Charging pile safety diagnosis method and system combined with artificial intelligence

Abstract

This invention provides a charging pile safety diagnosis method and system combining artificial intelligence. It acquires the original current and voltage traveling wave signal sequences at the moment a charging pile power transmission line fault occurs, and decomposes them into incident, reflected, and transmitted segments. A pre-trained fault evolution model is invoked to perform fault boundary inversion, generating boundary reflection and transmission coefficient distribution maps. Fault impedance and location feature codes are obtained through joint decoding of fault features. Based on the fault impedance and location feature codes, a safety response strategy mapping is performed to generate a safety protection command sequence containing power module shutdown instructions and fault area isolation instructions, which is then sent to the charging pile power controller and line switcher. This invention achieves rapid and accurate fault diagnosis and safety protection for charging piles, improving the safety and reliability of charging pile operation.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, and more specifically, to a method and system for safety diagnosis of charging piles that incorporates artificial intelligence. Background Technology

[0002] With the rapid development of new energy vehicles, charging piles, as crucial infrastructure, are of paramount importance in terms of safety and reliability. During operation, power transmission lines of charging piles may malfunction for various reasons, such as insulation aging, short circuits, and open circuits. If these faults are not diagnosed and handled promptly and accurately, they can not only damage the charging pile equipment but also cause safety accidents such as fires and electric shocks, seriously threatening the lives and property of users. Currently, charging pile fault diagnosis mainly relies on traditional protection devices and manual inspection methods. Traditional protection devices typically judge faults based on fixed thresholds, making it difficult to accurately identify the type and location of faults in complex and changing fault situations. Furthermore, their response speed is slow, failing to provide timely and effective equipment protection. Manual inspection methods are inefficient, unable to monitor the operating status of charging piles in real time, and unable to promptly detect potential fault hazards. In addition, most existing fault diagnosis methods rely solely on single electrical parameters for judgment, failing to fully consider the propagation characteristics of current and voltage traveling waves during fault occurrence, resulting in insufficient accuracy and comprehensiveness in fault diagnosis. Summary of the Invention

[0003] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a charging pile safety diagnosis method incorporating artificial intelligence, the method comprising:

[0004] The original current traveling wave signal sequence and the original voltage traveling wave signal sequence generated at the moment of the fault on the power transmission line of the charging pile are obtained. The original current traveling wave signal sequence includes the transient current change trajectory at the output end of the power module inside the charging pile, and the original voltage traveling wave signal sequence includes the transient voltage change trajectory at the input end of the power module inside the charging pile.

[0005] The original current traveling wave signal sequence is decomposed into a set of incident current traveling wave segments and a set of reflected current traveling wave segments, and the original voltage traveling wave signal sequence is decomposed into a set of incident voltage traveling wave segments and a set of transmitted voltage traveling wave segments;

[0006] The pre-trained fault evolution model is invoked to perform fault boundary inversion operation on the set of reflected current traveling wave segments and the set of transmitted voltage traveling wave segments, generating boundary reflection coefficient distribution map and boundary transmission coefficient distribution map of the fault point inside the charging pile;

[0007] Perform a joint fault feature decoding operation on the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map to obtain the fault impedance feature code and fault location feature code of the fault point inside the charging pile.

[0008] A safety response strategy mapping operation is performed on the fault impedance feature code and the fault location feature code to generate a safety protection command sequence containing a power module shutdown command and a fault area isolation command, and the safety protection command sequence is sent to the power controller and line switch of the charging pile.

[0009] In another aspect, embodiments of the present invention also provide a charging pile safety diagnostic system that incorporates artificial intelligence, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the machine-readable storage medium to implement the above-described method.

[0010] Based on the above, this embodiment of the invention acquires the original current traveling wave signal sequence and the original voltage traveling wave signal sequence at the moment of fault occurrence on the power transmission line of the charging pile. This allows for a comprehensive capture of the transient electrical characteristics at the time of fault occurrence. The original signal sequence is decomposed into sets of traveling wave segments of different types. A pre-trained fault evolution model is invoked to perform fault boundary inversion operations on the sets of reflected current traveling wave segments and transmitted voltage traveling wave segments, generating boundary reflection coefficient distribution maps and boundary transmission coefficient distribution maps. These maps accurately reflect the boundary characteristics of the fault point. A joint fault feature decoding operation is then performed on the boundary reflection coefficient distribution maps and boundary transmission coefficient distribution maps to obtain fault impedance feature codes and fault location feature codes, achieving accurate identification of the fault type and location. Finally, a safety response strategy mapping operation is performed on the fault impedance feature codes and fault location feature codes to generate a safety protection command sequence containing power module shutdown instructions and fault area isolation instructions. This sequence is then sent to the power controller and line switch of the charging pile, enabling timely and effective protection measures to prevent fault escalation and ensure the safe operation of the charging pile. This method has advantages such as accurate diagnosis, rapid response, and comprehensive protection, effectively improving the safety and reliability of charging piles. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the execution flow of the charging pile safety diagnosis method combined with artificial intelligence provided in an embodiment of the present invention.

[0012] Figure 2 This is a schematic diagram of exemplary hardware and software components of the charging pile safety diagnostic system that combines artificial intelligence, provided in an embodiment of the present invention. Detailed Implementation

[0013] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1This is a flowchart illustrating a charging pile safety diagnosis method incorporating artificial intelligence, provided in one embodiment of the present invention. The following is a detailed description of this charging pile safety diagnosis method incorporating artificial intelligence.

[0014] Step S110: Obtain the original current traveling wave signal sequence and the original voltage traveling wave signal sequence generated at the moment of the fault on the power transmission line of the charging pile. The original current traveling wave signal sequence includes the transient current change trajectory at the output end of the power module inside the charging pile, and the original voltage traveling wave signal sequence includes the transient voltage change trajectory at the input end of the power module inside the charging pile.

[0015] In this embodiment, to accurately diagnose internal faults in the charging pile, it is first necessary to acquire the key electrical signals at the time of the fault. Specifically, during normal operation of the charging pile, when a transient event indicating a potential fault is detected, the current and voltage traveling wave signals on the power transmission line are immediately acquired. These traveling wave signals can carry rich information at the moment of the fault, including key features such as the type and location of the fault. The original current traveling wave signal sequence reflects the transient change of the current at the output terminal of the power module inside the charging pile at the time of the fault, while the original voltage traveling wave signal sequence corresponds to the transient change trajectory of the voltage at the input terminal of the power module.

[0016] Step S111: Install a current traveling wave acquisition unit at the connection node between the output end of the charging pile power module and the transmission cable. The current traveling wave acquisition unit converts the current change on the power transmission line into a voltage signal through the principle of electromagnetic induction to obtain the original current traveling wave electrical signal.

[0017] In this embodiment, to obtain the original current traveling wave signal sequence, a dedicated current traveling wave acquisition unit needs to be installed at the connection node between the output end of the charging pile power module and the transmission cable. This acquisition unit operates based on the principle of electromagnetic induction. When the current on the power transmission line changes, an induced electromotive force is generated in the coil of the acquisition unit, thereby converting the current change into a corresponding voltage signal. This voltage signal is the original electrical signal of the current traveling wave. For example, in a DC charging pile system, the current at the output end of the power module remains relatively stable under normal conditions. When an internal short-circuit fault occurs, the current will increase rapidly in a very short time. At this time, the current traveling wave acquisition unit can capture this transient change and convert it into a corresponding voltage signal output.

[0018] Step S112: Install a voltage traveling wave acquisition unit at the connection node between the input end of the charging pile power module and the grid access point. The voltage traveling wave acquisition unit converts the voltage change on the power transmission line into a low-voltage signal through the principle of capacitive voltage division, and obtains the original voltage traveling wave electrical signal.

[0019] To acquire the original traveling voltage signal sequence, this embodiment installs a traveling voltage acquisition unit at the connection node between the input terminal of the charging pile power module and the grid access point. This unit uses the principle of capacitive voltage division, dividing the high voltage on the line through series capacitors, thereby converting the high voltage signal into a low voltage signal suitable for subsequent processing, i.e., the original traveling voltage signal. For example, in an AC charging pile system, the voltage at the grid access point is usually a relatively high AC voltage. When an internal fault causes transient voltage fluctuations, the traveling voltage acquisition unit can acquire these fluctuations in real time and convert them into a measurable low voltage signal through capacitive voltage division.

[0020] Step S113: Input the original current traveling wave signal into the first high-speed analog-to-digital converter. The first high-speed analog-to-digital converter discretizes the original current traveling wave signal at a fixed sampling frequency and outputs the original current traveling wave signal sequence. Each discrete point in the original current traveling wave signal sequence carries a sampling time identifier and a current amplitude quantization value.

[0021] After acquiring the raw traveling wave signal, it needs to be converted into a digital signal for subsequent analysis and processing. In this embodiment, the raw traveling wave signal is input to a first high-speed analog-to-digital converter. This converter samples and quantizes the continuous raw traveling wave signal at a fixed sampling frequency, achieving signal discretization. The selection of the sampling frequency needs to be determined based on the characteristics of the traveling wave signal to ensure accurate capture of the transient changes of the traveling wave. After discretization, the output raw traveling wave signal sequence consists of a series of discrete points, each containing a corresponding sampling time identifier and a quantized value of the current amplitude at that time. For example, if the sampling frequency is set to a certain value, then at each sampling interval, a quantized value of the current amplitude will be obtained, and the corresponding sampling time will be recorded, thus forming the raw traveling wave signal sequence.

[0022] Step S114: Input the original voltage traveling wave electrical signal into the second high-speed analog-to-digital converter. The second high-speed analog-to-digital converter discretizes the original voltage traveling wave electrical signal at the same fixed sampling frequency as the first high-speed analog-to-digital converter and outputs the original voltage traveling wave signal sequence. Each discrete point in the original voltage traveling wave signal sequence carries a sampling time identifier and a voltage amplitude quantization value.

[0023] Similarly, for the original voltage traveling wave signal, this embodiment inputs it into a second high-speed analog-to-digital converter (ADC). To ensure the time synchronization of the current and voltage traveling wave signals, the second high-speed ADC uses the same fixed sampling frequency as the first high-speed ADC. Thus, when discretizing the original voltage traveling wave signal, a sequence of original voltage traveling wave signals that is time-aligned with the original current traveling wave signal sequence can be obtained. Each discrete point in this sequence also carries a sampling time identifier and a voltage amplitude quantization value. For example, when the first high-speed ADC samples the current signal at a certain fixed frequency, the second high-speed ADC also samples the voltage signal at the same frequency, so that at the same sampling time, both current amplitude quantization values ​​and voltage amplitude quantization values ​​are present.

[0024] Step S115: Store the original current traveling wave signal sequence and the original voltage traveling wave signal sequence into a circular storage buffer. After the circular storage buffer is full, use a new traveling wave signal sequence to overwrite the earliest stored traveling wave signal sequence.

[0025] To temporarily store and manage the acquired raw current traveling wave signal sequences and raw voltage traveling wave signal sequences, this embodiment employs a circular storage buffer. This buffer can continuously receive and store both types of signal sequences. When the buffer's storage space is full, newly arriving traveling wave signal sequences will overwrite the oldest stored signal sequence according to a first-in, first-out (FIFO) principle. This ensures that the buffer always stores traveling wave signals from the most recent period, preventing unlimited growth of storage space and providing timely signal data for subsequent fault diagnosis. For example, assuming the circular storage buffer's capacity can store traveling wave signals for a certain duration, after that duration, new signal data will sequentially overwrite the oldest data, ensuring that the buffer always maintains the most up-to-date signal information.

[0026] Step S120: Decompose the original current traveling wave signal sequence into a set of incident current traveling wave segments and a set of reflected current traveling wave segments, and decompose the original voltage traveling wave signal sequence into a set of incident voltage traveling wave segments and a set of transmitted voltage traveling wave segments.

[0027] After obtaining the original current traveling wave signal sequence and the original voltage traveling wave signal sequence, they need to be decomposed to extract the traveling wave segments related to the fault. In this embodiment, the original current traveling wave signal sequence is decomposed into a set of incident current traveling wave segments and a set of reflected current traveling wave segments, and the original voltage traveling wave signal sequence is decomposed into a set of incident voltage traveling wave segments and a set of transmitted voltage traveling wave segments. The incident traveling wave is usually the traveling wave that propagates from the fault point to both ends of the line after the fault occurs, while the reflected and transmitted traveling waves are generated when the incident traveling wave encounters impedance discontinuities in the line (such as the fault point).

[0028] Step S121: Extract the rising edge trigger time of the current traveling wave waveform from the original current traveling wave signal sequence as the current traveling wave start reference time, extract the rising edge trigger time of the voltage traveling wave waveform from the original voltage traveling wave signal sequence as the voltage traveling wave start reference time, and use the time difference between the current traveling wave start reference time and the voltage traveling wave start reference time as the traveling wave propagation direction discrimination parameter.

[0029] To accurately decompose traveling wave signals, it is first necessary to determine the start time and propagation direction of the traveling wave. In this embodiment, the rising edge trigger time of the current traveling wave waveform is extracted from the original current traveling wave signal sequence as the reference time for the start of the current traveling wave. The rising edge trigger time refers to the moment when the traveling wave signal begins to rise from the baseline, marking the arrival of the traveling wave. Similarly, the rising edge trigger time of the voltage traveling wave waveform is extracted from the original voltage traveling wave signal sequence as the reference time for the start of the voltage traveling wave. Then, the time difference between the reference times for the start of the current traveling wave and the voltage traveling wave is calculated, and this difference is used as a parameter for determining the propagation direction of the traveling wave. This parameter can be used to determine the order in which the current and voltage traveling waves arrive at the acquisition point, and thus infer the propagation direction of the traveling waves. For example, if the reference time for the start of the current traveling wave is earlier than the reference time for the start of the voltage traveling wave, it indicates that the current traveling wave arrives at the acquisition point first, which may correspond to a specific fault propagation direction.

[0030] Step S122: When the traveling wave propagation direction discrimination parameter indicates that the current traveling wave arrives before the voltage traveling wave, the waveform segments in the original current traveling wave signal sequence within the first time window starting from the current traveling wave start reference time are taken as the incident current traveling wave segment set, and the waveform segments with opposite polarity that appear after the incident current traveling wave segment set in the original current traveling wave signal sequence are taken as the reflected current traveling wave segment set.

[0031] When the propagation direction discrimination parameter indicates that the current traveling wave arrives before the voltage traveling wave, the original current traveling wave signal sequence needs to be decomposed. In this embodiment, the waveform segments within the first time window starting from the reference time of the current traveling wave's initiation are determined as the incident current traveling wave segment set. The length of the first time window needs to be set according to factors such as the propagation speed of the traveling wave on the transmission line and the line length to ensure that the main part of the incident traveling wave is completely included. After the incident current traveling wave segment set, waveform segments with opposite polarities will appear in the original current traveling wave signal sequence. This is due to the reflection of the incident traveling wave at the fault point or other impedance discontinuities. These waveform segments with opposite polarities are taken as the reflected current traveling wave segment set. For example, if the incident current traveling wave is a positive waveform, after reflection at the fault point, the reflected current traveling wave is a negative waveform. By identifying this polarity change, the reflected current traveling wave segments can be accurately extracted.

[0032] Step S123: When the traveling wave propagation direction discrimination parameter indicates that the current traveling wave arrives before the voltage traveling wave, the waveform segment within the second time window starting from the voltage traveling wave in the original voltage traveling wave signal sequence is taken as the incident voltage traveling wave segment set, and the waveform segment with waveform broadening that appears after the incident voltage traveling wave segment set in the original voltage traveling wave signal sequence is taken as the transmission voltage traveling wave segment set.

[0033] When the current traveling wave arrives before the voltage traveling wave, the original voltage traveling wave signal sequence is decomposed as follows. In this embodiment, the waveform segments within the second time window starting from the reference time of the voltage traveling wave's inception are taken as the incident voltage traveling wave segment set. The length of the second time window is set similarly to the first time window, taking into account the propagation characteristics of the voltage traveling wave. After the incident voltage traveling wave segment set, waveform segments with broadened waveforms will appear in the original voltage traveling wave signal sequence. This is a manifestation of the transmission phenomenon during the propagation of the voltage traveling wave. These waveform-broadened segments are taken as the transmitted voltage traveling wave segment set. For example, when the incident voltage traveling wave passes through certain line components, its waveform will broaden. By detecting this waveform broadening characteristic, the transmitted voltage traveling wave segments can be extracted.

[0034] Step S124: When the traveling wave propagation direction discrimination parameter indicates that the voltage traveling wave arrives before the current traveling wave, the waveform segments in the original voltage traveling wave signal sequence within the first time window starting from the voltage traveling wave initiation reference time are taken as the incident voltage traveling wave segment set, the waveform segments with opposite polarity appearing after the incident voltage traveling wave segment set in the original voltage traveling wave signal sequence are taken as the reflected voltage traveling wave segment set, and the waveform segments with waveform broadening appearing after the current traveling wave in the original current traveling wave signal sequence are taken as the transmitted current traveling wave segment set.

[0035] When the propagation direction discrimination parameters indicate that the voltage traveling wave arrives before the current traveling wave, the decomposition method differs. In this case, the waveform segments within the first time window starting from the voltage traveling wave's initial reference time in the original voltage traveling wave signal sequence are taken as the incident voltage traveling wave segment set. Waveform segments with opposite polarity appearing after the incident voltage traveling wave segment set are taken as the reflected voltage traveling wave segment set. Simultaneously, for the original current traveling wave signal sequence, waveform segments with broadened waveforms appearing after the current traveling wave's initial reference time are taken as the transmitted current traveling wave segment set. In this case, the reflection characteristics of the voltage traveling wave and the transmission characteristics of the current traveling wave become the focus of the decomposition, and the corresponding traveling wave segment sets are extracted through corresponding waveform feature identification.

[0036] Step S125: The marked sets of incident current traveling wave segments, reflected current traveling wave segments, incident voltage traveling wave segments, and transmitted voltage traveling wave segments are arranged and aligned in chronological order to generate a time-synchronized traveling wave segment combination sequence. Each time position in the traveling wave segment combination sequence contains at least one of the following: incident current traveling wave segment, reflected current traveling wave segment, incident voltage traveling wave segment, and transmitted voltage traveling wave segment.

[0037] After labeling the various traveling wave segment sets, they need to be aligned temporally to form a unified traveling wave segment combination sequence. In this embodiment, the labeled incident current traveling wave segment sets, reflected current traveling wave segment sets, incident voltage traveling wave segment sets, and transmitted voltage traveling wave segment sets are sorted according to their respective time labels. Through time alignment, each time position in the traveling wave segment combination sequence can simultaneously contain at least one type of traveling wave segment. In this way, the subsequent fault evolution model can analyze and process different types of traveling wave segments in the same time dimension, improving the accuracy and efficiency of fault diagnosis. For example, at a certain time position, the traveling wave segment combination sequence may simultaneously contain incident current traveling wave segments and transmitted voltage traveling wave segments. These two segments are synchronous in time, facilitating joint analysis by the model.

[0038] Step S1251: Obtain the time start mark of each incident current traveling wave segment in the incident current traveling wave segment set, obtain the time start mark of each reflected current traveling wave segment in the reflected current traveling wave segment set, and pair the incident current traveling wave segments and reflected current traveling wave segments with a time start mark difference less than a preset time threshold as current traveling wave segment groups corresponding to the same fault event.

[0039] To achieve accurate pairing and time alignment of traveling wave segments, the current traveling wave segments are first processed. In this embodiment, the time start marker of each incident current traveling wave segment in the incident current traveling wave segment set and the time start marker of each reflected current traveling wave segment in the reflected current traveling wave segment set are obtained. Then, the difference between the time start markers of each incident current traveling wave segment and the reflected current traveling wave segment is calculated. When this difference is less than a preset time threshold, the two traveling wave segments are considered to belong to the same fault event and are paired as a current traveling wave segment group. The setting of the preset time threshold needs to take into account the propagation time and possible delay of the traveling wave inside the charging pile to ensure that the incident and reflected traveling waves generated by the same fault event can be correctly paired. For example, if the preset time threshold is a small time value, when the difference between the time start markers of the incident current traveling wave segment and the reflected current traveling wave segment is within this range, they are paired as a current traveling wave segment group of the same fault event.

[0040] Step S1252: Obtain the time start mark of each incident voltage traveling wave segment in the incident voltage traveling wave segment set, obtain the time start mark of each transmitted voltage traveling wave segment in the transmitted voltage traveling wave segment set, and pair the incident voltage traveling wave segments and transmitted voltage traveling wave segments with a time start mark difference less than a preset time threshold as voltage traveling wave segment groups corresponding to the same fault event.

[0041] For voltage traveling wave segments, a similar pairing method is used. The time start markers of each incident voltage traveling wave segment in the incident voltage traveling wave segment set and each transmitted voltage traveling wave segment in the transmitted voltage traveling wave segment set are obtained. The difference between the two time start markers is calculated. When the difference is less than a preset time threshold, the corresponding incident voltage traveling wave segment and transmitted voltage traveling wave segment are paired as a voltage traveling wave segment group corresponding to the same fault event. In this way, voltage traveling wave segments related to the same fault event can be grouped together.

[0042] Step S1253: Associate and pair the current traveling wave segment group with the group whose time start mark is closest to that of the voltage traveling wave segment group to form a fault event traveling wave segment group. The fault event traveling wave segment group includes incident current traveling wave segment, reflected current traveling wave segment, incident voltage traveling wave segment and transmitted voltage traveling wave segment.

[0043] After pairing the current traveling wave segment groups and voltage traveling wave segment groups, they need to be further correlated to form a complete fault event traveling wave segment group. In this embodiment, for each current traveling wave segment group, the group with the closest time start markers among all voltage traveling wave segment groups is found, and these two groups are correlated and paired to form a fault event traveling wave segment group that includes incident current traveling wave segments, reflected current traveling wave segments, incident voltage traveling wave segments, and transmitted voltage traveling wave segments. In this way, each fault event traveling wave segment group contains various types of traveling wave segments related to the same fault event, facilitating the analysis of fault characteristics from multiple perspectives.

[0044] Step S1254: Arrange all fault event traveling wave segment groups into a time-synchronized traveling wave segment combination sequence according to the order of the time start markers of each fault event traveling wave segment group. Each element in the traveling wave segment combination sequence corresponds to a fault event traveling wave segment group.

[0045] After forming the traveling wave segment groups of fault events, they need to be arranged in chronological order. In this embodiment, based on the order of the time start markers of each fault event traveling wave segment group, all fault event traveling wave segment groups are arranged sequentially to form a time-synchronized traveling wave segment combination sequence. Each element in the traveling wave segment combination sequence corresponds to a fault event traveling wave segment group, thus achieving an ordered arrangement of different fault events in the time dimension.

[0046] Step S1255: For a fault event traveling wave segment group that is missing a certain type of traveling wave segment in the time-synchronized traveling wave segment combination sequence, fill the missing position with an all-zero waveform segment.

[0047] In actual traveling wave signal acquisition and decomposition, some fault event traveling wave segment groups may lack certain types of traveling wave segments. For example, a fault event may only detect incident current and reflected current traveling wave segments, but not incident voltage and transmitted voltage traveling wave segments. To ensure the integrity of the traveling wave segment combination sequence and the uniformity of the data format, this embodiment uses all-zero waveform segments to fill the missing positions for such fault event traveling wave segment groups lacking traveling wave segments. All-zero waveform segments indicate that there is no corresponding type of traveling wave signal at that time position, thus ensuring that each element in the traveling wave segment combination sequence has the same data structure.

[0048] Step S130: Call the pre-trained fault evolution model to perform fault boundary inversion operation on the set of reflected current traveling wave segments and the set of transmitted voltage traveling wave segments to generate boundary reflection coefficient distribution map and boundary transmission coefficient distribution map of the fault points inside the charging pile.

[0049] After obtaining the sets of reflected current traveling wave segments and transmitted voltage traveling wave segments, this embodiment calls a pre-trained fault evolution model to perform fault boundary inversion operations on these two sets. The fault evolution model is trained based on a large amount of fault data and can invert the boundary characteristics of the fault point based on the characteristics of the input traveling wave segments. Through the fault boundary inversion operation, boundary reflection coefficient distribution maps and boundary transmission coefficient distribution maps of the fault points inside the charging pile are generated. The boundary reflection coefficient distribution map reflects the variation of the reflection characteristics of the fault point to the incident current traveling wave with frequency, while the boundary transmission coefficient distribution map reflects the variation of the transmission characteristics of the fault point to the incident voltage traveling wave with frequency. These two distribution maps contain important feature information of the fault point.

[0050] Step S131: Pair each reflected current traveling wave segment in the set of reflected current traveling wave segments with the incident current traveling wave segment in the set of incident current traveling wave segments to obtain multiple pairs of reflected and incident current traveling wave segments.

[0051] To perform fault boundary inversion, the reflected current traveling wave segments must first be paired with their corresponding incident current traveling wave segments. In this embodiment, each reflected current traveling wave segment is extracted from the set of reflected current traveling wave segments, and then the incident current traveling wave segment that corresponds to the reflected current traveling wave segment in time is found in the set of incident current traveling wave segments. The temporal correspondence can be determined by the previous time stamps; that is, reflected current traveling wave segments and incident current traveling wave segments with the same or similar time start stamps are considered paired. Through this pairing process, multiple pairs of reflected and incident current traveling wave segments are obtained, each pair containing one reflected current traveling wave segment and its corresponding incident current traveling wave segment.

[0052] Step S132: Perform a frequency domain transformation operation on each pair of reflected incident current traveling wave segments to convert the pair of reflected incident current traveling wave segments from the time domain representation to the frequency domain representation, thereby obtaining the frequency domain spectrum of the reflected current and the frequency domain spectrum of the incident current.

[0053] For each pair of reflected incident current traveling wave segments, it is necessary to transform them from the time domain to the frequency domain in order to analyze the frequency characteristics of the traveling wave. In this embodiment, a frequency domain transformation operation is performed on each pair of reflected incident current traveling wave segments. Common frequency domain transformation methods include Fourier transform. Through frequency domain transformation, the time-domain reflected current traveling wave segment is converted into the reflected current frequency domain spectrum, and the time-domain incident current traveling wave segment is converted into the incident current frequency domain spectrum. The frequency domain spectrum can clearly show the amplitude and phase information of the traveling wave at different frequency components, which is crucial for calculating parameters such as the reflection coefficient.

[0054] Step S133: The ratio of the amplitude of the reflected current frequency domain spectrum to the amplitude of the incident current frequency domain spectrum in each frequency component is taken as the reflection coefficient value corresponding to the frequency component. The reflection coefficient values ​​of all frequency components are arranged in frequency order to generate a boundary reflection coefficient distribution map.

[0055] After obtaining the frequency domain spectra of the reflected current and the incident current, the reflection coefficient value for each frequency component is calculated. In this embodiment, for each frequency component, the amplitude of the reflected current frequency domain spectrum at that frequency component is divided by the amplitude of the incident current frequency domain spectrum at the same frequency component to obtain the reflection coefficient value corresponding to that frequency component. Then, the reflection coefficient values ​​of all frequency components are arranged in ascending order of frequency to form a boundary reflection coefficient distribution map. This distribution map, with frequency as the horizontal axis and reflection coefficient value as the vertical axis, visually displays the reflection characteristics of the fault point at different frequencies.

[0056] Step S134: Pair each transmitted voltage traveling wave segment in the transmitted voltage traveling wave segment set with the incident voltage traveling wave segment in the incident voltage traveling wave segment set to obtain multiple pairs of transmitted incident voltage traveling wave segments.

[0057] Similar to the pairing of reflected current traveling wave segments, transmitted voltage traveling wave segments also need to be paired with corresponding incident voltage traveling wave segments. In this embodiment, each transmitted voltage traveling wave segment in the transmitted voltage traveling wave segment set is paired with the corresponding incident voltage traveling wave segment in the incident voltage traveling wave segment set, resulting in multiple pairs of transmitted and incident voltage traveling wave segments. The temporal correspondence is also determined by a time start marker to ensure that the paired traveling wave segments belong to the same fault event.

[0058] Step S135: Perform a frequency domain transformation operation on each transmitted incident voltage traveling wave segment to obtain the transmitted voltage frequency domain spectrum and the incident voltage frequency domain spectrum. Take the ratio of the amplitude of the transmitted voltage frequency domain spectrum to the amplitude of the incident voltage frequency domain spectrum at each frequency component as the transmission coefficient value corresponding to the frequency component. Arrange the transmission coefficient values ​​of all frequency components in frequency order to generate a boundary transmission coefficient distribution map.

[0059] For each transmitted incident voltage traveling wave segment, a frequency domain transformation is performed to obtain the transmitted voltage frequency domain spectrum and the incident voltage frequency domain spectrum. Then, for each frequency component, the amplitude of the transmitted voltage frequency domain spectrum is divided by the amplitude of the incident voltage frequency domain spectrum to obtain the transmission coefficient value corresponding to that frequency component. The transmission coefficient values ​​of all frequency components are arranged in frequency order to generate a boundary transmission coefficient distribution map. This distribution map is similar to the boundary reflection coefficient distribution map, showing the transmission characteristics of the fault point at different frequencies.

[0060] Step S136: Input the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map into the feature association layer of the fault evolution model. The feature association layer performs cross-validation on the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map, identifies the frequency feature peaks that match each other in the reflection coefficient distribution map and the transmission coefficient distribution map, and generates boundary reflection coefficient distribution map and boundary transmission coefficient distribution map that have passed consistency verification.

[0061] To improve the reliability of the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map, cross-validation and consistency verification are required. In this embodiment, these two distribution maps are input into the feature association layer of the fault evolution model. The feature association layer identifies matching feature peaks by analyzing the frequency characteristic peaks in the two distribution maps. Matching feature peaks refer to peaks that appear at the same or similar frequency positions and have a certain correlation. Through this cross-validation operation, some false feature peaks caused by noise or interference can be eliminated, retaining the true and valid feature information, thereby generating boundary reflection coefficient distribution maps and boundary transmission coefficient distribution maps that have passed consistency verification.

[0062] Step S1361: Represent the boundary reflection coefficient distribution map as a reflection coefficient sequence composed of frequency index and reflection coefficient value, and represent the boundary transmission coefficient distribution map as a transmission coefficient sequence composed of frequency index and transmission coefficient value, wherein the reflection coefficient sequence and the transmission coefficient sequence have the same frequency index range.

[0063] To facilitate cross-validation in the feature association layer, the boundary reflectance coefficient distribution map and the boundary transmission coefficient distribution map first need to be converted into sequence form. In this embodiment, the boundary reflectance coefficient distribution map is represented as a reflectance coefficient sequence, which consists of a series of frequency indices and corresponding reflectance coefficient values. Similarly, the boundary transmission coefficient distribution map is represented as a transmission coefficient sequence, consisting of frequency indices and corresponding transmission coefficient values. Furthermore, it is ensured that the reflectance coefficient sequence and the transmission coefficient sequence have the same frequency index range, that is, they contain the same frequency components, so that they can be compared and validated on the same frequency dimension.

[0064] Step S1362: For each frequency index position in the reflection coefficient sequence, obtain the reflection coefficient value corresponding to the frequency index position, and at the same time obtain the transmission coefficient value corresponding to the same frequency index position in the transmission coefficient sequence. Compare the sum of the squares of the reflection coefficient value and the squares of the transmission coefficient value with a preset physical constant. When the absolute value of the difference exceeds a preset error threshold, the frequency index position is identified as an inconsistent frequency point.

[0065] After obtaining the reflection coefficient sequence and the transmission coefficient sequence, a consistency check is performed on each frequency index position. In this embodiment, for each frequency index position in the reflection coefficient sequence, the corresponding reflection coefficient value and the transmission coefficient value at the same frequency index position in the transmission coefficient sequence are obtained. Then, the sum of the squares of the reflection coefficient value and the squares of the transmission coefficient value is calculated. According to physical principles, under ideal conditions without loss, this sum should equal a preset physical constant (for example, in some cases, this constant may be 1). The calculated sum is compared with the preset physical constant. If the absolute value of the difference exceeds a preset error threshold, the frequency index position is identified as an inconsistent frequency point. Inconsistent frequency points may be caused by measurement errors, noise interference, or model calculation deviations.

[0066] Step S1363: For the frequency index position identified as an inconsistent frequency point, if the reflection coefficient value is greater than the transmission coefficient value, keep the reflection coefficient value unchanged and redetermine the transmission coefficient value according to the physical constant; if the transmission coefficient value is greater than the reflection coefficient value, keep the transmission coefficient value unchanged and redetermine the reflection coefficient value according to the physical constant.

[0067] For identified inconsistent frequency points, corrections are needed to ensure physical consistency between the reflection and transmission coefficients. In this embodiment, when the reflection coefficient is greater than the transmission coefficient, the reflection coefficient is kept constant, and the transmission coefficient is recalculated based on preset physical constants and the reflection coefficient. Specifically, the transmission coefficient can be obtained by subtracting the square of the reflection coefficient from the physical constant and then taking the square root. Conversely, when the transmission coefficient is greater than the reflection coefficient, the transmission coefficient is kept constant, and the reflection coefficient is re-determined based on the physical constants. This method ensures that the corrected reflection and transmission coefficients satisfy the constraints of the physical constants, improving data reliability.

[0068] Step S1364: Rearrange the reflection coefficient values ​​of all frequency index positions after coefficient adjustment according to the frequency index order to generate a boundary reflection coefficient distribution map that has passed the consistency check. Rearrange the transmission coefficient values ​​of all frequency index positions after coefficient adjustment according to the frequency index order to generate a boundary transmission coefficient distribution map that has passed the consistency check.

[0069] After correcting all inconsistent frequency points, the adjusted reflection coefficient values ​​are rearranged according to frequency index order to form a boundary reflection coefficient distribution map that has passed consistency verification. Similarly, the adjusted transmission coefficient values ​​are rearranged according to frequency index order to generate a boundary transmission coefficient distribution map that has passed consistency verification. These two verified distribution maps eliminate inconsistent frequency points and more accurately reflect the boundary characteristics of the fault point.

[0070] Step S140: Perform a joint fault feature decoding operation on the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map to obtain the fault impedance feature code and fault location feature code of the fault point inside the charging pile.

[0071] After obtaining the boundary reflection coefficient distribution map and boundary transmission coefficient distribution map that have passed consistency verification, they need to be jointly decoded to extract the key features of the fault point. In this embodiment, through the joint decoding operation of fault features, the fault impedance feature code and fault location feature code of the fault point inside the charging pile are parsed from these two distribution maps. The fault impedance feature code reflects the impedance characteristics of the fault point, and different impedance characteristics correspond to different types of faults (such as short circuit faults, ground faults, etc.). The fault location feature code indicates the specific location information of the fault point inside the charging pile.

[0072] Step S141: Input the boundary reflection coefficient distribution map into the first decoding branch of the fault evolution model. The first decoding branch includes a reflection coefficient feature extraction network and an impedance transformation network. The reflection coefficient feature extraction network performs a multi-scale feature convolution operation on the boundary reflection coefficient distribution map to generate a reflection coefficient feature vector. The impedance transformation network performs a nonlinear mapping operation on the reflection coefficient feature vector to obtain the fault impedance feature code.

[0073] The first decoding branch of the fault evolution model processes the boundary reflection coefficient distribution map and extracts fault impedance feature codes. In this embodiment, the boundary reflection coefficient distribution map is input into the first decoding branch. This branch first performs a multi-scale feature convolution operation on the distribution map through a reflection coefficient feature extraction network. The multi-scale feature convolution operation can extract feature information at different scales using convolution kernels of different sizes. For example, a smaller convolution kernel can be used to extract local detail features, while a larger convolution kernel can be used to extract overall trend features. After multi-scale convolution processing, a reflection coefficient feature vector is generated. Then, an impedance transformation network performs a nonlinear mapping operation on this reflection coefficient feature vector, converting the feature vector into a fault impedance feature code. The nonlinear mapping operation can be implemented through a multi-layer neural network, mapping the high-dimensional feature vector to a specific fault impedance feature space to obtain an encoding that can characterize the fault impedance properties.

[0074] Step S142: Input the boundary transmission coefficient distribution map into the second decoding branch of the fault evolution model. The second decoding branch includes a transmission coefficient feature extraction network and a distance inversion network. The transmission coefficient feature extraction network performs a time-series feature extraction operation on the boundary transmission coefficient distribution map to generate a transmission coefficient feature vector. The distance inversion network performs a regression prediction operation on the transmission coefficient feature vector to obtain the fault location feature code.

[0075] Similar to the first decoding branch, the second decoding branch processes the boundary transmission coefficient distribution map and extracts the fault location feature code. In this embodiment, the boundary transmission coefficient distribution map is input into the second decoding branch. The transmission coefficient feature extraction network performs temporal feature extraction on it. Temporal feature extraction can be achieved through recurrent neural networks (such as LSTM, GRU, etc.), which can capture the changing trend and temporal relationship of the transmission coefficient at different frequencies, generating a transmission coefficient feature vector. Then, the distance inversion network performs regression prediction on the transmission coefficient feature vector, mapping the feature vector to the numerical space of the fault location through a regression model, obtaining the fault location feature code. This code directly reflects the distance information between the fault point and the reference location.

[0076] Step S143: Input the fault impedance feature code and the fault location feature code into the joint optimization layer of the fault evolution model. The joint optimization layer performs a consistency adjustment operation based on the physical constraint relationship between the fault impedance feature code and the fault location feature code. The physical constraint relationship is that the sum of the squares of the fault point reflection coefficient and the fault point transmission coefficient is equal to a fixed constant. The consistency adjustment operation generates fault impedance feature codes and fault location feature codes that have been corrected by physical constraints.

[0077] To ensure consistency between fault impedance feature codes and fault location feature codes, their physical constraints need to be considered. In this embodiment, the fault impedance feature codes and fault location feature codes are input into the joint optimization layer of the fault evolution model. This joint optimization layer performs consistency adjustments on the two feature codes based on the physical constraint that the sum of the squares of the fault point reflection coefficient and the fault point transmission coefficient equals a fixed constant. For example, when the reflection coefficient calculated from the fault impedance feature code and the transmission coefficient calculated from the fault location feature code do not satisfy the above physical constraint, the joint optimization layer fine-tunes the two feature codes so that the adjusted reflection coefficient and transmission coefficient satisfy the relationship that the sum of their squares equals a fixed constant, thereby generating fault impedance feature codes and fault location feature codes corrected for physical constraints.

[0078] Step S144: Input the fault impedance feature code, which has been corrected by physical constraints, into the impedance classification mapper. The impedance classification mapper maps the fault impedance feature code to a preset impedance category space and outputs a fault impedance category identifier.

[0079] The fault impedance feature code, corrected for physical constraints, needs to be mapped to a specific fault impedance category. In this embodiment, the fault impedance feature code is input into an impedance classification mapper. The impedance classification mapper contains a preset impedance category space, such as low impedance faults, high impedance faults, etc. A classification algorithm (such as a support vector machine, neural network classifier, etc.) is used to match the fault impedance feature code with the preset category, ultimately outputting a fault impedance category identifier. This identifier clearly identifies the impedance type of the fault.

[0080] Step S1441: Input the fault impedance feature code, which has been corrected by physical constraints, into the input feature receiving port of the impedance classification mapper. The impedance classification mapper includes a multi-layer fully connected feature extraction network and a classification output layer. The multi-layer fully connected feature extraction network performs a layer-by-layer feature abstraction operation on the fault impedance feature code to generate a high-dimensional feature representation vector.

[0081] The impedance classification mapper operates as follows: First, the fault impedance feature code, corrected for physical constraints, is input to the input feature receiving port. Internally, the impedance classification mapper contains a multi-layer fully connected feature extraction network, consisting of multiple fully connected layers. As the fault impedance feature code passes through this multi-layer fully connected network, each layer performs linear transformations and non-linear activations on the input features, achieving layer-by-layer feature abstraction. In this way, the original fault impedance feature code is converted into a more discriminative high-dimensional feature representation vector. This high-dimensional feature representation vector better reflects the differences between different impedance categories, improving classification accuracy.

[0082] Step S1442: The classification output layer receives the high-dimensional feature representation vector and performs a linear transformation operation on the high-dimensional feature representation vector to generate a first category score and a second category score.

[0083] The classification output layer receives the high-dimensional feature representation vector generated by the multi-layer fully connected feature extraction network and performs a linear transformation operation on it. The linear transformation is implemented through a weight matrix and a bias term, mapping the high-dimensional feature representation vector to a category score space. In this embodiment, assuming the assumed impedance category space contains two categories (e.g., low-impedance fault and high-impedance fault), the linear transformation operation generates a first category score and a second category score, corresponding to the probability scores of the two categories, respectively.

[0084] Step S1443: Input the first category score and the second category score into the normalized exponential function calculation unit. The normalized exponential function calculation unit transforms the first category score into a first normalized probability value and the second category score into a second normalized probability value.

[0085] To convert category scores into probability values ​​for category decision-making, the first and second category scores are input into a normalized exponential function calculation unit. A normalized exponential function (such as the softmax function) can convert multiple category scores into probability values ​​within the interval [0,1], with the sum of all probability values ​​being 1. Through this calculation unit, the first category score is transformed into a first normalized probability value, and the second category score is transformed into a second normalized probability value. These two probability values ​​represent the probability that the fault impedance feature code belongs to the corresponding category.

[0086] Step S1444: Compare the first normalized probability value with the second normalized probability value. When the first normalized probability value is greater than the second normalized probability value, output a low impedance fault category identifier. When the second normalized probability value is greater than the first normalized probability value, output a high impedance fault category identifier.

[0087] After obtaining the first and second normalized probability values, the fault impedance category is determined by comparing their magnitudes. When the first normalized probability value is greater than the second normalized probability value, it indicates that the fault impedance feature code is more likely to belong to the first category, and a low-impedance fault category identifier is output. Conversely, when the second normalized probability value is greater than the first normalized probability value, a high-impedance fault category identifier is output. In this way, the fault impedance feature code is mapped to a specific impedance category.

[0088] Step S1445: Pass the output fault impedance category identifier to the safety response strategy mapping operation step.

[0089] Finally, the output fault impedance category identifier is passed to the subsequent safety response strategy mapping operation steps, serving as one of the important bases for formulating the safety protection command sequence.

[0090] Step S145: Input the fault location feature code after physical constraint correction into the distance regression mapper. The distance regression mapper maps the fault location feature code to a preset distance value space and outputs the fault distance value code.

[0091] Similar to the processing of fault impedance feature encoding, the fault location feature encoding, after physical constraint correction, needs to be mapped to a specific distance value. In this embodiment, the fault location feature encoding is input into a distance regression mapper. The distance regression mapper uses a regression algorithm (such as linear regression, neural network regression, etc.) to map the fault location feature encoding to a preset distance value space, and outputs a fault distance value encoding. This encoding represents the distance between the fault point and a reference location (such as a fixed point of the charging pile power module) in numerical form.

[0092] Step S150: Perform a safety response strategy mapping operation on the fault impedance feature code and the fault location feature code to generate a safety protection command sequence containing a power module shutdown command and a fault area isolation command, and send the safety protection command sequence to the power controller and line switch of the charging pile.

[0093] After obtaining the fault impedance characteristic code and fault location characteristic code, a corresponding safety response strategy needs to be formulated based on these characteristics. In this embodiment, the fault impedance characteristic code and fault location characteristic code are mapped to a specific safety protection command sequence through a safety response strategy mapping operation. This command sequence includes a power module shutdown command and a fault area isolation command, aiming to quickly cut off the fault source, prevent the fault from escalating, and isolate the fault area to ensure the safe operation of the charging pile. Finally, the generated safety protection command sequence is sent to the charging pile's power controller and line switch to execute the corresponding protection actions.

[0094] Step S151: Input the fault impedance category identifier into the first index port of the security policy mapping table. The security policy mapping table stores the correspondence between impedance category identifiers and first response action sets. Retrieve the corresponding first response action set from the security policy mapping table according to the fault impedance category identifier. The first response action set includes the sending target address and sending time parameters of the power module shutdown command.

[0095] The security response policy mapping operation begins with the fault impedance category identifier. In this embodiment, the fault impedance category identifier is input into the first index port of the security policy mapping table. The security policy mapping table is a predefined mapping table that stores the correspondence between different impedance category identifiers and corresponding first response action sets. Based on the input fault impedance category identifier, the corresponding first response action set is retrieved from the security policy mapping table. The first response action set includes the target address for sending the power module shutdown command (i.e., the address information of the power module to be shut down) and the transmission time parameter (i.e., when to send the shutdown command). For example, if the fault impedance category identifier is a low impedance fault, the corresponding first response action set may indicate that all related power modules need to be shut down immediately, with a transmission time parameter of 0 delay.

[0096] Step S152: Input the fault distance numerical code into the second index port of the security policy mapping table. The security policy mapping table stores the correspondence between distance numerical intervals and second response action sets. Retrieve the corresponding second response action set from the security policy mapping table according to the distance numerical interval to which the fault distance numerical code belongs. The second response action set includes the isolation boundary identifier and isolation order parameters of the fault area isolation instruction.

[0097] Simultaneously, the fault distance numerical code is input into the second index port of the security policy mapping table. The security policy mapping table also stores the correspondence between distance numerical intervals and second response action sets. Based on the distance numerical interval to which the fault distance numerical code belongs, the corresponding second response action set is retrieved from the security policy mapping table. The second response action set contains the isolation boundary identifier (i.e., determining the boundary range of the isolation area) and isolation sequence parameters (i.e., the order in which isolation operations are performed). For example, if the distance corresponding to the fault distance numerical code falls within a specific interval, it indicates that the fault point is located on a specific line segment; in this case, the second response action set will indicate the corresponding isolation boundary and isolation sequence.

[0098] Step S153: Combine the sending target address and sending time parameter of the power module shutdown instruction in the first response action set with the isolation boundary identifier and isolation sequence parameter of the fault area isolation instruction in the second response action set to generate a security protection command sequence. The power module shutdown instruction and the fault area isolation instruction in the security protection command sequence are arranged according to the timing relationship determined by the sending time parameter and the isolation sequence parameter.

[0099] After obtaining the first and second response action sets, they need to be combined into a unified security protection command sequence. In this embodiment, the sending target address and sending time parameters of the power module shutdown command in the first response action set, and the isolation boundary identifier and isolation order parameters of the fault area isolation command in the second response action set, are combined and encoded. During the combination process, the timing relationship between the power module shutdown command and the fault area isolation command is determined according to the sending time parameters and the isolation order parameters. For example, if the sending time parameter of the power module shutdown command indicates that the power module should be shut down first and then the area isolation should be performed, then the power module shutdown command is arranged before the fault area isolation command; otherwise, the arrangement order is adjusted. In this way, a security protection command sequence is generated, in which the commands are arranged in an orderly manner according to a predetermined timing relationship.

[0100] For example, in step S1531: extract the sending target address field and sending time parameter field of the power module shutdown instruction from the first response action set, and extract the isolation boundary identifier field and isolation sequence parameter field of the fault area isolation instruction from the second response action set.

[0101] The first step in the combined encoding operation is to extract relevant fields. From the first set of response actions, the target address field and transmission time parameter field of the power module shutdown command are extracted; these fields specify the recipient and execution time of the shutdown command. Simultaneously, from the second set of response actions, the isolation boundary identifier field and isolation sequence parameter field of the fault area isolation command are extracted to determine the scope and order of the isolation operations.

[0102] Step S1532: Compare the delay duration value in the transmission time parameter field with the order index value in the isolation sequence parameter field. When the time corresponding to the delay duration value is earlier than the time corresponding to the order index value, arrange the power module shutdown command before the fault area isolation command.

[0103] To determine the order of instructions, it's necessary to compare the delay duration value in the transmission time parameter field with the corresponding time in the order index value in the isolation sequence parameter field. For example, the delay duration value in the transmission time parameter field represents the delay time from the current moment until the power module shutdown instruction is transmitted; this is converted into a specific time. The order index value in the isolation sequence parameter field indicates the position in the isolation operation sequence and also corresponds to a specific execution time. When the transmission time of the power module shutdown instruction is earlier than the execution time of the fault area isolation instruction, the power module shutdown instruction is placed before the fault area isolation instruction.

[0104] Step S1533: When the time corresponding to the delay duration value is later than the time corresponding to the sequence index value, the fault area isolation instruction is arranged before the power module shutdown instruction, and the difference between the delay duration value and the sequence index value is used as the waiting time parameter between instructions.

[0105] If the power module shutdown command is sent later than the fault area isolation command, the fault area isolation command will be scheduled before the power module shutdown command. Simultaneously, the difference between the delay duration and the corresponding time in the order index is calculated and used as the inter-command waiting time parameter. This ensures that after executing the previous command, a corresponding waiting time is allowed before executing the next command, guaranteeing operational coordination and safety.

[0106] Step S1534: Write the power module shutdown command and fault area isolation command after the arrangement is completed into the command sequence buffer in the order of arrangement. Each command occupies an independent storage unit in the command sequence buffer. The storage unit contains a command type identifier, a command parameter field and a command sending time field.

[0107] The power module shutdown instructions and fault area isolation instructions, after being arranged, are written into the command sequence buffer in a predetermined order. Each storage unit in the command sequence buffer corresponds to one instruction, and the storage unit contains an instruction type identifier (such as "power module shutdown" or "fault area isolation"), instruction parameter fields (such as sending target address, isolation boundary identifier, etc.), and instruction sending time field (the specific sending time determined according to the previous timing relationship). This ensures that the instructions can be accurately identified and executed when sent.

[0108] Step S1535: Determine the communication port number of the power controller based on the sending target address field of the power module shutdown command, determine the line switch number to be operated based on the isolation boundary identifier field of the fault area isolation command, and append the communication port number and the line switch number to the corresponding storage unit of the command sequence buffer.

[0109] To ensure that commands are correctly sent to the target device, the corresponding communication port number and device number need to be determined. Based on the destination address field of the power module shutdown command, the communication port number of the power controller can be determined, ensuring the shutdown command is sent to the correct power controller. Based on the isolation boundary identifier field of the fault area isolation command, the line switch number that needs to be operated can be determined, ensuring the isolation command is accurately applied to the target line switch.

[0110] Step S1536: Read the complete command sequence from the command sequence buffer as a security protection command sequence output.

[0111] Finally, the complete command sequence, which is arranged in order and contains all the necessary information, is read from the command sequence buffer and output as the security protection command sequence.

[0112] Step S154: Send the power module shutdown command in the safety protection command sequence to the command receiving port of the power controller of the charging pile according to the delay time specified by the sending time parameter.

[0113] After generating the safety protection command sequence, the power module shutdown command needs to be sent to the charging pile's power controller. In this embodiment, the command is sent to the power controller's command receiving port at the appropriate time according to the delay duration specified by the power module shutdown command's transmission time parameter. Upon receiving the command, the power controller executes the power module shutdown operation, cutting off the power supply to the fault source.

[0114] Step S155: Send the fault area isolation instructions in the safety protection command sequence to the instruction receiving port of the charging pile's line switch in the order specified by the isolation sequence parameter.

[0115] Simultaneously, the fault area isolation commands in the safety protection command sequence are sent to the command receiving port of the charging pile's line switch in the order specified by the isolation sequence parameter. The line switch executes the isolation operation according to the received command sequence, isolating the fault area from other normal areas and preventing further spread of the fault. Through this series of safety protection measures, the safe operation of the charging pile can be effectively guaranteed, and losses caused by faults can be reduced.

[0116] Figure 2 The illustration shows exemplary hardware and software components of an artificial intelligence-integrated charging pile safety diagnostic system 100, which can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 can be used in the artificial intelligence-integrated charging pile safety diagnostic system 100 and to perform the functions in this application.

[0117] The AI-integrated charging pile safety diagnostic system 100 can be a general-purpose server or a special-purpose server; both can be used to implement the AI-integrated charging pile safety diagnostic method of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0118] For example, the AI-integrated charging pile safety diagnostic system 100 may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the AI-integrated charging pile safety diagnostic system 100 may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The AI-integrated charging pile safety diagnostic system 100 also includes an I / O interface 150 between the computer and other input / output devices.

[0119] For ease of explanation, only one processor is described in the AI-integrated charging pile safety diagnostic system 100. However, it should be noted that the AI-integrated charging pile safety diagnostic system 100 of this application may also include multiple processors. Therefore, the steps performed by one processor described in this application may also be performed jointly or individually by multiple processors. For example, if the processor of the AI-integrated charging pile safety diagnostic system 100 performs steps A and B, it should be understood that steps A and B may also be performed jointly by two different processors or individually by one processor. For example, the first processor performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0120] Furthermore, this embodiment of the invention also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the above-mentioned charging pile safety diagnosis method combined with artificial intelligence is implemented.

[0121] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A safety diagnosis method for charging piles that incorporates artificial intelligence, characterized in that, The method includes: The original current traveling wave signal sequence and the original voltage traveling wave signal sequence generated at the moment of the fault on the power transmission line of the charging pile are obtained. The original current traveling wave signal sequence includes the transient current change trajectory at the output end of the power module inside the charging pile, and the original voltage traveling wave signal sequence includes the transient voltage change trajectory at the input end of the power module inside the charging pile. The original current traveling wave signal sequence is decomposed into a set of incident current traveling wave segments and a set of reflected current traveling wave segments, and the original voltage traveling wave signal sequence is decomposed into a set of incident voltage traveling wave segments and a set of transmitted voltage traveling wave segments; The pre-trained fault evolution model is invoked to perform fault boundary inversion operation on the set of reflected current traveling wave segments and the set of transmitted voltage traveling wave segments, generating boundary reflection coefficient distribution map and boundary transmission coefficient distribution map of the fault point inside the charging pile; Perform a joint fault feature decoding operation on the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map to obtain the fault impedance feature code and fault location feature code of the fault point inside the charging pile. A safety response strategy mapping operation is performed on the fault impedance feature code and the fault location feature code to generate a safety protection command sequence containing a power module shutdown command and a fault area isolation command, and the safety protection command sequence is sent to the power controller and line switch of the charging pile.

2. The artificial intelligence-based charging pile safety diagnosis method according to claim 1, characterized in that, The step of decomposing the original current traveling wave signal sequence into a set of incident current traveling wave segments and a set of reflected current traveling wave segments, and decomposing the original voltage traveling wave signal sequence into a set of incident voltage traveling wave segments and a set of transmitted voltage traveling wave segments, includes: The rising edge trigger time of the current traveling wave waveform is extracted from the original current traveling wave signal sequence as the current traveling wave start reference time. The rising edge trigger time of the voltage traveling wave waveform is extracted from the original voltage traveling wave signal sequence as the voltage traveling wave start reference time. The time difference between the current traveling wave start reference time and the voltage traveling wave start reference time is used as the traveling wave propagation direction discrimination parameter. When the traveling wave propagation direction discrimination parameter indicates that the current traveling wave arrives before the voltage traveling wave, the waveform segment in the original current traveling wave signal sequence within the first time window starting from the current traveling wave initiation reference time is taken as the incident current traveling wave segment set, and the waveform segment with opposite polarity that appears after the incident current traveling wave segment set in the original current traveling wave signal sequence is taken as the reflected current traveling wave segment set. When the traveling wave propagation direction discrimination parameter indicates that the current traveling wave arrives before the voltage traveling wave, the waveform segment within the second time window starting from the voltage traveling wave in the original voltage traveling wave signal sequence is taken as the incident voltage traveling wave segment set, and the waveform segment with waveform broadening that appears after the incident voltage traveling wave segment set in the original voltage traveling wave signal sequence is taken as the transmission voltage traveling wave segment set. When the traveling wave propagation direction discrimination parameter indicates that the voltage traveling wave arrives before the current traveling wave, the waveform segments in the original voltage traveling wave signal sequence within the first time window starting from the voltage traveling wave initiation reference time are taken as the incident voltage traveling wave segment set, the waveform segments with opposite polarity that appear after the incident voltage traveling wave segment set in the original voltage traveling wave signal sequence are taken as the reflected voltage traveling wave segment set, and the waveform segments with waveform broadening that appear after the current traveling wave in the original current traveling wave signal sequence are taken as the transmitted current traveling wave segment set. The marked sets of incident current traveling wave segments, reflected current traveling wave segments, incident voltage traveling wave segments, and transmitted voltage traveling wave segments are arranged and aligned in chronological order to generate a time-synchronized traveling wave segment combination sequence. Each time position in the traveling wave segment combination sequence contains at least one of the following: incident current traveling wave segment, reflected current traveling wave segment, incident voltage traveling wave segment, and transmitted voltage traveling wave segment.

3. The artificial intelligence-based charging pile safety diagnosis method according to claim 2, characterized in that, The pre-trained fault evolution model is invoked to perform fault boundary inversion operations on the set of reflected current traveling wave segments and the set of transmitted voltage traveling wave segments, generating boundary reflection coefficient distribution maps and boundary transmission coefficient distribution maps of fault points inside the charging pile, including: Each reflected current traveling wave segment in the set of reflected current traveling wave segments is paired with the time-corresponding incident current traveling wave segment in the set of incident current traveling wave segments to obtain multiple pairs of reflected incident current traveling wave segments. For each pair of reflected incident current traveling wave segments, a frequency domain transformation operation is performed to convert the pair of reflected incident current traveling wave segments from the time domain representation to the frequency domain representation, resulting in the frequency domain spectrum of the reflected current and the frequency domain spectrum of the incident current. The ratio of the amplitude of the reflected current frequency domain spectrum to the amplitude of the incident current frequency domain spectrum in each frequency component is taken as the reflection coefficient value corresponding to the frequency component. The reflection coefficient values ​​of all frequency components are arranged in frequency order to generate a boundary reflection coefficient distribution map. Each transmitted voltage traveling wave segment in the transmitted voltage traveling wave segment set is paired with the time-corresponding incident voltage traveling wave segment in the incident voltage traveling wave segment set to obtain multiple pairs of transmitted incident voltage traveling wave segments. For each transmitted incident voltage traveling wave segment, a frequency domain transformation operation is performed to obtain the transmitted voltage frequency domain spectrum and the incident voltage frequency domain spectrum. The ratio of the amplitude of the transmitted voltage frequency domain spectrum to the amplitude of the incident voltage frequency domain spectrum at each frequency component is taken as the transmission coefficient value corresponding to the frequency component. The transmission coefficient values ​​of all frequency components are arranged in frequency order to generate a boundary transmission coefficient distribution map. The boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map are input into the feature association layer of the fault evolution model. The feature association layer performs cross-validation on the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map, identifies the frequency feature peaks that match each other in the reflection coefficient distribution map and the transmission coefficient distribution map, and generates boundary reflection coefficient distribution map and boundary transmission coefficient distribution map that have passed consistency verification.

4. The artificial intelligence-based charging pile safety diagnosis method according to claim 3, characterized in that, The step of performing a joint fault feature decoding operation on the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map to obtain the fault impedance feature code and fault location feature code of the fault point inside the charging pile includes: The boundary reflection coefficient distribution map is input into the first decoding branch of the fault evolution model. The first decoding branch includes a reflection coefficient feature extraction network and an impedance transformation network. The reflection coefficient feature extraction network performs a multi-scale feature convolution operation on the boundary reflection coefficient distribution map to generate a reflection coefficient feature vector. The impedance transformation network performs a nonlinear mapping operation on the reflection coefficient feature vector to obtain the fault impedance feature code. The boundary transmission coefficient distribution map is input into the second decoding branch of the fault evolution model. The second decoding branch includes a transmission coefficient feature extraction network and a distance inversion network. The transmission coefficient feature extraction network performs a time-series feature extraction operation on the boundary transmission coefficient distribution map to generate a transmission coefficient feature vector. The distance inversion network performs a regression prediction operation on the transmission coefficient feature vector to obtain the fault location feature code. The fault impedance feature code and the fault location feature code are input into the joint optimization layer of the fault evolution model. The joint optimization layer performs a consistency adjustment operation based on the physical constraint relationship between the fault impedance feature code and the fault location feature code. The physical constraint relationship is that the sum of the squares of the fault point reflection coefficient and the fault point transmission coefficient is equal to a fixed constant. The consistency adjustment operation generates the fault impedance feature code and the fault location feature code that have been corrected by the physical constraints. The fault impedance feature code, which has been corrected by physical constraints, is input into the impedance classification mapper, which maps the fault impedance feature code to a preset impedance category space and outputs a fault impedance category identifier. The fault location feature code, which has been corrected by physical constraints, is input into the distance regression mapper, which maps the fault location feature code to a preset distance value space and outputs the fault distance value code.

5. The artificial intelligence-based charging pile safety diagnosis method according to claim 4, characterized in that, The step of performing a safety response strategy mapping operation on the fault impedance feature code and the fault location feature code to generate a safety protection command sequence containing a power module shutdown command and a fault area isolation command, and sending the safety protection command sequence to the power controller and line switch of the charging pile, includes: The fault impedance category identifier is input into the first index port of the security policy mapping table. The security policy mapping table stores the correspondence between impedance category identifiers and first response action sets. The corresponding first response action set is retrieved from the security policy mapping table according to the fault impedance category identifier. The first response action set includes the sending target address and sending time parameters of the power module shutdown command. The fault distance numerical code is input into the second index port of the security policy mapping table. The security policy mapping table stores the correspondence between distance numerical ranges and second response action sets. The corresponding second response action set is retrieved from the security policy mapping table according to the distance numerical range to which the fault distance numerical code belongs. The second response action set includes the isolation boundary identifier and isolation order parameters of the fault area isolation instruction. The sending target address and sending time parameter of the power module shutdown instruction in the first response action set are combined with the isolation boundary identifier and isolation sequence parameter of the fault area isolation instruction in the second response action set to generate a security protection command sequence. The power module shutdown instruction and the fault area isolation instruction in the security protection command sequence are arranged according to the timing relationship determined by the sending time parameter and the isolation sequence parameter. The power module shutdown command in the safety protection command sequence is sent to the command receiving port of the power controller of the charging pile according to the delay time specified by the sending time parameter; The fault area isolation commands in the safety protection command sequence are sent to the command receiving port of the charging pile's line switch in the order specified by the isolation sequence parameter.

6. The artificial intelligence-based charging pile safety diagnosis method according to claim 1, characterized in that, The acquisition of the original current traveling wave signal sequence and the original voltage traveling wave signal sequence generated at the moment of fault occurrence on the power transmission line of the charging pile, wherein the original current traveling wave signal sequence includes the transient current change trajectory at the output terminal of the power module inside the charging pile, and the original voltage traveling wave signal sequence includes the transient voltage change trajectory at the input terminal of the power module inside the charging pile, including: A current traveling wave acquisition unit is installed at the connection node between the output end of the charging pile power module and the transmission cable. The current traveling wave acquisition unit converts the current change on the power transmission line into a voltage signal through the principle of electromagnetic induction to obtain the original current traveling wave electrical signal. A voltage traveling wave acquisition unit is installed at the connection node between the input end of the charging pile power module and the grid access point. The voltage traveling wave acquisition unit converts the voltage change on the power transmission line into a low-voltage signal through the principle of capacitive voltage division, thereby obtaining the original voltage traveling wave electrical signal. The original current traveling wave signal is input to a first high-speed analog-to-digital converter. The first high-speed analog-to-digital converter discretizes the original current traveling wave signal at a fixed sampling frequency and outputs an original current traveling wave signal sequence. Each discrete point in the original current traveling wave signal sequence carries a sampling time identifier and a current amplitude quantization value. The original voltage traveling wave electrical signal is input to a second high-speed analog-to-digital converter. The second high-speed analog-to-digital converter discretizes the original voltage traveling wave electrical signal at the same fixed sampling frequency as the first high-speed analog-to-digital converter and outputs an original voltage traveling wave signal sequence. Each discrete point in the original voltage traveling wave signal sequence carries a sampling time identifier and a voltage amplitude quantization value. The original current traveling wave signal sequence and the original voltage traveling wave signal sequence are stored in a circular storage buffer. After the circular storage buffer is full, the earliest stored traveling wave signal sequence is overwritten with a new traveling wave signal sequence.

7. The artificial intelligence-based charging pile safety diagnosis method according to claim 2, characterized in that, The step involves aligning and scheduling the marked sets of incident current traveling wave segments, reflected current traveling wave segments, incident voltage traveling wave segments, and transmitted voltage traveling wave segments in chronological order to generate a time-synchronized sequence of traveling wave segments. Each time position in this sequence contains at least one of the following: incident current traveling wave segment, reflected current traveling wave segment, incident voltage traveling wave segment, and transmitted voltage traveling wave segment. Obtain the time start mark of each incident current traveling wave segment in the incident current traveling wave segment set, obtain the time start mark of each reflected current traveling wave segment in the reflected current traveling wave segment set, and pair the incident current traveling wave segments and reflected current traveling wave segments with a time start mark difference less than a preset time threshold as current traveling wave segment groups corresponding to the same fault event. Obtain the time start mark of each incident voltage traveling wave segment in the incident voltage traveling wave segment set, obtain the time start mark of each transmitted voltage traveling wave segment in the transmitted voltage traveling wave segment set, and pair the incident voltage traveling wave segments and transmitted voltage traveling wave segments with a time start mark difference less than a preset time threshold as voltage traveling wave segment groups corresponding to the same fault event. The current traveling wave segment group is associated and paired with the group whose time start mark is closest to that of the voltage traveling wave segment group to form a fault event traveling wave segment group. The fault event traveling wave segment group includes incident current traveling wave segment, reflected current traveling wave segment, incident voltage traveling wave segment and transmitted voltage traveling wave segment. According to the order of the time start markers of each fault event traveling wave segment group, all fault event traveling wave segment groups are arranged into a time-synchronized traveling wave segment combination sequence, and each element in the traveling wave segment combination sequence corresponds to a fault event traveling wave segment group. For a fault event traveling wave segment group that is missing a certain type of traveling wave segment in the time-synchronized traveling wave segment combination sequence, the missing position is filled with an all-zero waveform segment.

8. The artificial intelligence-based charging pile safety diagnosis method according to claim 3, characterized in that, The step of inputting the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map into the feature association layer of the fault evolution model, wherein the feature association layer performs cross-validation on the boundary reflection coefficient distribution map and the boundary transmission coefficient distribution map, identifies mutually matching frequency feature peaks in the reflection coefficient distribution map and the transmission coefficient distribution map, and generates boundary reflection coefficient distribution maps and boundary transmission coefficient distribution maps that have passed consistency verification, includes: The boundary reflection coefficient distribution map is represented as a reflection coefficient sequence composed of frequency index and reflection coefficient value, and the boundary transmission coefficient distribution map is represented as a transmission coefficient sequence composed of frequency index and transmission coefficient value, wherein the reflection coefficient sequence and the transmission coefficient sequence have the same frequency index range; For each frequency index position in the reflection coefficient sequence, the reflection coefficient value corresponding to the frequency index position is obtained. At the same time, the transmission coefficient value corresponding to the same frequency index position in the transmission coefficient sequence is obtained. The sum of the squares of the reflection coefficient value and the squares of the transmission coefficient value is compared with a preset physical constant. When the absolute value of the difference exceeds a preset error threshold, the frequency index position is identified as an inconsistent frequency point. For frequency index positions identified as inconsistent frequency points, if the reflection coefficient value is greater than the transmission coefficient value, the reflection coefficient value remains unchanged and the transmission coefficient value is re-determined based on physical constants; if the transmission coefficient value is greater than the reflection coefficient value, the transmission coefficient value remains unchanged and the reflection coefficient value is re-determined based on physical constants. The reflection coefficient values ​​of all frequency index positions after coefficient adjustment are rearranged according to the frequency index order to generate a boundary reflection coefficient distribution map that has passed the consistency check. The transmission coefficient values ​​of all frequency index positions after coefficient adjustment are rearranged according to the frequency index order to generate a boundary transmission coefficient distribution map that has passed the consistency check.

9. The artificial intelligence-based charging pile safety diagnosis method according to claim 4, characterized in that, The fault impedance feature encoding, corrected for physical constraints, is input into the impedance classification mapper. The impedance classification mapper maps the fault impedance feature encoding to a preset impedance category space and outputs a fault impedance category identifier, including: The fault impedance feature encoding, which has been corrected by physical constraints, is input to the input feature receiving port of the impedance classification mapper. The impedance classification mapper includes a multi-layer fully connected feature extraction network and a classification output layer. The multi-layer fully connected feature extraction network performs a layer-by-layer feature abstraction operation on the fault impedance feature encoding to generate a high-dimensional feature representation vector. The classification output layer receives the high-dimensional feature representation vector and performs a linear transformation operation on the high-dimensional feature representation vector to generate a first category score and a second category score. The first category score and the second category score are input into the normalized exponential function calculation unit, which transforms the first category score into a first normalized probability value and the second category score into a second normalized probability value. The first normalized probability value is compared with the second normalized probability value. When the first normalized probability value is greater than the second normalized probability value, a low impedance fault category identifier is output. When the second normalized probability value is greater than the first normalized probability value, a high impedance fault category identifier is output. The output fault impedance category identifier is passed to the safety response strategy mapping operation step.

10. A charging pile safety diagnostic system incorporating artificial intelligence, characterized in that, The artificial intelligence-integrated charging pile safety diagnosis system includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the artificial intelligence-integrated charging pile safety diagnosis method according to any one of claims 1-9.

Images