TWM654952U - Power transmission line fault location system - Google Patents

Abstract

A transmission line fault location system is used to locate a fault position of a transmission line to be tested and includes a processor electrically connected to a storage module and a receiving module. The processor uses a fault location estimation model stored in the storage module and established by deep learning based on a large amount of real-time digital simulation data related to multiple fault types to analyze three-phase parameter data received by the receiving module and associated with three-phase voltages and three-phase currents detected at opposite ends of the transmission line to be tested, so as to obtain fault location data corresponding to an estimated fault position of the transmission line to be tested.

Description

Transmission Line Fault Location System

The invention relates to a transmission line fault, and in particular to a transmission line fault location system.

Transmission lines play a key role in the transmission and distribution of electricity in the power system, and transmission line failure is one of the most common power equipment failures. Once a transmission line failure occurs, if it is not handled in time, it will seriously affect the stability of power supply. This will not only cause inconvenience to the daily lives of ordinary electricity users, but also affect the production operations of industrial users, and even worse, will lead to huge economic losses. Transmission line failures usually include four types, namely single-phase grounding fault (hereinafter referred to as LG), two-phase short circuit fault (hereinafter referred to as LL), two-phase grounding fault (hereinafter referred to as LLG) and three-phase short circuit (hereinafter referred to as LLL).

Existing protective relaying technology has been used to monitor faults in transmission lines in power systems. When a fault occurs in a transmission line, for example, distance protection detection is used to calculate the impedance value between the fault point and the protection point based on the changes in the current and voltage of the transmission line, and compare it with a preset fault location characteristic curve, and estimate the location of the fault based on the comparison result. However, when the fault point and the protection point are close (for example, the distance between the two is less than 5% of the entire transmission line length), due to the smaller impedance difference, the uncertainty of the estimated fault location increases (that is, the estimated fault location has a larger error).

Therefore, how to create a fault location method that can compensate for the problems existing in the existing protection pole technology during the above-mentioned distance protection and combine the protection pole technology to effectively improve the accuracy of fault location has become one of the issues that the relevant technical fields want to solve.

Therefore, the purpose of the present invention is to provide a power transmission line fault location system that can overcome at least one disadvantage of the prior art.

Therefore, a transmission line fault location system provided by the present invention is used to locate the fault position of a transmission line to be tested. The transmission line to be tested has a first end and a second end opposite to the first end. The transmission line fault location system includes a storage module, a receiving module and a processor electrically connected to the storage module and the receiving module.

The storage module stores a fault location estimation model that is pre-established based on a large amount of real-time digital simulation data related to various fault types and through deep learning.

The receiving module is used to receive three-phase parameter data related to the power transmission line to be tested, wherein the three-phase parameter data is associated with the three-phase voltage and three-phase current detected at the first and second ends respectively.

The processor uses the fault location estimation model stored in the storage module to analyze the three-phase parameter data received by the receiving module to obtain fault location data corresponding to an estimated fault location of the power transmission line to be tested.

In some embodiments, the three-phase parameter data includes a first characteristic parameter set regarding the A-phase current and the A-phase voltage at the first and second ends, a second characteristic parameter set regarding the B-phase current and the B-phase voltage at the first and second ends, and a third characteristic parameter set regarding the C-phase current and the C-phase voltage at the first and second ends.

In some embodiments, the processor further normalizes the first characteristic parameter set, the second characteristic parameter set, and the third characteristic parameter set before analyzing the three-phase parameter data.

In some embodiments, the fault location estimation model includes an input layer for receiving the three-phase parameter data, a plurality of hidden layers related to fault type, phase angle, impedance and line location and used to analyze the three-phase parameter data, and an output layer for outputting the fault location data.

In some embodiments, the fault location data output at the output layer of the fault location estimation model indicates the estimated fault location.

The effectiveness of the present invention is that since the fault location estimation model is based on a large amount of real-time digital simulation data related to various fault types and is trained through deep learning, this large amount of real-time digital simulation data can specifically and fully simulate the actual fault conditions of the transmission line. Therefore, the estimated fault location obtained by analyzing the three-phase parameter data of the transmission line to be tested through the fault location estimation model has a relatively high reliability, especially being able to compensate for the problems encountered by the existing protection post technology as mentioned above, and when used in combination with the protection post technology, it can indeed effectively improve the accuracy of transmission line fault location.

Before the present invention is described in detail, it should be noted that similar components are represented by the same reference numerals in the following description.

Referring to FIG. 1 , a transmission line fault location system 1 of the present novel embodiment is used to locate the fault position of a transmission line 3 to be tested. The transmission line 3 to be tested has a first end (e.g., a starting point) and a second end (e.g., an end point) opposite to the first end. It is worth noting that in this embodiment, the transmission line fault location system 1 must be used in conjunction with a protection switch system 2.

More specifically, the two detection modules (not shown) included in the protection powertrain system 2, which are respectively arranged at the first end and the second end of the transmission line 3 to be tested, will continuously detect the three-phase (i.e., A phase, B phase and C phase) current and voltage signals at the first end and the second end. Therefore, the protection powertrain system 2 can obtain the A phase current signal (represented by IA1), the B phase current signal (represented by IB1), the C phase current signal (represented by IC1), the A phase voltage signal (represented by VA1), the B phase voltage signal (represented by VB1) and the C phase voltage signal (represented by VC1) detected at the first end, and the C phase voltage signal (represented by VC1) detected at the second end through the detection modules. The received A-phase current signal (represented by IA2), B-phase current signal (represented by IB2), C-phase current signal (represented by IC2), A-phase voltage signal (represented by VA2), B-phase voltage signal (represented by VB2) and C-phase voltage signal (represented by VC2), where each current signal IA1/IB1/IC1/IA2/IB2/IC2 includes a current amplitude IA1 _Mag /IB1 _Mag /IC1 _Mag /IA2 _Mag /IB2 _Mag /IC2 _Mag and current phase angle IA1 _Ang /IB1 _Ang /IC1 _Ang /IA2 _Ang /IB2 _Ang /IC2 _Ang and each voltage signal VA1/VB1/VC1/VA2/VB2/VC2 includes a voltage amplitude VA1 _Mag /VB1 _Mag /VC1 _Mag /VA2 _Mag /VB2 _Mag /VC2 _Mag and a voltage phase angle VA1 _Ang /VB1 _Ang /VC1 _Ang /VA2 _Ang /VB2 _Ang /VC2 _Ang . Please note that in the present embodiment, the current amplitude IA1 _Mag and the current phase angle IA1 _Ang contained in the A-phase current signal IA1 related to the A-phase current at the first end, the voltage amplitude VA1 Mag and the voltage phase angle VA1 _Ang contained in the A-phase voltage signal VA1 related to the A-phase voltage at the first end, the current amplitude IA2 _Mag and the current phase angle IA2 Ang contained in the A-phase current signal IA2 related to the A-phase current at the second end, and the voltage amplitude VA2 _Mag and the voltage phase angle VA2 _Ang contained in the A-phase voltage signal VA2 related to the A-phase voltage at the second end are used as characteristic parameters contained in a first characteristic parameter group; the current amplitude IB1 _Mag and the current phase angle VA1 _Ang contained in the B-phase current signal IB1 related to the B-phase current at the first end _Mag and current phase angle IB1 _Ang , voltage amplitude VB1 _Mag and voltage phase angle VB1 Ang contained in the B-phase voltage signal VB1 related to the B-phase voltage at the first end, current amplitude IB2 _Mag and current phase angle IB2 _Ang contained in the B-phase current signal IB2 related to the B-phase current at the second end, and voltage amplitude VB2 Mag and voltage phase angle VB2 _Ang contained in the B-phase voltage signal VB2 related to the B-phase voltage at the second end as characteristic parameters contained in a second characteristic parameter set; and current amplitude IC1 _Mag and current phase angle _{IC1 Ang contained} in the C-phase current signal IC1 related to the C-phase current _at the first end , the voltage amplitude VC1 _Mag and the voltage phase angle VC1 _Ang contained in the C-phase voltage signal VC1 related to the C-phase voltage at the first end, the current amplitude IC2 _Mag and the current phase angle IC2 Ang contained in the C-phase current signal IC2 related to the C _- phase current at the second end, and the voltage amplitude VC2 _Mag and the voltage phase angle VC2 _Ang contained in the C-phase voltage signal VC2 related to the C-phase voltage at the second end as characteristic parameters contained in a third characteristic parameter group. In addition, the first, second, and third characteristic parameter groups together constitute three-phase parameter data. Therefore, all characteristic parameters (i.e., 24 characteristic parameters) contained in the three-phase parameter data of the transmission line 3 to be tested obtained by the protection powertrain system 2 can be clearly summarized in the following Table 1. Table 1 Feature parameter set Characteristic parameters Three-phase parameter data The first characteristic parameter set IA1 _Mag ，IA1 _Ang ，VA1 _Mag ，VA1 _Ang ，IA2 _Mag ，IA2 _Ang ，VA2 _Mag ，VA2 _Ang The second characteristic parameter set IB1 _Mag ，IB1 _Ang ，VB1 _Mag ，VB1 _Ang ，IB2 _Mag ，IB2 _Ang ，VB2 _Mag ，VB2 _Ang The third characteristic parameter set IC1 _Mag ，IC1 _Ang ，VC1 _Mag ，VC1 _Ang ，IC2 _Mag ，IC2 _Ang ，VC2 _Mag ，VC2 _Ang

The power transmission line fault location system 1 includes a storage module 11, a receiving module 12, and a processor 13 electrically connected to the storage module 11 and the receiving module 12.

The storage module 11 stores a fault location estimation model that is pre-established based on a large amount of real-time digital simulation data related to various fault types and through deep learning. Referring to FIG. 2 , the fault location estimation model has a type of neural network architecture and includes an input layer having multiple (e.g., 24) output terminals (i.e., X1 to X24) for receiving data, multiple hidden layers, and an output layer having an output terminal (i.e., Y) for outputting data. The number of neurons (nodes) in each hidden layer can be appropriately selected to avoid overfitting and underfitting, and an activation function such as a rectified linear unit (ReLU) function is used. In addition, each hidden layer can also appropriately add a dropout layer to effectively avoid overfitting. The output layer uses an activation function such as a softmax function. In this embodiment, the fault types include, for example, the following 10 fault types: A-phase ground fault (hereinafter referred to as AG), B-phase ground fault (hereinafter referred to as BG) and C-phase ground fault (hereinafter referred to as CG) belonging to the LG type; A&B phase short circuit fault (hereinafter referred to as AB), B&C phase short circuit fault (hereinafter referred to as BC) and A&C phase short circuit fault (hereinafter referred to as AC) belonging to the LL type; A&B phase ground fault (hereinafter referred to as ABG), B&C phase ground fault (hereinafter referred to as BCG) and A&C phase ground fault (hereinafter referred to as ACG) belonging to the LLG type; and A&B&C phase short circuit fault (hereinafter referred to as ABC) belonging to the LLL type. The large amount of real-time digital simulation data is a large amount of three-phase current and voltage data generated by the existing real-time digital simulator (RTDS) according to the simulation conditions of the 10 fault types, multiple different phase angles, multiple different impedances and multiple different (normalized) fault locations. Since RTDS can generate real-time simulation results (i.e., real-time digital simulation data) at a relatively fast speed compared to other simulation software, it is conducive to generating a large amount of real-time digital simulation data that can specifically and completely simulate the actual fault conditions of the transmission line.

It is worth noting that the large amount of real-time digital simulation data may include a training data set used in the model training phase and a test data set used in the model testing phase. For example, the RTDS generates the training data set and the test data set by simulation according to the simulation conditions shown in Table 2 below. Table 2 Simulation of the training dataset Simulation of the test dataset Fault Type AG, BG, CG, AB, BC, AC, ABG, BCG, ACG, LLL (10 types in total) AG, BG, CG, AB, BC, AC, ABG, BCG, ACG, LLL (10 types in total) Phase angle (degrees) 0, 30, 60, 90, 120, 150, 180 (7 types in total) 25, 50, 75, 100, 125 (5 types in total) impedance( ) 0.01, 10, 20, 30, 40, 50 (6 types in total) 5, 15, 25, 35, 45 (5 types in total) Fault location(%) 0, 2, 4, ..., 96, 98, 100 (51 types in total) 5, 10, 15, ..., 90, 95 (19 types in total) If RTDS generates simulation results corresponding to 0.5 seconds according to each simulation situation in Table 1 above, and samples the simulation results every 1.6 milliseconds, 310 pieces of real-time digital simulation data corresponding to 310 samples will be obtained. Therefore, the training data set finally obtained will contain, for example, 6,640,200 ( ) three-phase current and voltage data and the final test data set will contain, for example, 1,472,500 ( ) three-phase current and voltage data, wherein each three-phase current and voltage data contains 24 characteristic values corresponding to the 24 characteristic parameters in Table 1.

The following will further explain in detail how the fault location estimation model is established. First, the training data set must undergo further data processing. Specifically, for example, the above 6,640,200 three-phase current and voltage data (i.e., 24 feature parameters) are processed to remove the same data to obtain, for example, 5,438,902 three-phase current and voltage data. In addition, since the 24 feature values of the 24 feature parameters contained in each three-phase current and voltage data have different ranges, in order to make the range of each update smoother during the training process, the 24 feature values are normalized using, for example, the StandarScaler package so that their respective ranges are limited to between 0 and 1. Then, the multiple characteristic parameters contained in the normalized training data set are fed into the input layer one by one for iterative calculation. In each training iteration, this type of neural network architecture calculates the gradient according to a loss function such as the mean absolute error (MAE), and then uses an optimization algorithm such as Adam (Adaptive Moment Estimation) and an initial learning rate such as 0.001 to update the parameters of the hidden layer (i.e., the weights and biases of the connected neurons). The Adam optimization algorithm can also adjust the learning rate according to the characteristics of the gradient during the training process in order to balance the update speed of different parameters during the training process. Therefore, after a large number of training iterations, the fault location estimation model is trained and established. On the other hand, in the testing phase, after the test data set is similarly processed as described above, the multiple characteristic parameters contained in the normalized test data set are fed into the trained fault location estimation model, and the estimation accuracy of the fault location estimation model is obtained after model analysis.

The receiving module 12 is adapted to be connected to the protection powertrain system 2 to receive the three-phase parameter data obtained by the protection powertrain system 2 as shown in the above Table 1, and transmit the three-phase parameter data to the processor 13.

1 and 3 are referred to below to exemplarily illustrate how the processor 13 executes a transmission line fault location procedure.

First, in step S31, the processor 13 obtains the three-phase parameter data by receiving the three-phase parameter data from the receiving module 12.

Next, in step S32, the processor 13 performs normalization processing on the 24 characteristic values of the 24 characteristic parameters (see the above Table 1) contained in the three-phase parameter data as described above during modeling, so that the range of the characteristic value of each characteristic parameter is between 0 and 1.

Finally, in step S33, the processor 13 uses the fault location estimation model stored in the storage module 11 to analyze the normalized three-phase parameter data (i.e., the normalized 24 eigenvalues) to obtain fault location data corresponding to an estimated fault location of the transmission line 3 to be tested. More specifically, the processor 13 feeds the normalized 24 eigenvalues to the 24 input terminals X1-X24 (see FIG. 2) of the input layer of the fault location estimation model, and after calculation of the hidden layer, outputs the fault location data indicating the estimated fault location at the output terminal Y (see FIG. 2) of the output layer. In this embodiment, the fault location data is, for example, the length from the starting point of the transmission line 3 to the estimated fault position (i.e., the estimated fault point) to the total length of the transmission line 3 to be tested. At this point, the transmission line fault location program is completed. The transmission line fault location system 1 can further output the fault location data for subsequent related processing.

In summary, since the fault location estimation model is based on a large amount of real-time digital simulation data related to various fault types and is trained through deep learning, this large amount of real-time digital simulation data can specifically and fully simulate the actual fault conditions of the transmission line. Therefore, the estimated fault location obtained by analyzing the three-phase parameter data of the transmission line 3 to be tested through the fault location estimation model has a relatively high reliability, especially being able to compensate for the problems encountered by the existing protection post technology as mentioned above, and when used in combination with the protection post technology, it can effectively improve the accuracy of transmission line fault location. Therefore, the novel transmission line fault location system 1 can indeed achieve the purpose of the novel system.

However, the above is only an example of the implementation of the present invention, and it cannot be used to limit the scope of the implementation of the present invention. All simple equivalent changes and modifications made according to the scope of the patent application of the present invention and the content of the patent specification are still within the scope of the present patent.

1: Transmission line fault location system 11: Storage module 12: Receiving module 13: Processor 2: Protection power transmission system 3: Transmission line to be tested S31~S33: Steps

Other features and functions of the present invention will be clearly presented in the implementation method with reference to the drawings, wherein: FIG. 1 is a block diagram, exemplarily illustrating a transmission line fault location system of an embodiment of the present invention, and a protective electric post system used in conjunction therewith; FIG. 2 is a schematic diagram, exemplarily illustrating the structure of a fault location estimation model stored in a storage module of the embodiment; and FIG. 3 is a flow chart, exemplarily illustrating how a processor of the embodiment executes a transmission line fault location program.

1: Transmission line fault location system

11: Storage module

12: Receiving module

13: Processor

2: Protect the electric shock system

3: Transmission line to be tested

Claims (5)

A transmission line fault location system is used to locate the fault position of a transmission line to be tested, wherein the transmission line to be tested has a first end and a second end opposite to the first end. The transmission line fault location system comprises: a storage module storing a fault location estimation model established by deep learning based on a large amount of real-time digital simulation data related to various fault types in advance; a receiving module for receiving three-phase parameter data related to the transmission line to be tested, wherein the three-phase parameter data is associated with the three-phase voltage and three-phase current detected at the first and second ends respectively; and A processor is electrically connected to the storage module and the receiving module, and uses the fault location estimation model stored in the storage module to analyze the three-phase parameter data received by the receiving module to obtain fault location data corresponding to an estimated fault location of the transmission line to be tested. A transmission line fault location system as described in claim 1, wherein the three-phase parameter data includes a first characteristic parameter set regarding the A-phase current and the A-phase voltage at the first and second ends, a second characteristic parameter set regarding the B-phase current and the B-phase voltage at the first and second ends, and a third characteristic parameter set regarding the C-phase current and the C-phase voltage at the first and second ends. A power transmission line fault location system as described in claim 2, wherein the processor normalizes the first characteristic parameter group, the second characteristic parameter group and the third characteristic parameter group before analyzing the three-phase parameter data. A transmission line fault location system as described in claim 1, wherein the fault location estimation model includes an input layer for receiving the three-phase parameter data, a plurality of hidden layers related to the fault type, phase angle, impedance and line location and used for analyzing the three-phase parameter data, and an output layer for outputting the fault location data. A power transmission line fault location system as described in claim 4, wherein the fault location data outputted at the output layer of the fault location estimation model indicates the estimated fault location.

Images