Waveform classification optimization method and system for early warning of potential hazards in power secondary circuits

Abstract

This invention discloses a waveform classification optimization method and system for early warning of potential hazards in power secondary circuits. The method includes: acquiring three-phase current recording file window data; classifying and labeling the three-phase current recording file window data based on a preset numerical judgment principle to obtain labeled target three-phase current recording file window data; converting the target three-phase current recording file window data into two-dimensional images using Gram angle and field GASF, and combining the various two-dimensional images into at least one image unit; inputting at least one image unit into a preset VGG16 convolutional neural network for iterative training to obtain a fault diagnosis model for the neutral line of the current secondary circuit; inputting real-time fault recording time-series data into the fault diagnosis model for the neutral line of the current secondary circuit, and outputting the waveform classification result at the neutral line of the three-phase current circuit. This solves the problem that it is currently difficult to achieve efficient and accurate online detection and judgment of loose connections or breaks in the neutral line of the current secondary circuit.

Description

Technical Field

[0001] This invention belongs to the field of sampling circuit anomaly diagnosis technology, and in particular relates to a waveform classification optimization method and system for early warning of potential hazards in power secondary circuits. Background Technology

[0002] Relay protection devices should sensitively and reliably respond to faults and abnormal operating conditions of various electrical equipment in the power system, and selectively trip relevant circuit breakers quickly, thus playing a crucial role as guardians of power grid safety. When a fault occurs in the sampling circuit of a protection device, it often easily leads to power grid or equipment accidents. Existing secondary circuit anomaly diagnosis methods and systems can currently detect and alert to most types of secondary circuit operational hazards, but they heavily rely on relatively complete fault recording datasets and still have low identification rates for some circuit hazards, such as accurately determining whether the neutral wire in a current circuit is loose or broken.

[0003] As is well known, the health status of the neutral line in the current secondary circuit cannot be reflected under three-phase balance conditions. Even under slight external disturbances to a large power grid, the imbalance characteristics of the three-phase current are difficult to distinguish from normal grid disturbances. Considering the instability and time-varying nature of the neutral line transition resistance, as well as the imbalance caused by different secondary circuit loads, it is usually difficult to establish an accurate and reasonable simulation model for scenarios of loose or open neutral lines in the current circuit and to summarize and refine the mechanism formulas covering all scenarios. Currently, the commonly used approach of refining formulas based on experience has an accuracy rate of less than 50% in the actual production and operation of a certain provincial power grid company. Loose or open neutral lines in the current circuit occur frequently in actual production and can easily cause incorrect operation of related protection systems, posing incalculable risks to the high-quality operation of the new power system. Therefore, it is urgent to conduct intensive research and solutions to address this type of secondary circuit hazard. Summary of the Invention

[0004] This invention provides a waveform classification optimization method and system for early warning of potential hazards in power secondary circuits. It addresses the technical problem that it is currently difficult to achieve efficient and accurate online detection and judgment of loose or broken neutral lines in current secondary circuits, which leads to a decrease in the accuracy of current sampling circuit abnormality diagnosis and the existence of diagnostic blind spots.

[0005] In a first aspect, the present invention provides a waveform classification optimization method for early warning of potential hazards in power secondary circuits, comprising:

[0006] A large power grid simulation model is constructed based on RTDS, which includes a single-phase high-resistivity grounding fault simulation model and a neutral line fault simulation model.

[0007] Erratic simulations were performed on the single-phase high-resistance grounding fault simulation model and the neutral line fault simulation model to obtain the three-phase current waveform file window data under the condition of single-phase fault and superimposed neutral line fault of the transmission line.

[0008] Based on a preset numerical judgment principle, the three-phase current waveform file window data is classified and labeled to obtain the labeled target three-phase current waveform file window data.

[0009] The target three-phase current waveform file window data is converted into a two-dimensional image using Gram angle and field GASF, and the individual two-dimensional images are combined into at least one image unit.

[0010] The at least one image unit is input into a preset VGG16 convolutional neural network for iterative training to obtain a fault diagnosis model for the neutral line of the current secondary circuit.

[0011] Based on the fault recording master station platform, real-time fault recording time sequence data during a single-phase fault in the station-end transmission line is acquired, and the real-time fault recording time sequence data is input into the neutral line fault diagnosis model of the current secondary circuit. The neutral line fault diagnosis model of the current secondary circuit outputs the waveform classification results at the neutral line of the three-phase current circuit.

[0012] Secondly, the present invention provides a waveform classification and optimization system for early warning of potential hazards in power secondary circuits, comprising:

[0013] The construction module is configured to build a large power grid simulation model based on RTDS. The large power grid simulation model includes a single-phase high-resistivity grounding fault simulation model and a neutral line fault simulation model.

[0014] The simulation module is configured to perform ergonomic simulations on the single-phase high-resistance ground fault simulation model and the neutral line fault simulation model, and to obtain the three-phase current waveform file window data under the condition of single-phase fault and superimposed neutral line fault of the transmission line.

[0015] The annotation module is configured to classify and annotate the three-phase current waveform file window data based on a preset numerical judgment principle, so as to obtain the annotated target three-phase current waveform file window data.

[0016] The conversion module is configured to convert the target three-phase current waveform file window data into a two-dimensional image using Gram angle and field GASF, and combine the two-dimensional images into at least one image unit.

[0017] The training module is configured to input the at least one image unit into a preset VGG16 convolutional neural network for iterative training to obtain a fault diagnosis model for the neutral line of the current secondary circuit.

[0018] The output module is configured to acquire real-time fault recording time-series data during a single-phase fault in the transmission line at the station end based on the fault recording master station platform, and input the real-time fault recording time-series data into the neutral line fault diagnosis model of the current secondary circuit. The neutral line fault diagnosis model of the current secondary circuit outputs the waveform classification results at the neutral line of the three-phase current circuit.

[0019] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the waveform classification optimization method for early warning of potential hazards in power secondary circuits according to any embodiment of the present invention.

[0020] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the waveform classification optimization method for early warning of potential hazards in power secondary circuits according to any embodiment of the present invention.

[0021] This application presents a waveform classification optimization method and system for early warning of potential hazards in power secondary circuits. It is specifically designed to identify cases of loose or broken neutral lines in current circuits. Primarily, it is used by power secondary circuit hazard early warning devices to quickly classify cases where there are primary equipment faults in the power system itself, while simultaneously handling the extremely special anomaly of loose or broken neutral lines in the secondary current circuit. This fills the gap in rapid adaptive identification of anomalies in large-scale power secondary circuit data collection. It overcomes the limitations of traditional monitoring methods, such as reliance on mathematical mechanism derivation, difficulty in comprehensively testing and verifying new detection principles, and the need for source comparison. It strengthens and supplements the adaptive diagnosis of secondary circuit anomalies, enabling the anomaly identification method to better meet field needs and possess universality. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 A flowchart of a waveform classification optimization method for early warning of potential hazards in power secondary circuits, provided in an embodiment of the present invention;

[0024] Figure 2This is a structural block diagram of a waveform classification and optimization system for early warning of potential hazards in power secondary circuits, provided in an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Please see Figure 1 The diagram illustrates a flowchart of a waveform classification optimization method for early warning of potential hazards in power secondary circuits, as described in this application. This waveform classification optimization method for early warning of potential hazards in power secondary circuits is specifically designed for monitoring and early warning of loose connections and open circuits in the neutral line of three-phase current circuits. It can be used in power plants (hydropower, thermal power, and new energy) and substations with 110kV and above voltage levels and directly grounded systems containing various types and models of relay protection devices.

[0028] like Figure 1 As shown, the waveform classification optimization method for early warning of potential hazards in power secondary circuits specifically includes the following steps:

[0029] Step S101: Construct a large power grid simulation model based on RTDS, wherein the large power grid simulation model includes a single-phase high-resistivity grounding fault simulation model and a neutral line fault simulation model.

[0030] In this step, a large power grid simulation model is built using RTDS. A direct grounding method is employed to simulate single-phase metallic grounding faults and single-phase high-resistance grounding faults in the transmission line power grid. The single-phase high-resistance grounding fault simulation model is set with a simulation step size of 0.1 ohms for increasing grounding resistance, and the grounding transition resistance ranges from 0.1 to 300 ohms. A neutral line fault simulation model is also built, simulating a three-phase current loop neutral line fault through a high-voltage transformer to the relay protection device. The neutral line fault simulation model is set with a simulation step size of 0.1 ohms for increasing neutral line transition resistance, and the resistance ranges from 0.1 to 1000 ohms.

[0031] It should be noted that, for the RTDS fault simulation process, noise cases with ≤3 consecutive lost points and a total loss rate ≤10% are actively added to ensure that the normal waveform data accounts for 90%, the waveform data with consecutive lost points accounts for 5%, and the waveform data with total loss rate noise accounts for 5%. Specifically, the distribution of consecutive lost point samples should ensure that they appear in the first 30% of the fault waveform data, the middle 40% of the time period, and the last 30% of the time period, with the same distribution ratio. Samples with a total loss rate ≤10% are set according to random occurrence time periods.

[0032] Step S102: Perform ergonomic simulation on the single-phase high-resistance grounding fault simulation model and the neutral line fault simulation model to obtain the three-phase current waveform file window data under the condition of single-phase fault and superimposed neutral line fault of the transmission line.

[0033] In this step, a comprehensive simulation is performed on the single-phase high-resistivity grounding fault simulation model and the neutral line fault simulation model to simulate the three-phase current waveform file window data (including 5ms before the fault start time and 5ms after the fault end time) under the condition of single-phase fault of transmission line superimposed neutral line fault.

[0034] Step S103: Based on the preset numerical judgment principle, classify and label the three-phase current waveform file window data to obtain the labeled target three-phase current waveform file window data.

[0035] In this step, the RTDS simulation results data are extracted and converted into Comrade format waveform files. At the same time, the neutral line fault type is labeled for all the above waveform files (divided into: neutral line transition resistance range of 0.1 ohms-0.5 ohms is normal, resistance range of 0.6 ohms-10 ohms is high resistance, resistance range of 11 ohms-100 ohms is excessive resistance, and resistance range of 101-1000 ohms is complete disconnection).

[0036] For the current loops of line protection, busbar protection branches, and circuit breaker protection devices, the three-phase current waveform file window data is classified and labeled according to the first numerical judgment principle to obtain the labeled target three-phase current waveform file window data.

[0037] Specifically, it is determined whether the data in a certain three-phase current waveform file window meets a first preset condition, wherein the first preset condition is: under the condition of a single-phase short-circuit fault, there exists a non-faulty phase current that is less than 10% of the faulty phase current and in the opposite direction, corresponding to a second harmonic content of less than 10% for the non-faulty phase, and the formula for calculating the second harmonic content of the non-faulty phase is: the effective value of the second harmonic of the non-faulty phase current divided by the effective value of the fundamental wave of the non-faulty phase current; if the first preset condition is met, the data in the certain three-phase current waveform file window is marked as normal; it is then determined whether the data in the certain three-phase current waveform file window meets a second preset condition, wherein the second preset condition is: under the condition of a single-phase short-circuit fault, there exists a non-faulty phase current that is greater than 10% of the faulty phase current and in the opposite direction, corresponding to a second harmonic content of greater than 15% for the non-faulty phase and a duration of less than 20ms; if the second preset condition is met, the data in the certain three-phase current waveform file window is marked as normal. The waveform recording file window data is labeled as "high resistance". It is then determined whether the waveform recording file window data of a certain three-phase current meets a third preset condition. The third preset condition is: under the condition of a single-phase short circuit fault, there exists a non-faulty phase current that is 15% greater than the faulty phase current and in the opposite direction, corresponding to a second harmonic content of the non-faulty phase greater than 15%, and a duration greater than 20ms. If the third preset condition is met, the waveform recording file window data of that certain three-phase current is labeled as "excessive resistance". Next, it is determined whether the waveform recording file window data meets a fourth preset condition. The fourth preset condition is: under the condition of a single-phase short circuit fault, there exists a non-faulty phase current that is 15% greater than the faulty phase current and in the opposite direction, corresponding to a second harmonic content of the non-faulty phase greater than 30%, and a duration greater than 20ms. If the fourth preset condition is met, the waveform recording file window data of that certain three-phase current is labeled as "completely disconnected".

[0038] For main transformer protection, when a single-phase ground fault occurs on the high-voltage side of the main transformer grid, and the neutral line is loosely connected or disconnected in the medium-voltage side current secondary circuit of the main transformer protection device, the three-phase current waveform file window data is classified and labeled according to the second numerical judgment principle to obtain the labeled target three-phase current waveform file window data.

[0039] Specifically, if the main transformer uses YNynd11, the following judgment process is executed: when in the secondary circuit current recording of the voltage side of the main transformer in a certain three-phase current recording file window, there exists a non-fault phase current that is less than the fault phase current. Furthermore, in the opposite direction, when the third harmonic content of the non-faulty phase is less than 10%, the data in a certain three-phase current recording file window is marked as normal; when the secondary circuit current recording on the medium voltage side of the main transformer in a certain three-phase current recording file window shows that the current in the non-faulty phase is greater than the current in the faulty phase. Furthermore, if the direction is opposite, and the third harmonic content of the non-faulty phase is greater than 15% and the duration is less than 20ms, then the data in a certain three-phase current recording file window will be marked as a type with higher resistance; when the secondary circuit current recording of the main transformer medium voltage side of a certain three-phase current recording file window shows that the current in the non-faulty phase is greater than the current in the faulty phase. Furthermore, if the direction is opposite, and the third harmonic content of the non-faulty phase is greater than 15% and the duration is greater than 20ms, then the data in a certain three-phase current recording file window will be marked as excessive resistance; when the secondary circuit current recording of the main transformer medium voltage side of a certain three-phase current recording file window shows that the current in the non-faulty phase is greater than the current in the faulty phase, then... Furthermore, if the direction is opposite, and the third harmonic content of the non-faulty phase is greater than 30% and the duration is greater than 20ms, then the data in a certain three-phase current recording file window will be marked as a complete disconnection type.

[0040] If the main transformer uses YNyd11, the following judgment process is executed: When the secondary circuit current recording on the medium voltage side of the main transformer in a certain three-phase current recording window contains a non-fault phase current that is less than 30% of the fault phase current and leads the phase by 160°, and the corresponding fifth harmonic content of the non-fault phase is less than 10%, then the data in that three-phase current recording window is marked as normal. When the secondary circuit current recording on the medium voltage side of the main transformer in a certain three-phase current recording window contains a non-fault phase current that is greater than 30% of the fault phase current and leads the phase by 160°, and the corresponding fifth harmonic content of the non-fault phase is greater than 15%, and the duration is less than 20ms, then the data in that three-phase current recording window is marked as resistance. Larger type; When the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording window contains a non-fault phase current that is 30% greater than the fault phase current and has a phase lead of 160°, the corresponding fifth harmonic content of the non-fault phase is greater than 15%, and the duration is greater than 20ms, then the data in that three-phase current recording window will be marked as excessive resistance type; When the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording window contains a non-fault phase current that is 30% greater than the fault phase current and has a phase lead of 160°, the corresponding fifth harmonic content of the non-fault phase is greater than 30%, and the duration is greater than 20ms, then the data in that three-phase current recording window will be marked as completely disconnected type.

[0041] Step S104: Convert the target three-phase current waveform file window data into a two-dimensional image using Gram angle and field GASF, and combine the two-dimensional images into at least one image unit.

[0042] In this step, the target three-phase current waveform file window data is normalized according to a preset data processing rule to obtain normalized target three-phase current waveform file window data. The data processing rule is: (current sampling point of the window - average sampling point of the window) / standard deviation of the window sampling point or (current sampling point of the window - minimum sampling point of the window) / (maximum sampling point of the window - minimum sampling point of the window). The normalized target three-phase current waveform file window data is converted into a two-dimensional image using Gram angle and field GASF, and the various two-dimensional images are combined into at least one image unit.

[0043] It should be noted that when acquiring image units of all current loops with normal / abnormal (high resistance, excessive resistance, complete disconnection) neutral lines to be trained, verified, and tested, the setting of the GASF built-in parameter `image_size` should be dynamically adjusted according to the fault recording sampling rate to maintain consistent temporal resolution. Specifically:

[0044] The actual fault recording data of the large power grid comes from different fault recording devices and protection devices. The sampling rate of the device recording will be different. The data with a low sampling rate is smoother and blurrier, especially in the processing of high frequency components, which leads to different image granularity. The model may regard the image texture changes caused by the difference in sampling rate as classification characteristics and the real fault mode, which will ultimately reduce the performance of the model.

[0045] image_size = (fault recording sampling rate) The time window is calculated as the reference sampling rate. The fault recording sampling rate is generally divided into three cases: 1200 Hz, 4000 Hz, and 10000 Hz. The time window is automatically calculated based on the duration of each power grid fault, and is generally in the range of 20-30 (+10 ms). The reference sampling rate is 4000 Hz.

[0046] Step S105: Input the at least one image unit into a preset VGG16 convolutional neural network for iterative training to obtain a fault diagnosis model for the neutral line of the current secondary circuit.

[0047] In this step, a pre-trained VGG16 convolutional neural network is used for fault mode training, and data augmentation is used for feature extraction (selecting horizontal flipping to be applied to 50% of the randomly selected images; selecting to enlarge or reduce the image, with the scaling ratio randomly selected within the range of [−10%, +10%]). Finally, a fault diagnosis model for the neutral line of the current secondary circuit is obtained through training.

[0048] Step S106: Based on the fault recording master station platform, real-time fault recording time sequence data during a single-phase fault in the station-end transmission line is acquired, and the real-time fault recording time sequence data is input into the current secondary circuit neutral line fault diagnosis model. The current secondary circuit neutral line fault diagnosis model outputs the waveform classification results at the neutral line of the three-phase current circuit.

[0049] In summary, the method of this application requires that the points of occurrence of single-phase grounding faults and the points of occurrence of three-phase current loop neutral line loose connection and open circuit faults should not exceed two substations of the same voltage level in terms of electrical physical distance. It can specifically identify the situation of loose connection and open circuit of the neutral line in the current loop. It is mainly used for the early warning device of hidden dangers in the power secondary circuit when the power system itself has primary equipment faults, and at the same time, it is superimposed on the extremely special abnormal situation of loose connection or open circuit of the neutral line in the current secondary circuit. It fills the gap in rapid adaptive identification of abnormal situations in large-scale power secondary circuit data collection. It breaks through the limitations of traditional monitoring methods such as relying on mathematical mechanism derivation, difficulty in comprehensive testing and verification of new detection principles, and comparison with the same source. It strengthens and supplements the adaptive diagnosis of secondary circuit anomalies, so that the anomaly identification method can better meet the field needs and has universality.

[0050] Please see Figure 2 The diagram shows a structural block diagram of a waveform classification optimization system for early warning of potential hazards in power secondary circuits according to this application.

[0051] like Figure 2 As shown, the waveform classification optimization system 200 includes a construction module 210, a simulation module 220, a labeling module 230, a conversion module 240, a training module 250, and an output module 260.

[0052] The system includes: a construction module 210 configured to construct a large power grid simulation model based on RTDS, wherein the large power grid simulation model includes a single-phase high-resistivity grounding fault simulation model and a neutral line fault simulation model; a simulation module 220 configured to perform ergonomic simulation on the single-phase high-resistivity grounding fault simulation model and the neutral line fault simulation model to obtain three-phase current waveform file window data under single-phase fault conditions superimposed with neutral line fault conditions; an annotation module 230 configured to classify and annotate the three-phase current waveform file window data based on a preset numerical judgment principle to obtain annotated target three-phase current waveform file window data; and a conversion module 240 configured to convert the target three-phase current... The waveform recording file window data is converted into a two-dimensional image using Gram angle and field GASF, and the various two-dimensional images are combined into at least one image unit; the training module 250 is configured to input the at least one image unit into a preset VGG16 convolutional neural network for iterative training to obtain a current secondary loop neutral line fault diagnosis model; the output module 260 is configured to acquire real-time fault waveform recording time-series data during a single-phase fault in the transmission line at the station end based on the fault waveform recording master station platform, and input the real-time fault waveform recording time-series data into the current secondary loop neutral line fault diagnosis model, and the current secondary loop neutral line fault diagnosis model outputs waveform classification results at the neutral line of the three-phase current loop.

[0053] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0054] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the waveform classification optimization method for early warning of potential hazards in power secondary circuits in any of the above method embodiments.

[0055] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0056] A large power grid simulation model is constructed based on RTDS, which includes a single-phase high-resistivity grounding fault simulation model and a neutral line fault simulation model.

[0057] Erratic simulations were performed on the single-phase high-resistance grounding fault simulation model and the neutral line fault simulation model to obtain the three-phase current waveform file window data under the condition of single-phase fault and superimposed neutral line fault of the transmission line.

[0058] Based on a preset numerical judgment principle, the three-phase current waveform file window data is classified and labeled to obtain the labeled target three-phase current waveform file window data.

[0059] The target three-phase current waveform file window data is converted into a two-dimensional image using Gram angle and field GASF, and the individual two-dimensional images are combined into at least one image unit.

[0060] The at least one image unit is input into a preset VGG16 convolutional neural network for iterative training to obtain a fault diagnosis model for the neutral line of the current secondary circuit.

[0061] Based on the fault recording master station platform, real-time fault recording time sequence data during a single-phase fault in the station-end transmission line is acquired, and the real-time fault recording time sequence data is input into the neutral line fault diagnosis model of the current secondary circuit. The neutral line fault diagnosis model of the current secondary circuit outputs the waveform classification results at the neutral line of the three-phase current circuit.

[0062] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of a waveform classification and optimization system for early warning of potential hazards in power secondary circuits, etc. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely configured relative to a processor, which can be connected via a network to the waveform classification and optimization system for early warning of potential hazards in power secondary circuits. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0063] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the waveform classification optimization method for power secondary circuit hazard early warning as described in the above method embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the waveform classification optimization system for power secondary circuit hazard early warning. The output device 340 may include a display screen or other display device.

[0064] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0065] In one implementation, the aforementioned electronic device is applied to a waveform classification and optimization system for early warning of potential hazards in power secondary circuits. As a client, it includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to:

[0066] A large power grid simulation model is constructed based on RTDS, which includes a single-phase high-resistivity grounding fault simulation model and a neutral line fault simulation model.

[0067] Erratic simulations were performed on the single-phase high-resistance grounding fault simulation model and the neutral line fault simulation model to obtain the three-phase current waveform file window data under the condition of single-phase fault and superimposed neutral line fault of the transmission line.

[0068] Based on a preset numerical judgment principle, the three-phase current waveform file window data is classified and labeled to obtain the labeled target three-phase current waveform file window data.

[0069] The target three-phase current waveform file window data is converted into a two-dimensional image using Gram angle and field GASF, and the individual two-dimensional images are combined into at least one image unit.

[0070] The at least one image unit is input into a preset VGG16 convolutional neural network for iterative training to obtain a fault diagnosis model for the neutral line of the current secondary circuit.

[0071] Based on the fault recording master station platform, real-time fault recording time sequence data during a single-phase fault in the station-end transmission line is acquired, and the real-time fault recording time sequence data is input into the neutral line fault diagnosis model of the current secondary circuit. The neutral line fault diagnosis model of the current secondary circuit outputs the waveform classification results at the neutral line of the three-phase current circuit.

[0072] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A waveform classification optimization method for early warning of potential hazards in power secondary circuits, characterized in that, include: A large power grid simulation model is constructed based on RTDS, which includes a single-phase high-resistivity grounding fault simulation model and a neutral line fault simulation model. Erratic simulations were performed on the single-phase high-resistance grounding fault simulation model and the neutral line fault simulation model to obtain the three-phase current waveform file window data under the condition of single-phase fault and superimposed neutral line fault of the transmission line. The three-phase current recording file window data is classified and labeled based on a preset numerical judgment principle to obtain labeled target three-phase current recording file window data. The classification and labeling of the three-phase current recording file window data based on the preset numerical judgment principle to obtain labeled target three-phase current recording file window data includes: For the current loops of line protection, bus protection branches, and circuit breaker protection devices, the three-phase current waveform file window data is classified and labeled according to the first numerical judgment principle to obtain the labeled target three-phase current waveform file window data. For the protection of the main transformer, when the single-phase ground fault occurs on the high-voltage side of the main transformer grid, and the neutral line is loosely connected or disconnected in the medium-voltage side current secondary circuit of the main transformer protection device, the three-phase current waveform file window data is classified and labeled according to the second numerical judgment principle to obtain the labeled target three-phase current waveform file window data. The target three-phase current waveform file window data is converted into a two-dimensional image using Gram angle and field GASF, and the various two-dimensional images are combined into at least one image unit. The process of converting the target three-phase current waveform file window data into a two-dimensional image using Gram angle and field GASF, and combining the various two-dimensional images into at least one image unit includes: The target three-phase current waveform file window data is normalized according to the preset data processing rules to obtain the normalized target three-phase current waveform file window data. The data processing rules are: (current sampling point of the window - average sampling point of the window) / standard deviation of the window sampling point or (current sampling point of the window - minimum sampling point of the window) / (maximum sampling point of the window - minimum sampling point of the window). The normalized target three-phase current waveform file window data is converted into a two-dimensional image using Gram angle and field GASF, and the individual two-dimensional images are combined into at least one image unit. The at least one image unit is input into a preset VGG16 convolutional neural network for iterative training to obtain a fault diagnosis model for the neutral line of the current secondary circuit. Based on the fault recording master station platform, real-time fault recording time sequence data during a single-phase fault in the station-end transmission line is acquired, and the real-time fault recording time sequence data is input into the neutral line fault diagnosis model of the current secondary circuit. The neutral line fault diagnosis model of the current secondary circuit outputs the waveform classification results at the neutral line of the three-phase current circuit.

2. The waveform classification optimization method for early warning of potential hazards in power secondary circuits according to claim 1, characterized in that, The classification and labeling of the three-phase current recording file window data according to the first numerical judgment principle yields the labeled target three-phase current recording file window data, including: Determine whether the data in a certain three-phase current recording file window meets a first preset condition, wherein the first preset condition is: under the condition of a single-phase short circuit fault, there exists a non-fault phase current that is less than 10% of the fault phase current and in the opposite direction, and the corresponding non-fault phase second harmonic content is less than 10%; If the first preset condition is met, the data in the window of a certain three-phase current recording file will be marked as normal. Determine whether the data in a certain three-phase current recording file window meets the second preset condition, wherein the second preset condition is: under the condition of single-phase short circuit fault, there is a non-fault phase current that is 10% greater than the fault phase current and in the opposite direction, the corresponding non-fault phase second harmonic content is greater than 15%, and the duration is less than 20ms. If the second preset condition is met, the data in the window of a certain three-phase current recording file will be marked as a type with larger resistance, wherein the resistance range of the type with larger resistance is 0.6 ohms to 10 ohms; Determine whether the data in a certain three-phase current recording file window meets the third preset condition, wherein the third preset condition is: under the condition of single-phase short circuit fault, there is a non-fault phase current that is 15% greater than the fault phase current and in the opposite direction, the corresponding non-fault phase second harmonic content is greater than 15%, and the duration is greater than 20ms. If the third preset condition is met, the data in the window of a certain three-phase current recording file will be marked as having excessive resistance. Determine whether the data in a certain three-phase current recording file window meets the fourth preset condition. The fourth preset condition is: under the condition of single-phase short circuit fault, there is a non-fault phase current that is 15% greater than the fault phase current and in the opposite direction, the corresponding non-fault phase second harmonic content is greater than 30%, and the duration is greater than 20ms. If the fourth preset condition is met, the data in the window of a certain three-phase current recording file will be marked as a complete disconnection type.

3. The waveform classification optimization method for early warning of potential hazards in power secondary circuits according to claim 1, characterized in that, The classification and labeling of the three-phase current recording file window data according to the second numerical judgment principle yields the labeled target three-phase current recording file window data, including: If the main transformer uses YNynd11, then the following judgment process is executed: When in the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording file window, there is a non-fault phase current that is less than 2 / fault phase current and in the opposite direction, and the third harmonic content of the corresponding non-fault phase is less than 10%, then the data in that three-phase current recording file window is marked as normal. If, in the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording file window, there is a non-fault phase current that is greater than 2 / fault phase current and in the opposite direction, and the third harmonic content of the corresponding non-fault phase is greater than 15% and the duration is less than 20ms, then the data in the certain three-phase current recording file window is marked as a type with larger resistance. The resistance range of the type with larger resistance is 0.6 ohms to 10 ohms. If, in the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording file window, there is a non-fault phase current that is greater than 2 / fault phase current and in the opposite direction, and the third harmonic content of the corresponding non-fault phase is greater than 15% and the duration is greater than 20ms, then the data in the certain three-phase current recording file window will be marked as excessive resistance type. If, in the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording file window, there is a non-fault phase current that is greater than 2 / fault phase current and in the opposite direction, and the corresponding third harmonic content of the non-fault phase is greater than 30% and the duration is greater than 20ms, then the data in that three-phase current recording file window will be marked as a complete disconnection type.

4. The waveform classification optimization method for early warning of potential hazards in power secondary circuits according to claim 1, characterized in that, The classification and labeling of the three-phase current recording file window data according to the second numerical judgment principle yields the labeled target three-phase current recording file window data, including: If the main transformer uses YNyd11, then the following judgment process is executed: If, in the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording file window, there is a non-fault phase current that is less than 30% of the fault phase current and the phase is 160° ahead, and the corresponding fifth harmonic content of the non-fault phase is less than 10%, then the data in that three-phase current recording file window will be marked as normal. If, in the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording file window, there is a non-fault phase current that is 30% greater than the fault phase current and has a phase lead of 160°, the corresponding fifth harmonic content of the non-fault phase is greater than 15%, and the duration is less than 20ms, then the data in that three-phase current recording file window will be marked as a type with higher resistance. The resistance range of the type with higher resistance is 0.6 ohms to 10 ohms. If, in the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording file window, there is a non-fault phase current that is 30% greater than the fault phase current and has a phase lead of 160°, the corresponding fifth harmonic content of the non-fault phase is greater than 15%, and the duration is greater than 20ms, then the data in that three-phase current recording file window will be marked as having excessive resistance. If, in the secondary circuit current recording of the medium voltage side of the main transformer in a certain three-phase current recording file window, there is a non-fault phase current that is 30% greater than the fault phase current and has a phase lead of 160°, and the corresponding fifth harmonic content of the non-fault phase is greater than 30% and the duration is greater than 20ms, then the data in that three-phase current recording file window will be marked as a complete disconnection type.

5. A waveform classification and optimization system for early warning of potential hazards in power secondary circuits, characterized in that, include: The construction module is configured to build a large power grid simulation model based on RTDS. The large power grid simulation model includes a single-phase high-resistivity grounding fault simulation model and a neutral line fault simulation model. The simulation module is configured to perform ergonomic simulations on the single-phase high-resistance ground fault simulation model and the neutral line fault simulation model, and to obtain the three-phase current waveform file window data under the condition of single-phase fault and superimposed neutral line fault of the transmission line. The annotation module is configured to classify and annotate the three-phase current recording file window data based on a preset numerical judgment principle, thereby obtaining annotated target three-phase current recording file window data. The classification and annotation of the three-phase current recording file window data based on the preset numerical judgment principle to obtain the annotated target three-phase current recording file window data includes: For the current loops of line protection, bus protection branches, and circuit breaker protection devices, the three-phase current waveform file window data is classified and labeled according to the first numerical judgment principle to obtain the labeled target three-phase current waveform file window data. For the protection of the main transformer, when the single-phase ground fault occurs on the high-voltage side of the main transformer grid, and the neutral line is loosely connected or disconnected in the medium-voltage side current secondary circuit of the main transformer protection device, the three-phase current waveform file window data is classified and labeled according to the second numerical judgment principle to obtain the labeled target three-phase current waveform file window data. The conversion module is configured to convert the target three-phase current waveform file window data into a two-dimensional image using Gram angle and field GASF, and to combine the various two-dimensional images into at least one image unit. The step of converting the target three-phase current waveform file window data into a two-dimensional image using Gram angle and field GASF, and combining the various two-dimensional images into at least one image unit includes: The target three-phase current waveform file window data is normalized according to the preset data processing rules to obtain the normalized target three-phase current waveform file window data. The data processing rules are: (current sampling point of the window - average sampling point of the window) / standard deviation of the window sampling point or (current sampling point of the window - minimum sampling point of the window) / (maximum sampling point of the window - minimum sampling point of the window). The normalized target three-phase current waveform file window data is converted into a two-dimensional image using Gram angle and field GASF, and the individual two-dimensional images are combined into at least one image unit. The training module is configured to input the at least one image unit into a preset VGG16 convolutional neural network for iterative training to obtain a fault diagnosis model for the neutral line of the current secondary circuit. The output module is configured to acquire real-time fault recording time-series data during a single-phase fault in the transmission line at the station end based on the fault recording master station platform, and input the real-time fault recording time-series data into the neutral line fault diagnosis model of the current secondary circuit. The neutral line fault diagnosis model of the current secondary circuit outputs the waveform classification results at the neutral line of the three-phase current circuit.

6. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method described in any one of claims 1 to 4.

Images