Power distribution unit monitoring method and system

Abstract

This invention provides a power distribution unit monitoring method and system, relating to the field of power distribution unit monitoring technology. This invention triggers a monitoring window by exceeding a threshold current change rate, enabling refined analysis of suspected abnormal periods. This avoids data redundancy and computational burden caused by continuous data acquisition, while simultaneously capturing waveforms before and after the anomaly, laying the foundation for accurate diagnosis. By identifying multiple characteristic fluctuation moments within the monitoring window and determining whether other distribution branches experience synchronous fluctuations, it distinguishes whether the current anomaly originates from a branch fault or bus fluctuations, solving the problem of untraceability and preventing batch false alarms caused by bus fluctuations, ensuring business continuity. It converts time-domain waveforms into easily quantifiable and comparable frequency-domain information, combines it with a fault template library for feature matching, and ultimately determines the fault type, reducing reliance on the professional experience of maintenance personnel and improving fault diagnosis efficiency.

Description

Technical Field

[0001] This invention relates to the field of power distribution unit monitoring technology, specifically a power distribution unit monitoring method and system. Background Technology

[0002] Power distribution units are critical end-point power distribution devices in data centers, communication base stations, and industrial control sites. Their main function is to safely and reliably distribute front-end power to multiple servers, network devices, or industrial controllers within the rack. With the rapid development of cloud computing, big data, and artificial intelligence technologies, data center scale continues to expand, and rack power density continues to rise. Intelligent power distribution units have become an important component in ensuring the stable operation of critical infrastructure and are widely used in various power supply and distribution environments that require precise power distribution monitoring.

[0003] In existing technologies, threshold comparison algorithms are used for anomaly detection. This involves setting a current threshold for each branch, collecting current and voltage data in real time, and comparing it with a preset safety threshold. When the detected value exceeds the preset safety threshold, an alarm signal is triggered. This is currently the most basic and widely used existing technology. Some existing technologies employ fault waveform recording and offline analysis methods. This method collects data from each branch in real time, and when an anomaly is triggered, it stores the waveform data before and after the fault. Maintenance personnel can retrieve these stored waveforms through host computer software for manual analysis to determine the cause of the fault. Other existing technologies use rate-of-change detection methods. This method calculates the rate of change of current or voltage in real time using a differential algorithm. When the rate of change exceeds a set value, it is determined to be a load change or transient event, and the relevant data is recorded for subsequent analysis.

[0004] However, the aforementioned existing technologies have the following shortcomings: While existing threshold comparison algorithms and rate of change detection methods can detect anomalies, they can only detect the occurrence of anomalies and lack the ability to make logical judgments about the source of fluctuations. They cannot distinguish whether current fluctuations are caused by faults in the load equipment of this branch or by voltage disturbances on the upstream bus.

[0005] Existing fault recording and offline analysis methods are essentially passive data storage methods. Although they record the original waveforms, they lack the ability to extract features and recognize patterns from the waveform data. The system can only provide the original data and cannot output diagnostic results of the fault type. Fault identification relies heavily on the personal experience of maintenance personnel.

[0006] The existing data acquisition mechanism is relatively rigid. The continuous sampling mode causes data redundancy, while the simple threshold trigger mode is prone to missing early fault symptoms that have a large rate of change but whose amplitude has not exceeded the limit, and it is difficult to adapt to complex and ever-changing load conditions.

[0007] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0008] The purpose of this invention is to provide a power distribution unit monitoring method and system to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A method for monitoring a power distribution unit, comprising the following steps: The current change rate of all distribution branches of the power distribution unit is acquired in real time. When the current change rate of a distribution branch exceeds the set first threshold, the distribution branch is marked as a monitoring branch. The monitoring window is set with the current time as the center, and the current curves of all distribution branches are acquired within the monitoring window. Based on the current curve of the monitored branch within the monitoring window, identify all moments when the rate of change of current exceeds the set second threshold. If the second threshold is less than the first threshold, the identified moments are taken as characteristic fluctuation moments. For other distribution branches besides the monitoring branch, determine whether there is synchronous fluctuation in the time neighborhood of the current curve of each distribution branch at each moment of the characteristic fluctuation moment. If the number of other distribution branches with synchronous fluctuation reaches or exceeds a preset proportion, it is determined that the current fluctuation originates from the bus fluctuation; otherwise, it is determined that it originates from the monitoring branch fluctuation. If the fluctuation is determined to be a monitoring branch, the current curve of the monitoring branch within the monitoring window is subjected to Fourier transform to obtain the corresponding spectrum information. The frequency with amplitude exceeding the third threshold in the spectrum information is identified, and the identified frequency is used as the characteristic frequency and the corresponding amplitude is used as the characteristic amplitude. After obtaining the current curve of a known fault type and performing a Fourier transform, the frequency with an amplitude exceeding the fourth threshold in the spectrum information is identified as the standard frequency, and the corresponding amplitude is identified as the standard amplitude, thus forming a fault template library. The standard frequency and standard amplitude of each fault type in the fault template library are matched with the characteristic frequency and characteristic amplitude of the monitoring branch, and the fault type of the monitoring branch is determined based on the matching result.

[0010] Furthermore, the logic for setting the monitoring window includes: taking the moment when the rate of change of current in a distribution branch exceeds a first threshold as the center, extending forward and backward by a preset fixed duration to determine the start and end times of the monitoring window.

[0011] Furthermore, the time neighborhood refers to setting a time offset threshold, taking each moment in the characteristic fluctuation moments as the center, and considering the intervals before and after that moment as less than or equal to the time offset threshold as the time neighborhood corresponding to that moment.

[0012] Furthermore, it is determined whether there are synchronous fluctuations in the time neighborhood of the current curve of each distribution branch at each time point during the characteristic fluctuation moment, specifically including: For other distribution branches besides the monitoring branch, a fluctuation threshold is set. From the real-time current change rate, the maximum current change rate of each distribution branch in the time neighborhood of each moment in the characteristic fluctuation moment is extracted, and the maximum current change rate is compared with the fluctuation threshold. If the maximum current change rate exceeds the fluctuation threshold, it is determined that the distribution branch has current fluctuation at the characteristic fluctuation moment. Repeat the above process to count the number of characteristic fluctuation moments with current fluctuation in each distribution branch. Use this count as the number of synchronization moments. Set a synchronization threshold and compare the number of synchronization moments with the synchronization threshold. If the number of synchronization moments reaches or exceeds the synchronization threshold, it is determined that the distribution branch has synchronization fluctuations; otherwise, it is determined that the distribution branch does not have synchronization fluctuations.

[0013] Furthermore, the spectrum information is frequency domain data obtained by performing a Fourier transform on the current curve of the monitoring branch within the monitoring window, and the frequency domain data includes several frequencies and the amplitude corresponding to each frequency.

[0014] Furthermore, the standard frequency and standard amplitude of each fault type in the fault template library are matched with the characteristic frequency and characteristic amplitude of the monitoring branch, specifically including: Set a frequency offset threshold. Take each of the characteristic frequencies as the center and define the intervals before and after that frequency that are less than or equal to the frequency offset threshold as the frequency neighborhood of that frequency. For each fault type in the fault template library, obtain the standard frequency and standard amplitude of the fault type. For each standard frequency, determine whether there is a characteristic frequency in the characteristic frequency of the monitoring branch such that the standard frequency is in the frequency neighborhood of the characteristic frequency. The standard frequency that meets the condition is taken as the candidate frequency of the fault type.

[0015] Furthermore, it also includes: Amplitude tolerance range is set. Based on the candidate frequencies, it is further determined whether the difference between the standard amplitude corresponding to each candidate frequency and the characteristic amplitude of the characteristic frequency corresponding to the candidate frequency is within the amplitude tolerance range. The candidate frequencies that meet the above conditions are used as the matching frequencies for the fault type. The number of matching frequencies is counted as the number of matches for the fault type. Calculate the matching degree for each fault type. The matching degree is equal to the number of matches for that fault type divided by the total number of standard frequencies included in that fault type. Set a matching degree threshold and select fault types with a matching degree that reaches or exceeds the matching degree threshold as candidate fault types.

[0016] Furthermore, it also includes: If there are no candidate fault types that meet the above conditions, the fault type with the highest matching degree among the matching degrees that have not reached the threshold will be selected as the fault type of the monitoring branch, and a prompt will be added. If only one fault type is included among the candidate fault types, then that fault type is selected as the fault type of the monitoring branch. If the candidate fault types include multiple fault types, the fault type with the highest matching degree is selected as the fault type of the monitoring branch; if multiple fault types have the same highest matching degree, the fault type with the smallest cumulative deviation between the standard amplitude of the matching frequency and the characteristic amplitude of the corresponding monitoring branch characteristic frequency is selected as the fault type of the monitoring branch.

[0017] Furthermore, the fault type with the smallest cumulative deviation between the standard amplitude of the matching frequency and the characteristic amplitude of the corresponding monitoring branch characteristic frequency is selected, specifically including: The fault types with the highest matching degree among the candidate fault types are selected as parallel candidate fault types. The absolute value of the difference between the standard amplitude of each matching frequency and the characteristic amplitude of the corresponding monitoring branch characteristic frequency in each parallel candidate fault type is calculated as the amplitude deviation of the parallel candidate fault type. The amplitude deviations are summed to obtain the cumulative deviation of the parallel candidate fault type. The cumulative deviations of all parallel candidate fault types are compared, and the fault type with the smallest cumulative deviation is selected as the fault type of the monitoring branch.

[0018] The present invention also provides a power distribution unit monitoring system, the system being used to implement the above-described power distribution unit monitoring method, comprising: Real-time monitoring module: Used to obtain the current change rate of all distribution branches of the power distribution unit in real time. When the current change rate of a distribution branch exceeds the set first threshold, the distribution branch is marked as a monitoring branch. The monitoring window is set with the current time as the center, and the current curves of all distribution branches are obtained within the monitoring window. Feature extraction module: used to identify all times when the rate of change of current exceeds a set second threshold based on the current curve of the monitored branch within the monitoring window. The second threshold is less than the first threshold, and the identified times are used as feature fluctuation times. Fluctuation identification module: For other distribution branches besides the monitoring branch, it is used to determine whether there is synchronous fluctuation in the time neighborhood of the current curve of each distribution branch at each moment of the characteristic fluctuation moment. If the number of other distribution branches with synchronous fluctuation reaches or exceeds a preset ratio, it is determined that the current fluctuation originates from the bus fluctuation; otherwise, it is determined that it originates from the monitoring branch fluctuation. Spectrum analysis module: If the fluctuation of the monitored branch is determined, the current curve of the monitored branch within the monitoring window is subjected to Fourier transform to obtain the corresponding spectrum information, the frequency with amplitude exceeding the third threshold in the spectrum information is identified, the identified frequency is used as the characteristic frequency, and the corresponding amplitude is used as the characteristic amplitude. Fault matching module: It is used to acquire the current curve of known fault types, perform Fourier transform on it, identify the frequency with amplitude exceeding the fourth threshold in the spectrum information as the standard frequency, and the corresponding amplitude as the standard amplitude to form a fault template library. It matches the standard frequency and standard amplitude of each fault type in the fault template library with the characteristic frequency and characteristic amplitude of the monitoring branch, and determines the fault type of the monitoring branch based on the matching result.

[0019] Compared with the prior art, the beneficial effects of the present invention are: This invention uses the current change rate exceeding a first threshold as a trigger condition and sets a monitoring window to achieve refined analysis only for the time period of suspected anomalies. This avoids data redundancy and computational burden caused by continuous data acquisition, while ensuring complete capture of waveform data before and after the anomaly occurs, laying the foundation for accurate diagnosis.

[0020] This invention also identifies multiple characteristic fluctuation moments within the monitoring window and determines whether other distribution branches experience synchronous fluctuations. This enables the differentiation between current anomalies originating from monitoring branch faults and bus fluctuations. When multiple distribution branches fluctuate synchronously, it is determined to be bus fluctuation; when only the monitoring branch fluctuates, it is determined to be a monitoring branch fault. This fundamentally solves the problem of existing technologies being unable to trace the source, avoids batch false alarms caused by bus fluctuations, and ensures business continuity.

[0021] This invention transforms complex time-domain waveforms into frequency-domain information that is easy to quantify and compare, providing a basis for fault classification. It also constructs a fault template library of known fault types and matches the characteristic frequencies and characteristic amplitudes of the monitored branches with the fault template library to ultimately determine the fault type. This significantly reduces the reliance on the professional experience of maintenance personnel and improves the efficiency of fault diagnosis. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a structural block diagram of the overall system of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0024] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0025] Example: Please see Figure 1 The present invention provides a technical solution: A method for monitoring a power distribution unit, comprising the following steps: Step 1: Real-time acquisition of the current change rate of all distribution branches of the power distribution unit. When the current change rate of a distribution branch exceeds the set first threshold, the distribution branch is marked as a monitoring branch. The monitoring window is set with the current time as the center, and the current curves of all distribution branches are acquired within the monitoring window. The reason for using the current change rate exceeding the first threshold as the trigger condition, instead of the traditional current threshold, is that the current change rate can more sensitively reflect sudden changes in the load state. The current exceeding the threshold is often the result of a fault already formed, while the current change rate exceeding the threshold can capture the process of the fault occurring at the moment, enabling earlier detection of abnormalities. At the same time, using the current change rate as the trigger quantity avoids frequent false triggers caused by slow load changes.

[0026] The system only initiates refined data acquisition within the monitoring window when the current change rate exceeds the threshold. At other times, it only needs to perform low-overhead change rate calculation and threshold comparison, thus avoiding resource consumption caused by continuous data acquisition and large-scale data storage.

[0027] Before the system of this invention is put into real-time monitoring, the current curves of each distribution branch of the power distribution unit are first obtained during the actual operation. Based on the experience of the operation and maintenance personnel, the current curve data under normal operation is selected. In order to ensure statistical representativeness, the collection time of normal operation data should cover at least one complete working cycle, such as 24 consecutive hours.

[0028] For each distribution branch, extract the instantaneous current value from its normal operating data, calculate the current change rate point by point, and for each distribution branch, find the maximum value of its current change rate under normal operating conditions, and then apply the formula... ,in For the first The first threshold for each distribution branch is problematic in practical applications. Different distribution branches within the same power distribution unit may be connected to different types of loads, and the current change rate of different loads varies greatly under normal operating conditions. If a uniform threshold is applied to all branches, it will inevitably lead to frequent false triggers for branches with large fluctuations and missed early anomalies for branches with small fluctuations. For the first The maximum rate of change of current under normal operating conditions of a branch circuit. The safety factor is determined based on the load characteristics of the assigned branch.

[0029] The maximum current change rate under normal operating conditions is the maximum value statistically obtained within a specific time period. It reflects the upper limit of fluctuation that may be reached under normal conditions. However, in actual operation, various factors may cause the current change rate to occasionally exceed this historical maximum value, and these situations are still normal. If we directly... As a trigger threshold, any fluctuation exceeding the historical maximum value will trigger the monitoring window, leading to frequent false triggers and rendering anomaly detection meaningless. Multiplied by a safety factor, this is equivalent to... A buffer zone is set up above the system. The width of this buffer zone is determined by a safety factor, which ensures that the system only responds to drastic fluctuations that significantly exceed the normal level. This avoids frequent false triggers caused by occasional overshooting of normal fluctuations and improves the reliability of anomaly detection.

[0030] In this embodiment, taking the distribution branch that connects to the load of a regular server as an example, we take... A value of 3.5 strikes a good balance between sensitivity and false alarm resistance, effectively capturing genuine anomalies while avoiding false triggers caused by normal fluctuations. The specific value can be determined based on the load characteristics of the assigned branch and the site requirements. Within the range of adjustment, for branches with small load fluctuations and stable operating conditions, such as critical equipment that is sensitive to abnormalities, a smaller value can be taken to make the threshold closer to the upper limit of normal fluctuations, resulting in higher trigger sensitivity, which is suitable for scenarios that require early warning; for branches with large load fluctuations, such as equipment that starts frequently, a larger value can be taken to reserve sufficient margin for drastic fluctuations under normal operating conditions, effectively avoiding frequent false triggers.

[0031] The logic for setting the monitoring window includes: taking the moment when the rate of change of current in a distribution branch exceeds a first threshold as the center, extending forward and backward by a preset fixed duration to determine the start and end times of the monitoring window.

[0032] The monitoring window is set because a fault is a continuous process rather than an isolated point in time. The monitoring window ensures that the waveform of the entire process before, during, and after the anomaly is fully captured, providing a data foundation for subsequent steps. Setting a window of limited length only after the event is triggered avoids continuously storing massive amounts of raw data.

[0033] When setting the monitoring window, extending it forward by 0.2 seconds ensures that important early warning information is not lost due to a window that is too short, nor introduces too much useless redundant data due to a window that is too long. Extending it backward by 1.0 second covers the complete evolution of most faults. For faults with longer durations, although the fault may still be ongoing when the monitoring window ends, the 1.0 second of data captured is sufficient to identify the fault type. The total window length of 1.2 seconds is within a reasonable range for storage and computation, and will not put too much pressure on the system. If the monitoring window is too short, although resource consumption is low, the accuracy of judgment will decrease. If the monitoring window is too long, resources will be wasted and irrelevant noise may be introduced.

[0034] Step 2: Based on the current curve of the monitored branch within the monitoring window, identify all moments when the rate of change of current exceeds the set second threshold. If the second threshold is less than the first threshold, the identified moments are taken as characteristic fluctuation moments. Since the monitoring window contains a large number of moments, using all of these moments without filtering for subsequent synchronous fluctuation analysis would impose a huge computational burden. Moreover, not all moments carry effective information. During periods of stable current, the rate of change is close to zero, and these moments do not contribute to determining whether synchronous fluctuations exist. Only those moments where the current changes significantly truly reflect the occurrence of fluctuations. By setting a second threshold for filtering, the analysis can be focused on characteristic fluctuation moments, achieving precise allocation of computational resources.

[0035] The first threshold is set based on the maximum fluctuation during normal operation. It is used to filter out noteworthy abnormal events and avoid the system frequently starting the monitoring window due to minor daily fluctuations. Once the monitoring window is started, it means that an anomaly has been confirmed. At this time, it is necessary to fully capture all key fluctuation points within the abnormal event, including those secondary fluctuations whose amplitude does not exceed the first threshold, and to conduct refined analysis of the anomaly. Therefore, the second threshold is smaller than the first threshold.

[0036] In this embodiment, the second threshold is set using a proportional method based on the first threshold. This proportional method establishes a clear numerical relationship between the second threshold and the first threshold, and each allocation branch independently sets its corresponding second threshold. Assign branches according to the formula Calculate the second threshold, where For the first The second threshold for the assigned branch, As a proportionality coefficient, this embodiment takes... A value of 0.35 strikes a good balance between capturing secondary fluctuations and the number of control characteristic moments, making it suitable for most common fault diagnosis scenarios. The specific value can be determined according to the actual application scenario. Within the range of adjustment, a smaller value can be taken for scenarios with low tolerance for missed faults and a desire to capture as many secondary fluctuations as possible; a larger value can be taken for scenarios that are sensitive to false alarms and a desire to reduce the number of feature moments to reduce the burden of subsequent analysis.

[0037] Step 3: For other distribution branches besides the monitoring branch, determine whether there is synchronous fluctuation in the time neighborhood of the current curve of each distribution branch at each moment of the characteristic fluctuation time. If the number of other distribution branches with synchronous fluctuation reaches or exceeds the preset ratio, it is determined that the current fluctuation originates from the bus fluctuation; otherwise, it is determined that it originates from the monitoring branch fluctuation. The time neighborhood refers to a time offset threshold that is set, and the intervals before and after each moment in the characteristic fluctuation are less than or equal to the time offset threshold, which are taken as the time neighborhood of that moment.

[0038] Setting a time neighborhood is necessary because there can be a slight time difference when fluctuations propagate from the source to other distribution branches. If the fluctuations of other distribution branches are required to occur at the exact same instant as the characteristic fluctuation moment to be considered synchronized, then a large number of truly synchronized fluctuations will be missed due to these slight time differences, leading to serious inaccuracies in locating the fluctuation source. By setting a time neighborhood, the judgment of synchronization is relaxed from the same instant to the same time period. As long as the fluctuations of other distribution branches occur within the time neighborhood of the characteristic moment, the accuracy of synchronization judgment is improved.

[0039] In this embodiment, the time offset threshold is set to 0.5 milliseconds. 0.5 milliseconds is sufficient to cover the propagation delay and measurement difference in most cases, ensuring that true synchronization fluctuations are not missed. The total width of 1.0 milliseconds is very narrow relative to the duration of the fault process, and will not mistakenly associate irrelevant fluctuations that are far apart in time, thus ensuring the reliability of fluctuation source location. In practical applications, this value can be adjusted within a reasonable range according to the specific situation on site. For scenarios with long lines in large data centers, it can be appropriately increased, and for special scenarios with extremely high real-time requirements and short lines, it can be appropriately decreased.

[0040] Determine whether there are synchronous fluctuations in the time neighborhood of the current curve of each distribution branch at each time point within the characteristic fluctuation period, specifically including: For other distribution branches besides the monitoring branch, a fluctuation threshold is set. From the real-time current change rate, the maximum current change rate of each distribution branch in the time neighborhood of each moment in the characteristic fluctuation moment is extracted, and the maximum current change rate is compared with the fluctuation threshold. If the maximum current change rate exceeds the fluctuation threshold, it is determined that the distribution branch has current fluctuation at the characteristic fluctuation moment. Setting a fluctuation threshold provides a quantitative and objective standard for judging whether fluctuations exist. In this embodiment, the fluctuation threshold is set using a statistical method based on the normal operating fluctuation level of the distribution branch. Each distribution branch has its own independently set fluctuation threshold because different distribution branches connect to different load types, resulting in differences in their normal operating fluctuation levels. The fluctuation threshold is set according to the formula... ,in For the first The fluctuation threshold of each assigned branch, No. The mean absolute value of the rate of change of current in each distribution branch under normal operating conditions. No. The variance of the absolute value of the rate of change of current in a branch under normal operating conditions is used to calculate the mean and standard deviation. This is because the absolute value directly measures the intensity of fluctuations, while the original rate of change contains irrelevant directional information. The mean of the absolute value can truly reflect the average level of fluctuations, while the original mean may be distorted due to the cancellation of positive and negative values.

[0041] In this embodiment, the multiplier is used as a factor. A value of 4 ensures that only fluctuations significantly exceeding the normal range are counted as valid fluctuations, while not omitting the vast majority of true fluctuations, achieving a good balance between sensitivity and noise resistance. The specific values ​​can be determined based on the on-site requirements for sensitivity and false alarm rate. Within the range of adjustment, a smaller value can be selected for scenarios with low tolerance for missed fault reports and a desire to capture as many abnormal signs as possible, while a larger value can be selected for scenarios that are sensitive to false alarms and a desire to minimize invalid alarms.

[0042] Repeat the above process to count the number of characteristic fluctuation moments with current fluctuation in each distribution branch. Use this count as the number of synchronization moments. Set a synchronization threshold and compare the number of synchronization moments with the synchronization threshold. If the number of synchronization moments reaches or exceeds the synchronization threshold, it is determined that the distribution branch has synchronization fluctuations; otherwise, it is determined that the distribution branch does not have synchronization fluctuations.

[0043] In this embodiment, the synchronization threshold is determined proportionally based on the total number of characteristic fluctuation moments. Let the total number of characteristic fluctuation moments be... Then the synchronization threshold is ,in This is the proportionality coefficient, and its value range is... This means that the distribution branch is considered to have synchronous fluctuations only if it fluctuates at least half of the characteristic fluctuation times. If a distribution branch fluctuates only at a few characteristic fluctuation times, it may just be a coincidence. Only when it fluctuates at most characteristic fluctuation times can it be said that the distribution branch and the monitoring branch have a true synchronous relationship.

[0044] Step 4: If it is determined to be a fluctuation in the monitoring branch, perform a Fourier transform on the current curve of the monitoring branch within the monitoring window to obtain the corresponding spectrum information, identify the frequency in the spectrum information whose amplitude exceeds the third threshold, take the identified frequency as the characteristic frequency, and take the corresponding amplitude as the characteristic amplitude. Because time-domain waveforms are complex and variable, it is difficult to directly distinguish fault types. Fourier transform converts time-domain waveforms into easily quantifiable spectral information, which is more suitable as a basis for fault identification. The spectral information is the frequency domain data obtained by performing Fourier transform on the current curve of the monitoring branch within the monitoring window. The frequency domain data includes several frequencies and the amplitude corresponding to each frequency.

[0045] After obtaining the spectral information by completing the Fourier transform, similar to step 2 which requires filtering characteristic fluctuation moments from a large number of moments, the spectrum contains a large number of frequencies. If all of them are used for subsequent fault template matching, it will bring a huge computational burden. Moreover, the amplitude of most of these frequencies is very low, which does not contribute to fault identification and will instead interfere with the matching results. Therefore, it is necessary to identify characteristic frequencies and filter out the key information that truly reflects the essence of the fault from a large number of frequencies.

[0046] In this embodiment, the third threshold is set using a proportional method based on the maximum amplitude. This method uses the maximum amplitude as a benchmark, ensuring that the filtering is proportional to the signal strength. Regardless of the fault strength, it retains the key frequency components that truly reflect the essence of the fault, exhibiting adaptive capability. Let the maximum amplitude corresponding to all frequencies in the spectrum be... Then the third threshold ,in As a proportionality coefficient, this embodiment takes... A value of 0.1 effectively filters out secondary frequencies while retaining the main fault characteristic frequencies, achieving a good balance between fault diagnosis accuracy and computational efficiency. The specific value can be determined according to the actual application scenario. Within the range of adjustment, a smaller value can be selected for scenarios with high diagnostic accuracy requirements and a desire to retain more detailed features; a larger value can be selected for scenarios with high computational efficiency requirements and a desire to quickly complete fault diagnosis.

[0047] Step 5: Obtain the current curve of the known fault type and perform Fourier transform on it. Identify the frequency in the spectrum information whose amplitude exceeds the fourth threshold as the standard frequency and the corresponding amplitude as the standard amplitude to form a fault template library. Match the standard frequency and standard amplitude of each fault type in the fault template library with the characteristic frequency and characteristic amplitude of the monitoring branch. Determine the fault type of the monitoring branch based on the matching result.

[0048] In this embodiment, the fourth threshold is set using the same method and proportional coefficient as the third threshold to ensure that the feature frequencies extracted from the monitoring branch are on the same order of magnitude as the standard frequencies in the fault template library, making the subsequent matching degree calculation meaningful. The proportional coefficient of the fourth threshold can be adjusted according to the actual application scenario. Adjust within the range, consistent with the adjustment range of the third threshold. For scenarios that need to retain more detailed features, a smaller value can be used, while for scenarios that need to improve filtering efficiency, a larger value can be used.

[0049] The current curves of known fault types are subjected to Fourier transform to ensure they are in the same data format as the characteristic frequencies extracted in step 4, facilitating subsequent matching calculations. Once the fault template library is built, it is not limited to a single fault event or a specific branch. Instead, it is a globally shared and reusable library that can be repeatedly called during system operation for fault type identification after each monitored branch is calibrated. The fault template library is not static after its construction. In actual operation, if a new fault type is encountered and confirmed by maintenance personnel, its spectrum information can be added to the template library to achieve continuous optimization and expansion of the fault template library.

[0050] The standard frequency and standard amplitude of each fault type in the fault template library are matched with the characteristic frequency and characteristic amplitude of the monitoring branch. Specifically, this includes: A frequency offset threshold is set, and for each of the characteristic frequencies, the intervals before and after that frequency that are less than or equal to the frequency offset threshold are defined as the frequency neighborhood of that frequency. The setting of the frequency neighborhood is essentially the same as that of the time neighborhood. By introducing a tolerance range, the strict equality condition is relaxed to fall within the neighborhood, thereby accommodating the unavoidable small deviations in reality and ensuring that true matching relationships are not missed, which significantly improves the accuracy of fault template matching.

[0051] In this embodiment, the frequency offset threshold is set to 1.0Hz. This value effectively accommodates minor frequency deviations that may occur during actual operation, ensuring that true matching relationships are not missed, while maintaining the selectivity of matching and preventing irrelevant frequencies from being mistakenly identified as matches. The specific value of the frequency offset threshold can be determined according to the actual application scenario. Adjust within the Hz range. For scenarios requiring high matching accuracy, a smaller value can be used, while for scenarios where the frequency deviation may be large, a larger value can be used.

[0052] For each fault type in the fault template library, obtain the standard frequency and standard amplitude of the fault type. For each standard frequency, determine whether there is a characteristic frequency in the characteristic frequency of the monitoring branch such that the standard frequency is in the frequency neighborhood of the characteristic frequency. The standard frequency that meets the condition is taken as the candidate frequency of the fault type.

[0053] Amplitude tolerance range is set. Based on the candidate frequencies, it is further determined whether the difference between the standard amplitude corresponding to each candidate frequency and the characteristic amplitude of the characteristic frequency corresponding to the candidate frequency is within the amplitude tolerance range. The candidate frequencies that meet the above conditions are used as the matching frequencies for the fault type. The number of matching frequencies is counted as the number of matches for the fault type. Since frequency matching can only determine whether the standard frequency in the fault template is close to the characteristic frequency of the monitoring branch in terms of location, but cannot verify whether the intensity of the two is consistent, it is considered a true match only when the frequency and amplitude are close. By setting an amplitude tolerance range, a certain deviation is allowed between the characteristic amplitude of the monitoring branch and the standard amplitude of the fault template, rather than requiring them to be strictly equal. This significantly improves the accuracy and robustness of fault identification. The number of matches is the basis for the matching degree calculation. Only frequencies that have passed the amplitude matching verification are included in the number of matches, ensuring the reliability of the matching degree calculation.

[0054] In this embodiment, the amplitude tolerance range is set to 80% to 120% of the standard amplitude, meaning that the characteristic amplitude of the monitored branch is allowed to be between 80% and 120% of the standard amplitude. This range can effectively accommodate normal differences, ensuring that true matching relationships are not missed. At the same time, this range is not too large, which can effectively distinguish different fault types and avoid misjudgment. This value achieves a good balance between inclusiveness and selectivity. The specific value of this amplitude tolerance range can be adjusted within the range of 70% to 130% of the standard amplitude according to the actual application scenario. For scenarios with high matching accuracy requirements and good signal stability, a narrower range can be used, such as 90% to 110% of the standard amplitude. For scenarios with high inclusiveness requirements and large signal differences, a wider range can be used, such as 70% to 130% of the standard amplitude.

[0055] Calculate the matching degree for each fault type. The matching degree is equal to the number of matches for that fault type divided by the total number of standard frequencies included in that fault type. Set a matching degree threshold and select fault types with a matching degree that reaches or exceeds the matching degree threshold as candidate fault types.

[0056] In this embodiment, the matching degree threshold is set to 60%. This value achieves a good balance between the accuracy and inclusiveness of fault identification. A matching degree of 60% means that more than half of the frequencies of the fault type and their corresponding amplitudes match the monitoring branch, which has a high degree of reliability. This threshold will not cause real faults to be missed due to slightly low matching degree because the requirements are too high. It can effectively contain the influence of signal differences in actual operation. The specific value of the matching degree threshold can be adjusted in the range of 50% to 70% according to the actual application scenario. A smaller value can be taken for scenarios with low tolerance for missed detection and the desire to capture as many candidate fault types as possible. A larger value can be taken for scenarios that are sensitive to misjudgment and the desire to improve the accuracy of candidates.

[0057] If there are no candidate fault types that meet the above conditions, the fault type with the highest matching degree among the matching degrees that have not reached the threshold will be selected as the fault type of the monitoring branch, and a prompt will be added: "The matching degree is low, it is recommended to make a judgment based on the on-site situation." If only one fault type is included among the candidate fault types, then that fault type is selected as the fault type of the monitoring branch. If the candidate fault types include multiple fault types, the fault type with the highest matching degree is selected as the fault type of the monitoring branch; if multiple fault types have the same highest matching degree, the fault type with the smallest cumulative deviation between the standard amplitude of the matching frequency and the characteristic amplitude of the corresponding monitoring branch characteristic frequency is selected as the fault type of the monitoring branch.

[0058] The fault type with the smallest cumulative deviation between the standard amplitude of the matching frequency and the characteristic amplitude of the corresponding monitoring branch's characteristic frequency is selected, specifically including: The fault types with the highest matching degree among the candidate fault types are selected as parallel candidate fault types. The absolute value of the difference between the standard amplitude of each matching frequency and the characteristic amplitude of the corresponding monitoring branch characteristic frequency in each parallel candidate fault type is calculated as the amplitude deviation of the parallel candidate fault type. The amplitude deviations are summed to obtain the cumulative deviation of the parallel candidate fault type. The cumulative deviations of all parallel candidate fault types are compared, and the fault type with the smallest cumulative deviation is selected as the fault type of the monitoring branch.

[0059] Since the matching degree only counts the proportion of the number of matched standard frequencies to the total number of standard frequencies for that fault type, when multiple fault types have the same matching degree, the matching degree alone cannot distinguish them. The amplitude deviation reflects the degree of difference between the characteristic amplitude of the monitoring branch and the standard amplitude of the fault template. The smaller the deviation, the closer the matching frequency is to the fault template in terms of intensity. By summing the amplitude deviations of all matching frequencies, the similarity of the fault type to the monitoring branch in terms of overall amplitude distribution can be comprehensively evaluated. The cumulative deviation is the sum of the amplitude deviations of each matching frequency. It comprehensively reflects the closeness of the fault type to the overall amplitude of the monitoring branch across all matching frequencies. The fault type with the smallest cumulative deviation means that it is closest to the monitoring branch in terms of amplitude distribution and is the optimal choice when the matching degree is the same.

[0060] Please see Figure 2 A power distribution unit monitoring system, the system being used to implement the aforementioned power distribution unit monitoring method, comprising: Real-time monitoring module: Used to obtain the current change rate of all distribution branches of the power distribution unit in real time. When the current change rate of a distribution branch exceeds the set first threshold, the distribution branch is marked as a monitoring branch. The monitoring window is set with the current time as the center, and the current curves of all distribution branches are obtained within the monitoring window. Feature extraction module: used to identify all times when the rate of change of current exceeds a set second threshold based on the current curve of the monitored branch within the monitoring window. The second threshold is less than the first threshold, and the identified times are used as feature fluctuation times. Fluctuation identification module: For other distribution branches besides the monitoring branch, it is used to determine whether there is synchronous fluctuation in the time neighborhood of the current curve of each distribution branch at each moment of the characteristic fluctuation moment. If the number of other distribution branches with synchronous fluctuation reaches or exceeds a preset ratio, it is determined that the current fluctuation originates from the bus fluctuation; otherwise, it is determined that it originates from the monitoring branch fluctuation. Spectrum analysis module: If the fluctuation of the monitored branch is determined, the current curve of the monitored branch within the monitoring window is subjected to Fourier transform to obtain the corresponding spectrum information, the frequency with amplitude exceeding the third threshold in the spectrum information is identified, the identified frequency is used as the characteristic frequency, and the corresponding amplitude is used as the characteristic amplitude. Fault matching module: It is used to acquire the current curve of known fault types, perform Fourier transform on it, identify the frequency with amplitude exceeding the fourth threshold in the spectrum information as the standard frequency, and the corresponding amplitude as the standard amplitude to form a fault template library. It matches the standard frequency and standard amplitude of each fault type in the fault template library with the characteristic frequency and characteristic amplitude of the monitoring branch, and determines the fault type of the monitoring branch based on the matching result.

[0061] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0062] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0063] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0064] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for monitoring a power distribution unit, characterized in that, The specific steps include: Step 1: Real-time acquisition of the current change rate of all distribution branches of the power distribution unit. When the current change rate of a distribution branch exceeds the set first threshold, the distribution branch is marked as a monitoring branch. The monitoring window is set with the current time as the center, and the current curves of all distribution branches are acquired within the monitoring window. Step 2: Based on the current curve of the monitored branch within the monitoring window, identify all moments when the rate of change of current exceeds the set second threshold. If the second threshold is less than the first threshold, the identified moments are taken as characteristic fluctuation moments. Step 3: For other distribution branches besides the monitoring branch, determine whether there is synchronous fluctuation in the time neighborhood of the current curve of each distribution branch at each moment of the characteristic fluctuation time. If the number of other distribution branches with synchronous fluctuation reaches or exceeds the preset ratio, it is determined that the current fluctuation originates from the bus fluctuation; otherwise, it is determined that it originates from the monitoring branch fluctuation. Step 4: If it is determined to be a fluctuation in the monitoring branch, perform a Fourier transform on the current curve of the monitoring branch within the monitoring window to obtain the corresponding spectrum information, identify the frequency in the spectrum information whose amplitude exceeds the third threshold, take the identified frequency as the characteristic frequency, and take the corresponding amplitude as the characteristic amplitude. Step 5: Obtain the current curve of the known fault type and perform Fourier transform on it. Identify the frequency in the spectrum information whose amplitude exceeds the fourth threshold as the standard frequency and the corresponding amplitude as the standard amplitude to form a fault template library. Match the standard frequency and standard amplitude of each fault type in the fault template library with the characteristic frequency and characteristic amplitude of the monitoring branch. Determine the fault type of the monitoring branch based on the matching result.

2. The power distribution unit monitoring method according to claim 1, characterized in that: The logic for setting the monitoring window includes: taking the moment when the rate of change of current in a distribution branch exceeds a first threshold as the center, extending forward and backward by a preset fixed duration to determine the start and end times of the monitoring window.

3. The power distribution unit monitoring method according to claim 1, characterized in that: The time neighborhood refers to a time offset threshold that is set, and the intervals before and after each moment in the characteristic fluctuation are less than or equal to the time offset threshold, which are taken as the time neighborhood of that moment.

4. The power distribution unit monitoring method according to claim 3, characterized in that: Determine whether there are synchronous fluctuations in the time neighborhood of the current curve of each distribution branch at each time point within the characteristic fluctuation period, specifically including: For other distribution branches besides the monitoring branch, a fluctuation threshold is set. From the real-time current change rate, the maximum current change rate of each distribution branch in the time neighborhood of each moment in the characteristic fluctuation moment is extracted, and the maximum current change rate is compared with the fluctuation threshold. If the maximum current change rate exceeds the fluctuation threshold, it is determined that the distribution branch has current fluctuation at the characteristic fluctuation moment. Repeat the above process to count the number of characteristic fluctuation moments with current fluctuation in each distribution branch. Use this count as the number of synchronization moments. Set a synchronization threshold and compare the number of synchronization moments with the synchronization threshold. If the number of synchronization moments reaches or exceeds the synchronization threshold, it is determined that the distribution branch has synchronization fluctuations; otherwise, it is determined that the distribution branch does not have synchronization fluctuations.

5. The power distribution unit monitoring method according to claim 1, characterized in that: The spectrum information is frequency domain data obtained by performing a Fourier transform on the current curve of the monitoring branch within the monitoring window. The frequency domain data includes several frequencies and the amplitude corresponding to each frequency.

6. The power distribution unit monitoring method according to claim 1, characterized in that: The standard frequency and standard amplitude of each fault type in the fault template library are matched with the characteristic frequency and characteristic amplitude of the monitoring branch. Specifically, this includes: Set a frequency offset threshold. Take each of the characteristic frequencies as the center and define the intervals before and after that frequency that are less than or equal to the frequency offset threshold as the frequency neighborhood of that frequency. For each fault type in the fault template library, obtain the standard frequency and standard amplitude of the fault type. For each standard frequency, determine whether there is a characteristic frequency in the characteristic frequency of the monitoring branch such that the standard frequency is in the frequency neighborhood of the characteristic frequency. The standard frequency that meets the condition is taken as the candidate frequency of the fault type.

7. A power distribution unit monitoring method according to claim 6, characterized in that: Also includes: Amplitude tolerance range is set. Based on the candidate frequencies, it is further determined whether the difference between the standard amplitude corresponding to each candidate frequency and the characteristic amplitude of the characteristic frequency corresponding to the candidate frequency is within the amplitude tolerance range. The candidate frequencies that meet the above conditions are used as the matching frequencies for the fault type. The number of matching frequencies is counted as the number of matches for the fault type. Calculate the matching degree for each fault type. The matching degree is equal to the number of matches for that fault type divided by the total number of standard frequencies included in that fault type. Set a matching degree threshold and select fault types with a matching degree that reaches or exceeds the matching degree threshold as candidate fault types.

8. A power distribution unit monitoring method according to claim 7, characterized in that: Also includes: If there are no candidate fault types that meet the above conditions, the fault type with the highest matching degree among the matching degrees that have not reached the threshold will be selected as the fault type of the monitoring branch, and a prompt will be added. If only one fault type is included among the candidate fault types, then that fault type is selected as the fault type of the monitoring branch. If the candidate fault types include multiple fault types, the fault type with the highest matching degree is selected as the fault type of the monitoring branch; if multiple fault types have the same highest matching degree, the fault type with the smallest cumulative deviation between the standard amplitude of the matching frequency and the characteristic amplitude of the corresponding monitoring branch characteristic frequency is selected as the fault type of the monitoring branch.

9. A power distribution unit monitoring method according to claim 8, characterized in that: The fault type with the smallest cumulative deviation between the standard amplitude of the matching frequency and the characteristic amplitude of the corresponding monitoring branch's characteristic frequency is selected, specifically including: The fault types with the highest matching degree among the candidate fault types are selected as parallel candidate fault types. The absolute value of the difference between the standard amplitude of each matching frequency and the characteristic amplitude of the corresponding monitoring branch characteristic frequency in each parallel candidate fault type is calculated as the amplitude deviation of the parallel candidate fault type. The amplitude deviations are summed to obtain the cumulative deviation of the parallel candidate fault type. The cumulative deviations of all parallel candidate fault types are compared, and the fault type with the smallest cumulative deviation is selected as the fault type of the monitoring branch.

10. A power distribution unit monitoring system, the system being used to implement the power distribution unit monitoring method according to any one of claims 1-9, characterized in that, include: Real-time monitoring module: Used to obtain the current change rate of all distribution branches of the power distribution unit in real time. When the current change rate of a distribution branch exceeds the set first threshold, the distribution branch is marked as a monitoring branch. The monitoring window is set with the current time as the center, and the current curves of all distribution branches are obtained within the monitoring window. Feature extraction module: used to identify all times when the rate of change of current exceeds a set second threshold based on the current curve of the monitored branch within the monitoring window. The second threshold is less than the first threshold, and the identified times are used as feature fluctuation times. Fluctuation identification module: For other distribution branches besides the monitoring branch, it is used to determine whether there is synchronous fluctuation in the time neighborhood of the current curve of each distribution branch at each moment of the characteristic fluctuation moment. If the number of other distribution branches with synchronous fluctuation reaches or exceeds a preset ratio, it is determined that the current fluctuation originates from the bus fluctuation; otherwise, it is determined that it originates from the monitoring branch fluctuation. Spectrum analysis module: If the fluctuation of the monitored branch is determined, the current curve of the monitored branch within the monitoring window is subjected to Fourier transform to obtain the corresponding spectrum information, the frequency with amplitude exceeding the third threshold in the spectrum information is identified, the identified frequency is used as the characteristic frequency, and the corresponding amplitude is used as the characteristic amplitude. Fault matching module: It is used to acquire the current curve of known fault types, perform Fourier transform on it, identify the frequency with amplitude exceeding the fourth threshold in the spectrum information as the standard frequency, and the corresponding amplitude as the standard amplitude to form a fault template library. It matches the standard frequency and standard amplitude of each fault type in the fault template library with the characteristic frequency and characteristic amplitude of the monitoring branch, and determines the fault type of the monitoring branch based on the matching result.

Images