection system and method that can analyze multiple online data simultaneously to compensate for the shortcomings of offline analysis and the inherent limitations of single online sensor data. .

Substation is located between the transmission line and distribution line and converts high transmission voltage to low distribution voltage. For this purpose, a switchgear is installed in the substation to connect or separate the transmission line and distribution line. This device is called a distribution line switchgear.

Distribution line switchgear is divided into hydraulic, magnetic, and pneumatic types depending on its type. In particular, gas insulated switchgear (GIS) using SF6 gas, which has excellent dielectric strength, is widely used.

However, even if the insulation performance of the gas insulated load switch device is excellent, due to the nature of the power system, a single insulation accident can have a very large ripple effect such as social chaos or economic loss, so thorough inspection to prevent insulation accidents is essential.

For this reason, research on technology for automatically detecting partial discharge phenomenon, which is a typical cause of deterioration of gas-insulated load switch devices, is being widely conducted. Partial discharge (PD) is a discharge phenomenon that occurs locally around or inside an insulator under high voltage stress. It is often caused by a naturally formed physical gap (void) or a partial crack (crack) caused by the deterioration of the switchgear. ) occurs in

Partial discharge results in continuous power loss due to power leakage, and if the effects accumulate, it can cause irreversible physical or chemical changes to the insulating material. Therefore, when partial discharge occurs, power supply through distribution lines may be completely stopped or equipment may explode.

Several methods are being used to detect such partial discharge (PD). For understanding, examples include technology to detect partial discharge at the hardware level using sensors, technology to detect partial discharge using online single sensor data, and technology to detect using offline equipment.

However, in the case of partial discharge (PD) detection technology at the hardware level utilizing sensors, When partial discharge (PD) occurs in a gas insulated switch (GIS), energy emission is detected in the ultra-high frequency (UHF) band. Therefore, the same phenomenon occurs even in the case of noise rather than actual PD, which has the disadvantage of having a high false detection rate.

Additionally, in the case of PD detection technology utilizing online single sensor data, Partial discharge is detected through analysis of PRPD (Pulse-Resolved Partial Discharge) pattern of UHF (Ultra High Frequency) band signal. Therefore, in the case of low-resolution online data, there is a disadvantage that it is impossible to detect noise of the actual partial discharge-like waveform.

In addition, in the case of PD detection technology using offline equipment, PD detection, type, and location of occurrence are analyzed through PRPD (Phase Resolved Partial Discharge), frequency, and ToA (Time of Arrival) analysis for high-resolution signals in the UHF band. Therefore, there are disadvantages in that offline acquisition involves additional costs and real-time analysis is not possible.

1. Korean Patent No. 10-0853725 (registration date: August 18, 2008)

The present invention was proposed to solve the problems caused by the above background technology, and is a partial discharge detection system utilizing sensor data that can simultaneously analyze multiple online data to compensate for the shortcomings of offline analysis and the inherent limitations of single online sensor data. The purpose is to provide methods and methods.

In addition, the present invention analyzes the similarity of PRPD (Phase Resolved Partial Discharge) characteristic factors of multiple sensors to utilize the characteristics of partial discharge (occurring inside the distribution line switchgear) and noise (generated outside the distribution line switchgear) to identify the signal generator. Another purpose is to provide a partial discharge detection system and method using sensor data that can detect partial discharge by determining whether the location is inside or outside the distribution line switchgear.

In order to achieve the tasks presented above, the present invention provides a partial discharge detection system utilizing sensor data that can simultaneously analyze multiple online data to compensate for the shortcomings of offline analysis and the inherent limitations of single online sensor data.

The partial discharge detection system using the sensor data,

A sensor system that generates multiple sensor data by sensing distribution line switches;

a terminal that converts a plurality of sensor data obtained from the sensor system into characteristic data and uses the characteristic data to determine whether partial discharge or noise has occurred in the distribution line switch; and

A network that communicates and connects the sensor system and the terminal.

In addition, the sensor system includes a nearby sensor installed on the inner surface of the distribution line switch;

a target sensor installed toward the distribution line switch to determine occurrence of the partial discharge; and

A noise sensor installed on the outer surface of the distribution line switch and close to the target sensor.

In addition, the plurality of sensor data is two-dimensional PRPS (Phase Resolved Pulse Sequence) data consisting of time and phase axes, and is rearranged according to the current phase to represent the waveform characteristics of the partial discharge.

In addition, the characteristic data is characterized as PRPD (Phase Resolved Partial Discharge) data that is projected and converted in the time direction in order to focus on the phase characteristics of the PRPS data.

In addition, the projection is not a simple overlap of the PRPS data, but is a projection in which the size and number in a specific phase are emphasized by multiplying the number and maximum size of the PRPS data observed for a certain period of time at a specific phase point.

Additionally, the terminal may include an acquisition unit that acquires the sensor data; An extraction module that extracts PRPD (Phase Resolved Partial Discharge)-based characteristic factors using the PRPD data; and an analysis module that generates a similarity result by applying the characteristic factor to an MLP (Multi-layer perceptron) and uses the similarity result to determine whether partial discharge or noise has occurred in the distribution line switch. It is characterized by including.

At this time, the characteristic factors are characterized by skewness, kurtosis, entropy, and FFT (Fast Fourier Transform) ratio.

(여기서, X : PRPD 데이터, μ: 평균, σ: 표준편차, E[] : 기댓값)으로 정의되는 것을 특징으로 한다.In addition, the skewness is expressed by the equation

(Here, X: PRPD data, μ: mean, σ: standard deviation, E[]: expected value).

으로 정의되는 것을 특징으로 한다.In addition, the kurtosis is expressed by the equation

It is characterized by being defined as.

(여기서,

, X는 PRPD 데이터, i는 양의 정수이다)로 정의되는 것을 특징으로 한다.In addition, the entropy is expressed by the equation

(here,

, X is PRPD data, and i is a positive integer).

( 여기서, X_fi는 PRPD의 FFT 결과 i번째 성분이며, i는 양의 정수이다)으로 정의되는 것을 특징으로 한다.In addition, the FFT ratio is expressed by the equation

(Here, X _fi is the ith component of the FFT result of PRPD, and i is a positive integer).

In addition, among the above characteristic factors, the target sensor characteristic factor ( ) and nearby sensor characteristic factors ( The first distance between ) is set to a preset basic distance, and the target sensor characteristic factor ( ) and noise sensor characteristic factors ( ) The second distance between the distances is set to be farther than the basic distance.

(여기서, (f_near)는 인근 센서 특성 변수, (f_target)는 타겟 센서 특성 변수, (f_noise)는 노이즈 센서 특성 변수이며, β는 가중치이다)을 이용하여 상기 부분 방전이 발생했는지 또는 상기 노이즈가 발생했는지를 판단하는 것을 특징으로 한다.In addition, the analysis module uses the equation

(Here, (f _near ) is a nearby sensor characteristic variable, (f _target ) is a target sensor characteristic variable, (f _noise ) is a noise sensor characteristic variable, and β is a weight) to determine whether the partial discharge has occurred or It is characterized by determining whether noise has occurred.

In addition, the analysis module determines that the first distance between the nearby sensor characteristic variable (f _near ) and the target sensor characteristic variable (f _target ) is the distance between the noise sensor characteristic variable (f _noise ) and the target sensor characteristic variable (f _target ). If it is less than the second distance, it is determined as the partial discharge, and the first distance between the nearby sensor characteristic variable (f _near ) and the target sensor characteristic variable (f _target ) is equal to the noise sensor characteristic variable (f _noise ) and the target. If it is greater than the second distance between the sensor characteristic variables (f _target ), it is characterized as noise.

In addition, the analysis module is characterized in that it displays the selection of the single-line diagram of the distribution line switch, the neighborhood sensor, the target sensor, and the noise sensor on the upper screen of the user interface.

In addition, the analysis module is characterized in that it displays similarity information between the neighborhood sensor, the target sensor, and the noise sensor on a lower screen of the user interface.

On the other hand, another embodiment of the present invention includes (a) generating a plurality of sensor data by a sensor system sensing a distribution line switch; (b) converting a plurality of sensor data obtained from the sensor system into characteristic data by a terminal connected to the sensor system through a network; and (c) determining, by the terminal, whether a partial discharge or noise has occurred in the distribution line switch using the characteristic data. It provides a partial discharge detection method using sensor data, comprising: .

According to the present invention, it is possible to diagnose partial discharge that is robust against noise showing phase characteristics similar to partial discharge (PD), which cannot be determined using existing online single sensor data.

In addition, another effect of the present invention is that in addition to detection using similarity, three-dimensional analysis and/or rough estimation of the location of PD occurrence are possible by presenting a similarity matrix that can simultaneously observe the similarity of multiple sensor data. .

Figure 1 is a block diagram of a partial discharge detection system using sensor data according to an embodiment of the present invention.
Figure 2 is a flowchart showing a process for detecting partial discharge using sensor data according to an embodiment of the present invention.
Figure 3 is a flowchart detailing the sensor data acquisition steps according to an embodiment of the present invention.
FIG. 4 is an example of sensor data shown in FIG. 2.
FIG. 5 is a conceptual diagram of acquiring PRPD data through multiple PRPS temporal projections shown in FIG. 2.
Figure 6 is a graph showing the results of Fast Fourier Transform of partial discharge data according to an embodiment of the present invention.
Figure 7 is a graph showing the results of fast Fourier transformation of noise data according to an embodiment of the present invention.
Figure 8 is a flowchart showing a partial discharge determination process using a similarity learning model according to an embodiment of the present invention.
Figure 9 is an example of a user interface configuration using similarity analysis according to an embodiment of the present invention.

The present invention can make various changes and have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all changes, equivalents, and substitutes included in the spirit and technical scope of the present invention.

When describing each drawing, similar reference numerals are used for similar components. Terms such as first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

For example, a first component may be referred to as a second component, and similarly, the second component may be referred to as a first component without departing from the scope of the present invention. The term “and/or” includes any of a plurality of related stated items or a combination of a plurality of related stated items.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by a person of ordinary skill in the technical field to which the present invention pertains.

Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and unless clearly defined in the present application, should not be interpreted in an ideal or excessively formal sense. It shouldn't be.

Hereinafter, a partial discharge detection system and method using sensor data according to an embodiment of the present invention will be described in detail with reference to the attached drawings.

Figure 1 is a block diagram of a partial discharge detection system 100 using sensor data according to an embodiment of the present invention. Referring to FIG. 1, the sensor data utilization partial discharge detection system 100 acquires a plurality of sensor data from the sensor system 110 and the sensor system 110, which generates a plurality of sensor data by sensing a distribution line switch. It may be configured to include a terminal 120 that determines whether it is a partial discharge or noise, a network 10 that communicates and connects the sensor system 110 and the terminal 120, etc.

Distribution line switchgear is divided into hydraulic, magnetic, and pneumatic types, and gas insulated switchgear (GIS), which uses SF6 gas with excellent dielectric strength, is widely used. In one embodiment of the present invention, GIS is explained as an example, but it is not limited to this and other load switches, etc. can also be applied.

The sensor system 110 may be composed of a noise sensor, a nearby sensor, a target sensor, etc.

The noise sensor is installed on the inner surface of the distribution line switch, the target sensor is installed toward the distribution line switch and is installed to determine the occurrence of partial discharge, and the nearby sensor is installed on the outer surface of the distribution line switch and the above It is installed close to the target sensor. The sensor 110 is an external electromagnetic wave sensor with a detection band of 500 ~ 1,500MHz. It has a sensitivity that can detect IEC60270 apparent partial discharge of 5pC or less. To minimize the inflow of external noise, a shielding material is installed as a spacer and a gasket is installed on the sensor installation surface. It is desirable to insert it in the form.

Generally, the frequency detection band is 500 to 1,500 MHz within the UHF (Utra High Frequency) band (300 to 3,000 MHz). Below 500 MHz is a frequency band with strong external noise, and over 1,500 MHz shows rapid attenuation due to the characteristics of high frequencies, so take this into consideration when determining the frequency. band can be detected.

The terminal 120 may be configured to include an acquisition unit 121, an extraction module 122, an analysis module 123, and an output unit 124. The acquisition unit 121 performs a function of connecting to the network 10 through wired or wireless communication. For this purpose, a microprocessor, modem, LAN card, circuit, etc. are configured.

The extraction module 122 collects information such as sensor data acquired through the acquisition unit 121 and stores it in the storage unit 130.

The analysis module 1230 uses sensor data to determine whether partial discharge or noise has occurred in the distribution line switch 901.

The output unit 124 functions to display information processed by the extraction module 122, analysis module 123, etc. Accordingly, the output unit 124 can generate notification information using a combination of text, voice, and graphics. To this end, the output unit 124 may be configured to include a display, a sound system, etc.

The display may be a Liquid Crystal Display (LCD), Light Emitting Diode (LED) display, Plasma Display Panel (PDP), Organic LED (OLED) display, touch screen, Cathode Ray Tube (CRT), or flexible display. In the case of a touch screen, it can function as an input means.

The extraction module 122 and analysis module 123 shown in FIG. 2 refer to units that process at least one function or operation, and may be implemented in software and/or hardware. In hardware implementation, an application specific integrated circuit (ASIC), digital signal processing (DSP), programmable logic device (PLD), field programmable gate array (FPGA), processor, microprocessor, and other devices designed to perform the above-described functions. It may be implemented as an electronic unit or a combination thereof.

In software implementation, software composition components (elements), object-oriented software composition components, class composition components and task composition components, processes, functions, properties, procedures, subroutines, segments of program code, drivers, firmware, microcode, data. , databases, data structures, tables, arrays, and variables. Software, data, etc. can be stored in memory and executed by a processor. The memory or processor may employ various means well known to those skilled in the art.

The storage unit 130 performs the function of storing sensor data, notification information, similarity analysis model, etc. The storage unit 130 may be configured within the terminal 120 or may be configured on a separate server. Accordingly, the storage unit 130 may be a flash memory type, a hard disk type, a multimedia card micro type, or a card type memory (e.g., SD (Secure Digital) or eXtreme Digital (XD) memory, etc.), Random Access Memory (RAM), Static Random Access Memory (SRAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Programmable Read It may include at least one type of storage medium among (Only Memory), magnetic memory, magnetic disk, and optical disk. Additionally, it may operate in connection with web storage and cloud servers that perform storage functions on the Internet.

The network 10 performs a function of communicating and connecting the sensor system 110 and the terminal 120. Communication connections can be RS232, RS485, Modbus, CC-Link communication, Ethernet communication, etc. Of course, the network 10 refers to a connection structure that allows information exchange between each node, such as a plurality of terminals and servers, such as the public switched telephone network (PSTN), public switched data network (PSDN), and comprehensive information and communication network ( ISDN (Integrated Services Digital Networks), Broadband ISDN (BISDN), Local Area Network (LAN), Metropolitan Area Network (MAN), Wide LAN (WLAN), etc. However, the present invention is not limited to this, and is applicable to wireless communication networks such as CDMA (Code Division Multiple Access), WCDMA (Wideband Code Division Multiple Access), Wibro (Wireless Broadband), WiFi (Wireless Fidelity), and HSDPA (High It can be a Speed Downlink Packet Access (Speed Downlink Packet Access) network, Bluetooth, NFC (Near Field Communication) network, satellite broadcasting network, analog broadcasting network, DMB (Digital Multimedia Broadcasting) network, etc. Alternatively, it may be a combination of these wired communication networks and wireless communication networks.

Figure 2 is a flowchart showing a process for detecting partial discharge using sensor data according to an embodiment of the present invention. Referring to Figure 2, the acquisition module 121 acquires sensor data through the sensor system 110 (step S210). The sensor system 110 includes a target sensor, a nearby sensor, a noise sensor, etc. Sensor The data may be PRPS (Phase Resolved Pulse Sequence) data, which is two-dimensional data consisting of time and phase axes, and data acquired in time series are rearranged according to the current phase to represent the waveform characteristics of the partial discharge. This is data.

Thereafter, in order to focus on the phase characteristics of the PRPS data, the extraction module 122 projects it in the time direction and converts it into PRPD (Phase Resolved Partial Discharge) data to obtain it (step S220). In other words, in order to focus on the phase characteristics of PRPS data, it is projected in the time direction and expressed in PRPD form. In the process of projecting in the time direction, rather than simply overlapping the data, the number of data observed during one minute at a specific phase is multiplied by the maximum size, and the size and number of data in a specific phase are projected to be emphasized.

Afterwards, the extraction module 122 extracts PRPD (Phase Resolved Partial Discharge)-based characteristic factors (step S230). In detail, four characteristic factors are extracted using PRPD data. The four characteristic factors include skewness, kurtosis, entropy, and FFT (Fast Fourier Transform) rate.

Skewness is defined as follows:

Here, X: PRPD data, μ: mean, σ: standard deviation, E[]: expected value

Additionally, kurtosis is defined as follows:

Entropy is defined as follows:

P _i is defined as follows:

Here, X is PRPD data and i is a positive integer.

The FFT ratio is defined as follows:

Here, X _fi is the ith component of the FFT result of PRPD, and i is a positive integer.

Continuing to refer to FIG. 2, the analysis module 123 performs multi-layer perceptron (MLP)-based similarity learning using PRPD characteristic factors (step S240). The MLP model constructs two perceptrons in one layer and constructs a new perceptron connecting these two perceptrons. Therefore, it consists of an input layer that inputs input values, an output layer that outputs results, and a hidden layer that is an internal layer. In particular, in the case of the MLP model, there are two hidden layers.

Therefore, the distance between the target sensor characteristic factor (F _target ) and the potential variable of the nearby sensor characteristic factor (F _near ) is close, and the distance between the target sensor characteristic factor and the noise sensor characteristic factor (F _noise ) is learned far.

(여기서, β는 가중치이다)로 선정하고 이를 최소화하는 방향으로 Adam Optimizer 기반으로 학습 진행한다.At this time, the loss function is

(here, β is the weight) is selected and learning is performed based on Adam Optimizer in the direction of minimizing it.

Afterwards, when the similarity result is calculated according to this learning, the analysis module 123 determines whether partial discharge or simple noise has occurred based on the similarity result, and outputs this result as notification information to the output unit 124. . Notification information may be a combination of graphics, text, and voice.

Figure 3 is a flowchart detailing the sensor data acquisition steps according to an embodiment of the present invention. Referring to FIG. 3, the sensor data acquisition step includes the processes of acquiring a nearby sensor PRPS signal, acquiring a target sensor PRPS signal, and acquiring a noise sensor PRPS signal (steps S301, S302, and S303).

Since the remaining steps (S310, S320, and S330) have already been described in FIG. 2, further description will be omitted.

Afterwards, the analysis module 123 determines the similarity result (step S340). As a result of the determination, if the similarity according to the nearby sensor PRPS signal acquisition, target sensor PRPS signal acquisition, and noise sensor PRPS signal in step S340 satisfies a certain condition, partial discharge (PD) is determined (step S350).

In contrast, the analysis module 1230 determines it as noise if the similarity according to the nearby sensor PRPS signal acquisition, target sensor PRPS signal acquisition, and noise sensor PRPS signal does not meet certain conditions in step S340 (step S340). These certain conditions is shown in FIG. 8 and will be described later.

FIG. 4 is an example of sensor data shown in FIG. 2. Referring to FIG. 4, sensor data becomes PRPS data. PRPS data is two-dimensional data composed of time and phase axes, and is data obtained in time series and rearranged according to the current phase to represent the waveform characteristics of the partial discharge. Therefore, in order to compare the inside/outside of the target signal of the switch, PRPS data is acquired from the target sensor aiming at partial discharge determination, a nearby sensor, and a noise sensor located close to the target sensor. The horizontal axis of the PRPS data 400 represents time (unit is seconds), and the vertical axis represents phase.

FIG. 5 is a conceptual diagram of acquiring PRPD data 500 through multiple PRPS temporal projections shown in FIG. 2. Referring to FIG. 5, PRPD data 500 is generated by applying a weighted projection to the PRPS data 400. In order to focus on the phase characteristics of PRPS data, it is projected in the time direction and expressed in PRPD form. In the process of projecting in the time direction, rather than simply overlapping PRPS data, the number of PRPS data observed for 1 minute at a specific phase is multiplied by the highest size, and the size and number of data in a specific phase are projected to be emphasized. In the PRPD data 500 of FIG. 5, the horizontal axis represents phase, and the vertical axis represents entropy.

Figure 6 is a graph showing the results of Fast Fourier Transform of partial discharge data according to an embodiment of the present invention. Referring to FIG. 6, the PRPD data 500 obtained in FIG. 5 is partial discharge data. The horizontal axis represents frequency, and the vertical axis represents size. In the case of partial discharge, it is almost flat after the peak occurs once, as shown in the shaded area on the left.

Figure 7 is a graph showing the results of fast Fourier transformation of noise data according to an embodiment of the present invention. Referring to FIG. 7, this is a case where the PRPD data 500 obtained in FIG. 5 is noise data. In the case of noise, large and small peaks occur several times, as shown in the shaded area on the right.

(여기서, β는 가중치이다)로 선정하고 이를 최소화하는 방향으로 Adam Optimizer 기반으로 이루어진다.Figure 8 is a flowchart showing a partial discharge determination process using a similarity learning model according to an embodiment of the present invention. Referring to FIG. 8, a nearby sensor characteristic factor (F _near ), a target sensor characteristic factor (F _target ), and a noise sensor characteristic factor (F _noise ) are extracted from the PRPD data 500 (steps S801, S802, and S803). Afterwards, the characteristic factors (F _near , F _target , F _noise ) are applied to the MLP model to calculate the characteristic variables (f _near , f _target , f _noise ) through learning (steps S811, S812, S813). The learning process uses the loss function

(here, β is the weight) is selected and minimized based on Adam Optimizer.

As a result of this learning, a nearby sensor characteristic variable (f _near ), a target sensor characteristic variable (f _target ), and a noise sensor characteristic variable (f _noise ) are created.

Afterwards, it is determined whether it is a partial discharge or noise through the distances between the characteristic variables (f _near , f _target , f _noise ) calculated from the learned model (step S820). To elaborate, the distance between characteristic variables is compared, and if it is close to a nearby sensor, it is judged as partial discharge, and if it is close to a noise sensor, it is judged as noise (steps S821 and S830). For this purpose, the judgment conditions are as follows.

In other words, the analysis module 123 determines that the distance between the nearby sensor characteristic variable (f _near ) and the target sensor characteristic variable (f _target ) is smaller than the distance between the noise sensor characteristic variable (f _noise ) and the target sensor characteristic variable (f _target ). It is judged to be a partial discharge, and if the distance between the nearby sensor characteristic variable (f _near ) and the target sensor characteristic variable (f _target ) is greater than the distance between the noise sensor characteristic variable (f _noise ) and the target sensor characteristic variable (f _target ), it is considered noise. judge.

Figure 9 is an example of a user interface configuration using similarity analysis according to an embodiment of the present invention. Referring to FIG. 9, in a user interface configuration example, a single line diagram and selection of sensors 911, 912, and 913 are displayed on the upper screen 900. To elaborate, the upper screen 900 displays a switch 901 connected to the bus 930, and based on this switch 901, a nearby sensor 911, a target sensor 912, and a noise sensor 913 are displayed. It is placed. Additionally, the bus 930 consists of a 65 bus 931 and a 60 bus 932.

Meanwhile, the lower screen 9920 displays a similarity matrix expressed as similarity information between sensors. That is, the columns include the noise sensor 913, the neighborhood sensor 911, and the target sensor 912 in that order, and the rows include the noise sensor 913, the neighborhood sensor 911, and the target sensor 912 in that order.

To elaborate, while the content of Equation 6 directly determines whether it is PD or noise, the similarity matrix shows one of other possible use cases. That is, when two different sensors are selected (e.g., 913, 911), the distance between the information (f_noise, f_near) extracted from each is calculated through the statistical distance (mahalanobis distance), and then the normal distribution function (gaussian distribution probability density function) ) This is the result of calculating the similarity by mapping.

Meanwhile, it is possible to diagnose PD robustly against noise that shows phase characteristics similar to PD, which cannot be determined using existing online single sensor data. To elaborate, the probability of PD false detection due to the limitations of a single sensor is about 40%, the probability of additional PD detection through comparison of existing false detection data and multi-sensor data is 83.75%, and the increase in PD detection probability using multiple sensors is 0.4*0.8375 = approximately 33.5%.

Additionally, the steps of the method or algorithm described in relation to the embodiments disclosed herein are implemented in the form of program instructions that can be executed through various computer means such as a microprocessor, processor, CPU (Central Processing Unit), etc., and are computer readable. Can be recorded on any available medium. The computer-readable medium may include program (instruction) codes, data files, data structures, etc., singly or in combination.

The program (instruction) code recorded on the medium may be specially designed and constructed for the present invention, or may be known and usable by those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROM, DVD, and Blu-ray, and ROM and RAM. Semiconductor memory elements specially configured to store and execute program (instruction) code, such as RAM), flash memory, etc., may be included.

Here, examples of program (instruction) code include not only machine language code such as that created by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

100: Partial discharge detection system
110: sensor system
120: terminal
121: Acquisition department
122: Extraction module
123: Analysis module
124: output unit
130: storage unit
901: Switchgear
911: Nearby sensor
912: Target sensor
913: Noise sensor
911: Target sensor

Claims (16)

A sensor system (110) that generates a plurality of sensor data by sensing the distribution line switch (901);
A terminal ( 120); and
A network 10 that communicates and connects the sensor system 110 and the terminal 120;
A partial discharge detection system utilizing sensor data, comprising:
According to claim 1,
The sensor system 110,
A nearby sensor 911 installed on the inner surface of the distribution line switch 901;
a target sensor 912 installed toward the distribution line switch 901 to determine occurrence of the partial discharge; and
A noise sensor 913 installed on the outer surface of the distribution line switch 901 and close to the target sensor 912; A partial discharge detection system utilizing sensor data, comprising:
According to claim 1,
A plurality of the sensor data is two-dimensional PRPS data consisting of time and phase axes, and is rearranged according to the current phase to indicate the waveform characteristics of the partial discharge. Partial discharge detection system utilizing sensor data.
According to claim 3,
The characteristic data is PRPD (Phase Resolved Partial Discharge) data that is projected and converted in the time direction in order to focus on the phase characteristics of the PRPS data. A partial discharge detection system utilizing sensor data.
According to claim 4,
The projection is not a simple overlap of the PRPS data, but is a projection in which the size and number in a specific phase are emphasized by multiplying the number and maximum size of the PRPS data observed for a certain time at a specific phase point in time. Partial discharge using sensor data Detection system.
According to claim 4,
The terminal 120,
An acquisition unit 121 that acquires the sensor data;
An extraction module 122 that extracts PRPD (Phase Resolved Partial Discharge)-based characteristic factors using the PRPD data; and
An analysis module that generates a similarity result by applying the characteristic factor to an MLP (Multi-layer perceptron) and determines whether partial discharge or noise has occurred in the distribution line switch 901 using the similarity result. (123); A partial discharge detection system utilizing sensor data, wherein the characteristic factors include skewness, kurtosis, entropy, and FFT (Fast Fourier Transform) ratio.
According to claim 6,
The skewness is expressed by the equation

(Where,
According to claim 7,
The kurtosis is expressed by the equation

A partial discharge detection system utilizing sensor data, characterized in that it is defined as:
According to claim 6,
The entropy is expressed by the equation

(here,

, X is PRPD data, and i is a positive integer).
According to claim 6,
The FFT ratio is expressed by the equation

(Where _,
According to claim 6,
Among the above characteristic factors, the target sensor characteristic factor ( ) and nearby sensor characteristic factors ( The first distance between ) is set to a preset basic distance, and the target sensor characteristic factor ( ) and noise sensor characteristic factors ( ) Partial discharge detection system utilizing sensor data, wherein the second distance is set to be farther than the basic distance.
According to claim 6,
The analysis module 123 uses the equation

(Here, (f _near ) is a nearby sensor characteristic variable, (f _target ) is a target sensor characteristic variable, (f _noise ) is a noise sensor characteristic variable, and β is a weight) to determine whether the partial discharge has occurred or A partial discharge detection system utilizing sensor data that determines whether noise has occurred.
According to claim 12,
The analysis module 123 determines that the first distance between the nearby sensor characteristic variable (f _near ) and the target sensor characteristic variable (f _target ) is the noise sensor characteristic variable (f _noise ) and the target sensor characteristic variable (f _target ). If it is less than the second distance between, it is determined as the partial discharge, and the first distance between the nearby sensor characteristic variable (f _near ) and the target sensor characteristic variable (f _target ) is the noise sensor characteristic variable (f _noise ) and the A partial discharge detection system utilizing sensor data, characterized in that it is judged as noise if it is greater than the second distance between target sensor characteristic variables (f _target ).
According to claim 6,
The analysis module 123 selects the single-line diagram of the distribution line switch 901, the neighborhood sensor 911, the target sensor 912, and the noise sensor 913 on the upper screen 900 of the user interface. A partial discharge detection system utilizing sensor data, characterized in that the display.
According to claim 6,
The analysis module 123 utilizes sensor data to display similarity information between the neighborhood sensor 911, the target sensor 912, and the noise sensor 913 on the lower screen 920 of the user interface. Partial discharge detection system.
(a) the sensor system 110 generating a plurality of sensor data by sensing the distribution line switch 901;
(b) converting a plurality of sensor data obtained from the sensor system 110 into characteristic data by a terminal 120 connected to the sensor system 110 through a network 10; and
(c) the terminal 120 determining whether partial discharge or noise has occurred in the distribution line switch 901 using the characteristic data;
Partial discharge detection method using sensor data, comprising: