Impedance spectrum data processing method, device and system of current transformer and medium
By employing robust statistics to determine the cleaning threshold and data repair method in current transformers, the problem of inaccurate data cleaning caused by extreme outliers in online monitoring of current transformers is solved, achieving efficient data cleaning and abnormal feature repair under harsh operating conditions.
Patent Information
- Application Number
- CN202610778447.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-01
- Publication Date
- 2026-08-25
AI Technical Summary
In the online monitoring of current transformers, the accuracy of data cleaning algorithms is affected by extreme outliers when facing the high-intensity electromagnetic shock environment in substations, resulting in inaccurate data cleaning.
A cleaning threshold method based on robust statistics is adopted. By acquiring broadband impedance spectrum data of current transformers, the cleaning threshold is determined by median absolute deviation and preset penalty coefficient. Transient abnormal features are identified and repaired. Data is stored in a ring buffer and repaired by first-order linear interpolation or cross-channel mapping.
It improves the ability to resist pulse interference under harsh operating conditions, ensures the accuracy of data cleaning, reduces false alarms and false negatives, and achieves accurate rejection of transient anomalies.
Smart Images
Figure CN122634441A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system sensing and measurement technology, and in particular to a method, device, system, and medium for processing impedance spectrum data of a current transformer. Background Technology
[0002] Currently, real-time acquisition and analysis of high-frequency sensor data are widely used in the condition assessment of smart grid equipment, such as in the online monitoring and fault diagnosis system of current transformers in substations. In a typical online monitoring architecture for current transformers, a broadband excitation signal generator injects multi-frequency excitation into the secondary circuit of the current transformer. The sensing acquisition unit acquires the raw broadband impedance spectrum data (such as waveform data) in real time and transmits it to the edge terminal in the form of streaming data. The edge terminal can preprocess the streaming data before inputting it to the backend host device or server for online evaluation or error compensation.
[0003] In related technologies, data cleaning algorithms based on mean and variance (such as moving average filtering algorithms) are used for data preprocessing. While these algorithms can smooth data to a certain extent under stable operating conditions, extreme outliers (such as spikes) can severely skew the mean and variance when faced with high-intensity electromagnetic shock environments caused by frequent circuit breaker switching and lightning strikes in substations. This can lead to the failure of the data cleaning algorithm itself and affect the accuracy of data cleaning. Therefore, improving the ability to resist pulse interference under harsh operating conditions and ensuring the accuracy of data cleaning is an urgent problem to be solved. Summary of the Invention
[0004] The purpose of this invention is to provide a method, device, system, and computer-readable storage medium for processing impedance spectrum data of current transformers, so as to improve the anti-pulse interference capability under harsh operating conditions and ensure the accuracy of data cleaning.
[0005] To solve the above-mentioned technical problems, the present invention provides a method for processing impedance spectrum data of a current transformer, comprising: Acquire and store broadband impedance spectrum data of current transformers; wherein, the broadband impedance spectrum data includes impedance characteristics corresponding to each frequency in a preset frequency set; Based on the robust statistics corresponding to each impedance feature sequence in the target impedance spectrum data, the cleaning threshold corresponding to each target frequency is determined; wherein, the target impedance spectrum data includes the impedance feature sequence within a preset time period corresponding to each target frequency in the broadband impedance spectrum data, and the target frequency is the frequency in the preset frequency set. Based on the cleaning threshold, determine the transient anomaly characteristics corresponding to each of the target frequencies in the broadband impedance spectrum data; Based on the target impedance characteristics in the broadband impedance spectrum data, the transient anomaly characteristics are repaired to obtain the target feature matrix; wherein, the target impedance characteristics include impedance characteristics adjacent to the transient anomaly characteristics.
[0006] On the other hand, acquiring and storing the broadband impedance spectrum data of the current transformer includes: Receive broadband impedance spectrum data of the current transformer collected by the sensing and acquisition unit according to a preset sampling period; The broadband impedance spectrum data is stored using a circular buffer; wherein the target impedance spectrum data is the impedance characteristic in a multidimensional sliding data window in the circular buffer, the time window length of the multidimensional sliding data window is a preset number of preset sampling periods, and the multidimensional sliding data window is updated internally in a first-in-first-out manner.
[0007] On the other hand, the robust statistic is the median absolute deviation; the step of determining the cleaning threshold corresponding to each target frequency based on the robust statistic corresponding to each impedance characteristic sequence in the target impedance spectrum data includes: Extract the median from the current impedance feature sequence; wherein the current impedance feature sequence is any of the impedance feature sequences described above; Based on the median, calculate the absolute deviation between each impedance feature in the current impedance feature sequence and the median; Extract the median from the absolute deviations to obtain the median absolute deviation corresponding to the current impedance characteristic sequence; The cleaning threshold corresponding to the current impedance characteristic sequence is determined based on the median absolute deviation of the current impedance characteristic sequence.
[0008] On the other hand, determining the cleaning threshold corresponding to the current impedance feature sequence based on the median absolute deviation of the current impedance feature sequence includes: The cleaning threshold corresponding to the current impedance feature sequence is determined based on the median absolute deviation, the preset penalty coefficient, and the preset normal distribution scaling constant.
[0009] On the other hand, determining the transient anomaly characteristics corresponding to each of the target frequencies in the broadband impedance spectrum data based on the cleaning threshold includes: If the absolute value of the difference between the current impedance feature and the target median is greater than the target cleaning threshold, then the current impedance feature is determined to be the transient anomaly feature; wherein, the current impedance feature is any impedance feature corresponding to the target frequency in the broadband impedance spectrum data; the target median is the median of the impedance feature sequence corresponding to the target frequency of the current impedance feature; and the target cleaning threshold is the cleaning threshold corresponding to the target frequency of the current impedance feature.
[0010] On the other hand, the step of repairing the transient anomaly features based on the target impedance features in the broadband impedance spectrum data to obtain the target feature matrix includes: If the number of consecutive anomalous features corresponding to the current transient anomalous feature is less than or equal to the continuous interference threshold, then the current transient anomalous feature is repaired using first-order linear interpolation based on the target impedance feature corresponding to the current transient anomalous feature in the target impedance feature sequence; wherein, the current transient anomalous feature is any unrepaired transient anomalous feature, and the number of consecutive anomalous features is the length of the transient anomalous feature sequence including the current transient anomalous feature in the target impedance feature sequence; the target impedance feature sequence is the sequence of impedance features in the broadband impedance spectrum data corresponding to the frequency of the current transient anomalous feature; If the number of consecutive abnormal features corresponding to the current transient abnormal feature is greater than the continuous interference threshold, then the current transient abnormal feature is repaired by cross-channel mapping based on the target impedance feature corresponding to the current transient abnormal feature in the adjacent impedance feature sequence; wherein, the adjacent impedance feature sequence is at least one impedance feature sequence in the broadband impedance spectrum data that is adjacent to the target impedance feature sequence.
[0011] On the other hand, the step of repairing the current transient anomaly feature using first-order linear interpolation based on the target impedance feature corresponding to the current transient anomaly feature in the target impedance feature sequence includes: pass Calculate the repaired impedance characteristics corresponding to the current transient anomaly characteristics; where, This refers to the sampling time of the current transient anomaly feature; For the target impedance characteristic sequence The impedance feature that is not the transient anomaly feature at the latest sampling time; For the target impedance characteristic sequence The impedance feature that is not the transient anomaly feature at the earliest sampling time thereafter; for The sampling time; for The sampling time.
[0012] The present invention also provides an impedance spectrum data processing device for a current transformer, comprising: Memory, used to store computer programs; A processor is used to execute the computer program to implement the steps of the impedance spectrum data processing method for current transformers as described above.
[0013] The present invention also provides an impedance spectrum data processing system for a current transformer, comprising: a sensing acquisition unit and an edge device; The edge device is an impedance spectrum data processing device for a current transformer as described above; the sensing and acquisition unit is used to collect and send broadband impedance spectrum data of the current transformer to the edge device.
[0014] Furthermore, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the impedance spectrum data processing method for a current transformer as described above.
[0015] The present invention provides a method for processing impedance spectrum data of a current transformer, comprising: acquiring and storing broadband impedance spectrum data of the current transformer; wherein the broadband impedance spectrum data includes impedance characteristics corresponding to each frequency in a preset frequency set; determining a cleaning threshold corresponding to each target frequency based on the robust statistics corresponding to each impedance characteristic sequence in the target impedance spectrum data; wherein the target impedance spectrum data includes impedance characteristic sequences within a preset time period corresponding to each target frequency in the broadband impedance spectrum data, and the target frequencies are frequencies in the preset frequency set; determining transient abnormal characteristics corresponding to each target frequency in the broadband impedance spectrum data based on the cleaning thresholds; repairing the transient abnormal characteristics based on the target impedance characteristics in the broadband impedance spectrum data to obtain a target feature matrix; wherein the target impedance characteristics include impedance characteristics adjacent to the transient abnormal characteristics.
[0016] As can be seen, this invention, by determining the transient anomaly characteristics corresponding to each target frequency in broadband impedance spectrum data based on a cleaning threshold, can quickly and accurately identify robust transient anomaly characteristics in broadband impedance spectrum data using a cleaning threshold determined based on robust statistics. This allows for the repair and removal of transient anomaly characteristics, immunizing against the influence of extreme outliers on the overall judgment baseline. It achieves precise rejection of transient anomaly characteristics, improves the resistance to pulse interference under harsh operating conditions, and ensures the accuracy of data cleaning. Furthermore, this invention also provides an impedance spectrum data processing device, system, and computer-readable storage medium for current transformers, which also possess the aforementioned beneficial effects. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0018] Figure 1 This is a flowchart of an impedance spectrum data processing method for a current transformer provided in an embodiment of the present invention; Figure 2This is a flowchart illustrating another impedance spectrum data processing method for a current transformer provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the structure of an impedance spectrum data processing device for a current transformer provided in an embodiment of the present invention. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0020] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a method for processing impedance spectrum data of a current transformer, provided in an embodiment of the present invention. The method may include: Step 101: Acquire and store the broadband impedance spectrum data of the current transformer; wherein, the broadband impedance spectrum data includes the impedance characteristics corresponding to each frequency in the preset frequency set.
[0021] It is understood that the broadband impedance spectrum data of the current transformer in this embodiment can be the original broadband impedance spectrum data collected by the sensing and acquisition unit, such as the original broadband impedance spectrum data collected in real time by the sensing and acquisition unit after the broadband excitation signal generator injects multi-frequency excitation into the secondary circuit of the current transformer. The broadband impedance spectrum data can include the impedance characteristics corresponding to each frequency in the preset frequency set, such as amplitude or phase. That is, the broadband impedance spectrum data can be an impedance characteristic matrix of M rows and A columns, where M can be the number of frequencies in the preset frequency set, and A can be the number of samples of the stored broadband impedance spectrum data. If the sensing and acquisition unit collects the broadband impedance spectrum data of the current transformer according to a preset sampling period, this step can acquire and store broadband impedance spectrum data for A preset sampling periods (i.e., (the characteristic matrix).
[0022] Correspondingly, the method provided in this embodiment can be applied to edge terminals. That is, the edge terminal can process the raw broadband impedance spectrum data collected by the sensing unit to identify and repair transient anomalies, obtaining a target feature matrix, which is then sent to the backend host device or server for online evaluation or error compensation. The method provided in this embodiment can also be applied to host devices or servers. For example, a server can process the raw broadband impedance spectrum data collected by the sensing unit sent by the edge terminal to identify and repair transient anomalies, obtaining a target feature matrix for subsequent online evaluation or error compensation.
[0023] Accordingly, the specific method for acquiring and storing the broadband impedance spectrum data of the current transformer in this step can be set by the designer according to the practical scenario and user needs. For example, when the method provided in this embodiment is applied to an edge terminal, the edge terminal can receive and store the broadband impedance spectrum data of the current transformer collected by the sensing acquisition unit according to a preset sampling period in this step; when the method provided in this embodiment is applied to a server, the server can receive and store the broadband impedance spectrum data of the current transformer collected by the sensing acquisition unit according to a preset sampling period sent by the edge device in this step.
[0024] Correspondingly, in this embodiment, the broadband impedance spectrum data of the current transformer can be streaming data. For example, the sensing acquisition unit can send broadband impedance spectrum data to the edge terminal in streaming data form. To facilitate the storage and updating of broadband impedance spectrum data and the updating of target impedance spectrum data, this embodiment can utilize a ring buffer to store broadband impedance spectrum data, that is, use a ring buffer mechanism to cache broadband impedance spectrum data. This results in extremely low computational and storage resource consumption for the entire streaming data cleaning pipeline, making it easy to deploy in low-power edge terminals. This embodiment does not impose any limitations on this.
[0025] Step 102: Determine the cleaning threshold corresponding to each target frequency based on the robust statistics corresponding to each impedance characteristic sequence in the target impedance spectrum data.
[0026] The target impedance spectrum data includes the impedance characteristic sequence of each target frequency within a preset time period in the broadband impedance spectrum data, and the target frequency is the frequency in the preset frequency set.
[0027] It is understood that the target frequency in this embodiment can be all or part of the frequencies in a preset frequency set, that is, the frequencies corresponding to the impedance features in the broadband impedance spectrum data that need to be cleaned (i.e., transient anomaly feature identification). For example, the target frequency can be all the frequencies in the preset frequency set. The target impedance spectrum data can be the impedance features in the broadband impedance spectrum data used to determine the cleaning threshold for each target frequency; the target impedance spectrum data can include the impedance feature sequence within a preset time period corresponding to each target frequency in the broadband impedance spectrum data, that is, the cleaning threshold corresponding to the target frequency is determined by using the impedance feature sequence corresponding to each target frequency in the target impedance spectrum data.
[0028] Correspondingly, the specific settings for the aforementioned preset time period can be customized by the designer based on the practical scenario and user needs. For example, the preset time period can be a time period containing a preset number (N) of sampling times, ending at a certain sampling time, corresponding to the stored broadband impedance spectrum data. That is, the preset time period can include N consecutive sampling times. If the target frequency can be all M frequencies in the preset frequency set, the target impedance spectrum data can be... The feature matrix; to reduce computation, the preset time period can be a time period containing N sampling times ending at the next sampling time (such as the second new sampling time); to improve accuracy, the preset time period can be a time period containing N sampling times ending at the latest sampling time. This embodiment does not impose any restrictions on this.
[0029] Furthermore, to facilitate the selection and use of target impedance spectrum data, this embodiment can utilize a multi-dimensional sliding data window to determine the target impedance spectrum data. That is, the target impedance spectrum data can be the impedance characteristics within the multi-dimensional sliding data window. The time window length of the multi-dimensional sliding data window is a preset number (N) of preset sampling periods. For example, if the target frequency can be all M frequencies in a preset frequency set, the multi-dimensional sliding data window can conveniently determine the target impedance spectrum data. The characteristic matrix of .
[0030] Correspondingly, when using a circular buffer to store broadband impedance spectrum data, the target impedance spectrum data is the impedance characteristics in a multidimensional sliding data window within the circular buffer. The multidimensional sliding data window can update its internal data in a first-in-first-out (FIFO) manner. That is, after the data frame (i.e., impedance characteristics at M frequencies) of a new sampling moment is stored in the circular buffer, the multidimensional sliding data window is updated in a first-in-first-out (FIFO) manner, which can achieve microsecond-level displacement updates and provide data slices for subsequent real-time detection. Moreover, since seasonal temperature changes in the environment can cause the impedance spectrum to drift slowly, the dynamic window sliding mechanism of the multidimensional sliding data window in this embodiment can naturally follow this slow change and strictly distinguish it from transient anomalies. This allows a robust "data purification firewall" to be established between the hardware sensor and the subsequent AI (artificial intelligence) evaluation model, significantly reducing the false alarm and false negative rates of the system.
[0031] It should be noted that the robust statistics in this embodiment can be statistical measures used to describe data anomalies. Robust statistics can be statistics determined based on ordinal statistics (such as median or quartiles), such as Median Absolute Deviation (MAD) determined based on the median. The cleaning threshold in this step can be a threshold used to identify transient anomaly characteristics.
[0032] Correspondingly, the specific method for determining the cleaning threshold corresponding to each target frequency based on the robust statistics corresponding to each impedance feature sequence in the target impedance spectrum data in this step can be set by the designer. For example, if the robust statistics are median absolute deviations determined based on the median, the median in the current impedance feature sequence can be extracted in this step; where the current impedance feature sequence is any impedance feature sequence; based on the median, the absolute deviations of each impedance feature in the current impedance feature sequence from the median are calculated; the median of the absolute deviations is extracted to obtain the median absolute deviation corresponding to the current impedance feature sequence; and the cleaning threshold corresponding to the current impedance feature sequence is determined based on the median absolute deviation corresponding to the current impedance feature sequence. If the robust statistics are statistics determined based on other order statistics (such as quartiles), a similar method can be used to determine the corresponding cleaning threshold, and this embodiment does not impose any restrictions on this.
[0033] The specific method for determining the cleaning threshold corresponding to the current impedance characteristic sequence based on the median absolute deviation can be set by the designer. For example, the cleaning threshold can be determined based on the median absolute deviation, a preset penalty coefficient, and a preset normal distribution scaling constant. The preset penalty coefficient can be a pre-set value, such as a fixed value, or an adaptively adjusted penalty coefficient based on the background electromagnetic noise, for example, a value of 3. The preset normal distribution scaling constant can be a pre-set normal distribution scaling constant, such as 1.4826. For instance, the cleaning threshold (i.e., the value of the cleaning threshold) can be determined by calculating the product of the median absolute deviation, the preset penalty coefficient, and the preset normal distribution scaling constant. );in, This can be the cleaning threshold corresponding to the current impedance feature sequence. You can preset the penalty coefficient. A preset normal distribution scaling constant can be provided. The cleaning threshold can be the median absolute deviation of the current impedance characteristic sequence; alternatively, it can be determined by calculating the product of the median absolute deviation of the current impedance characteristic sequence and a preset penalty coefficient, a preset normal distribution scaling constant, and the frequency coefficient of the current impedance characteristic sequence. This embodiment does not impose any limitations on this.
[0034] Correspondingly, the target impedance spectrum data in this embodiment may include the repaired impedance characteristics, such as the impedance characteristics repaired when they were previously identified as transient abnormal characteristics.
[0035] For example, the method provided in this embodiment can be deployed as a software algorithm in the processor of an edge terminal at a substation site. During system operation, a wideband excitation signal generator injects multi-frequency excitation into the secondary circuit of the current transformer. The sensing and acquisition unit acquires the raw high-frequency impedance spectrum data in real time and sends it to the edge terminal in the form of streaming data. The edge terminal can use an edge buffer (i.e., a circular buffer) to store the received high-frequency impedance spectrum data and use a multi-dimensional sliding data window based on the edge buffer to update the impedance characteristic sequence in the target impedance spectrum data in real time. Figure 2 As shown, the method provided in this embodiment may further include the process of constructing a multi-dimensional sliding data window based on edge caching, such as the edge terminal opening a circular buffer in memory; let the current time be... For the collected A set of discrete frequency points (i.e., a preset frequency set) The time window length of the multi-dimensional sliding data window is set to N preset sampling periods. For any specific frequency... Its time series within the time window is defined as As new data frames arrive, the multidimensional sliding data window is updated with microsecond-level displacement in a first-in-first-out (FIFO) manner, enabling it to naturally follow the slow evolution characteristics of the impedance baseline.
[0036] Correspondingly, such as Figure 2 As shown, the edge terminal can extract the median value within a multi-dimensional sliding data window, calculate the median absolute deviation (MAD), and establish a dynamic cleaning threshold. For example, it can be used for specific frequencies. time series First, calculate the median of the N data points (i.e., impedance characteristics) within the window as the reference value: Then, the absolute deviation of each data point from the median is calculated, and the median of the set of absolute deviations is obtained to get the median absolute deviation: Based on this, establish a specific frequency. Corresponding cleaning threshold .in, The penalty coefficient can be adaptively adjusted according to the background electromagnetic noise of the site, such as a value of 3; The preset normal distribution scaling constant can be set to a value of 1.4826.
[0037] Step 103: Based on the cleaning threshold, determine the transient anomaly characteristics corresponding to each target frequency in the broadband impedance spectrum data.
[0038] It is understandable that the transient anomaly features in this step can be robust anomaly impedance features determined using the cleaning threshold, i.e., extreme outliers. The specific method for determining the transient anomaly features corresponding to each target frequency in the broadband impedance spectrum data based on the cleaning threshold in this step can be set by the designer according to the practical scenario and user needs. For example, the transient anomaly features corresponding to each target frequency in the broadband impedance spectrum data can be determined based on the cleaning threshold and robust statistics. For instance, when the robust statistics are median absolute deviation, this step can determine the current impedance feature as a transient anomaly feature when the absolute value of the difference between the current impedance feature and the target median is greater than the target cleaning threshold; and determine the current impedance feature as a normal impedance feature when the absolute value of the difference between the current impedance feature and the target median is not greater than the target cleaning threshold. Here, the current impedance feature is any impedance feature corresponding to the target frequency in the broadband impedance spectrum data; the target median is the median of the impedance feature sequence corresponding to the target frequency of the current impedance feature; and the target cleaning threshold is the cleaning threshold corresponding to the target frequency of the current impedance feature.
[0039] Correspondingly, in this step, transient anomaly detection can be performed on impedance features in broadband impedance spectrum data that have not been previously detected using the corresponding cleaning threshold (such as...). Figure 2 (The deviation judgment in the process) reduces the amount of detection. For example, in this step, the transient anomaly characteristics corresponding to each target frequency in the latest impedance spectrum data can be determined based on the cleaning threshold; among them, the impedance characteristics corresponding to the latest sampling time in the broadband impedance spectrum data, i.e. Each impedance characteristic. In other words, after the latest impedance spectrum data arrives and is stored, the content of the target impedance spectrum data can be updated, thereby updating the cleaning threshold accordingly. The updated cleaning threshold can then be used to detect transient anomalies in the latest impedance spectrum data.
[0040] This step can also detect transient anomalies in the impedance characteristics prior to the target sampling time in the broadband impedance spectrum data. The target sampling time can be a sampling time after a preset time period. This embodiment does not impose any restrictions on this.
[0041] For example, such as Figure 2 As shown, this step can determine the deviation of the most recently acquired streaming data in the edge buffer based on the cleaning threshold, and mark and remove transient abnormal data points (i.e., transient abnormal features). For example, for a certain impedance characteristic of the most recently transmitted data from the sensor acquisition unit... The edge terminal calculates its deviation from the corresponding median; if the anomaly detection condition is met: If the impedance characteristic is not detected, it is determined to be a transient anomaly, i.e., a transient distortion spike caused by strong electromagnetic pulse interference such as the switching on or off of a circuit breaker or a lightning strike. Accordingly, the edge terminal can remove this data point (i.e., the transient anomaly) from the valid queue and update its timestamp and frequency coordinates. The void state is marked as NaN to be repaired.
[0042] Step 104: Based on the target impedance characteristics in the broadband impedance spectrum data, repair the transient anomaly characteristics to obtain the target feature matrix; wherein, the target impedance characteristics include the impedance characteristics adjacent to the transient anomaly characteristics.
[0043] It is understandable that in this step, the target impedance features (such as adjacent impedance features) corresponding to each transient anomaly feature (such as the data points in the missing state mentioned above) in the broadband impedance spectrum data can be used to repair the transient anomaly features, thereby obtaining a high-quality feature matrix (i.e., the target feature matrix) with a complete topological structure and smooth changes.
[0044] Correspondingly, the target impedance feature in this embodiment can be the impedance feature required to repair the transient anomaly feature. This embodiment does not limit the specific setting of the target impedance feature. For example, the target impedance feature may include all or part of the impedance features adjacent to the transient anomaly feature. The impedance features adjacent to the transient anomaly feature in the target impedance feature may be impedance features that are not transient anomalies (e.g., the original impedance features or the repaired impedance features).
[0045] It should be noted that the specific method for repairing transient anomaly features and obtaining the target feature matrix based on the target impedance characteristics in the broadband impedance spectrum data in this step can be set by the designer according to the practical scenario and user needs. For example, the same repair method (such as first-order linear interpolation) can be directly used to repair each transient anomaly feature. In order to solve the distortion and divergence problems that are prone to occur when single-time dimension interpolation faces persistent data anomalies, this embodiment can utilize the physical correlation of the broadband impedance spectrum to perform cross-channel mapping compensation when reconstructing persistent data anomalies, so as to realize a two-dimensional adaptive compensation closed loop of priority interpolation for slight time domain missingness and cross-channel mapping for severe frequency domain missingness. In other words, in this step, when the number of consecutive abnormal features corresponding to the current transient abnormal feature is less than or equal to the continuous interference threshold, the current transient abnormal feature can be repaired using first-order linear interpolation based on the target impedance feature corresponding to the current transient abnormal feature in the target impedance feature sequence; when the number of consecutive abnormal features corresponding to the current transient abnormal feature is greater than the continuous interference threshold, the current transient abnormal feature can be repaired using cross-channel mapping based on the target impedance feature corresponding to the current transient abnormal feature in the adjacent impedance feature sequence; where the current transient abnormal feature is any unrepaired transient abnormal feature (such as the data point marked as missing above), the number of consecutive abnormal features is the length of the transient abnormal feature sequence including the current transient abnormal feature in the target impedance feature sequence; the target impedance feature sequence is the sequence of impedance features in the broadband impedance spectrum data corresponding to the frequency of the current transient abnormal feature; and the adjacent impedance feature sequence is at least one impedance feature sequence in the broadband impedance spectrum data adjacent to the target impedance feature sequence.
[0046] Correspondingly, the repair process in this step can be carried out after the transient abnormal feature sequence including the current transient abnormal feature is complete or greater than the repair threshold. The repair threshold can be greater than the continuous interference threshold. For example, in this step, the transient abnormal feature in the transient abnormal feature sequence can be repaired when the impedance features adjacent to both ends of the transient abnormal feature sequence are not transient abnormal features.
[0047] The specific method for repairing the current transient anomaly feature using first-order linear interpolation based on the target impedance feature corresponding to the current transient anomaly feature in the target impedance feature sequence can be set by the designer. For example, it can be achieved through... Calculate the repaired impedance characteristics corresponding to the current transient anomaly characteristics; where, This refers to the sampling time of the current transient anomaly feature; For the target impedance characteristic sequence The impedance characteristic that is not a transient anomaly at the latest sampling time; For the target impedance characteristic sequence The impedance feature that is not a transient anomaly at the earliest sampling time thereafter; for The sampling time; for The sampling time. For example, if the number of consecutive anomalous features corresponding to the current transient anomaly is 1, meaning the gap only occurs at a single timestamp and the data before and after it is valid, the two adjacent impedance features in the target impedance feature sequence can be directly used to repair the current transient anomaly. If the number of consecutive anomalous features corresponding to the current transient anomaly is greater than 1 and less than or equal to the continuous interference threshold, the two adjacent impedance features at both ends of the transient anomaly feature sequence corresponding to the current transient anomaly can be directly used to repair all transient anomalies in the transient anomaly feature sequence. That is, the impedance features after repair of all transient anomalies in the transient anomaly feature sequence are the same. and It can be the original impedance characteristics; or it can be that each transient anomaly feature in the transient anomaly feature sequence is repaired sequentially by interpolation (e.g., from left to right), as described above. and It can include the repaired impedance characteristics, such as one being the original impedance characteristic and the other being the repaired impedance characteristic.
[0048] Similarly, this embodiment does not limit the specific method of repairing the current transient anomaly feature through cross-channel mapping based on the target impedance feature corresponding to the current transient anomaly feature in the adjacent impedance feature sequence. For example, it can use the healthy data features (i.e., impedance features that are not transient anomalies) in the adjacent impedance feature sequence that have the same sampling time as the current transient anomaly feature to perform cross-channel mapping using the continuity curve of impedance versus frequency, and calculate the repaired feature vector corresponding to the current transient anomaly feature. Alternatively, it can use the healthy data features at each sampling time corresponding to the transient anomaly feature sequence in the adjacent impedance feature sequence to perform cross-channel mapping using the continuity curve of impedance versus frequency, and directly repair the transient anomaly feature sequence.
[0049] For example, such as Figure 2As shown, this step can determine the length of consecutive missing data (i.e., the number of consecutive anomalous features), adaptively trigger time-domain first-order interpolation or frequency-domain cross-channel mapping, and complete two-dimensional joint reconstruction. For example, the edge terminal counts the number L of data points (i.e., transient anomalous feature sequences) continuously marked as missing (NaN) in the current frequency channel, and sets a continuous interference threshold P; if This indicates that the interference encountered is an isolated transient event; the edge terminal can extract the preceding valid points at the missing location. and subsequent effective points Fast time-domain repair can be achieved using first-order linear interpolation, and the calculation formula is as follows: ; This can be the impedance characteristic after repair of the current transient anomaly characteristics. If This indicates that continuous strong interference has been encountered, and there is a risk of divergence in time-domain interpolation. At this time, the edge terminal can adaptively switch to frequency-domain reconstruction mode to extract adjacent frequency channels at the same time (such as adjacent frequencies). and The health data characteristics of the data are used to perform cross-channel projection using the inherent polynomial physical correlation curves between different frequency points, and the missing segment is calculated and filled in.
[0050] Furthermore, to ensure the authenticity of the target feature matrix data, the method provided in this embodiment may also include a process of outputting an alarm message when the number of target abnormal feature sequences in the stored broadband impedance spectrum data exceeds a threshold. The target abnormal feature sequences may include transient abnormal feature sequences where the number of consecutive abnormal features corresponding to each frequency exceeds the abnormal feature threshold. This allows for the cessation of repairs and the direct reporting of a "serious communication / interference failure" alarm message to the host device or server when large-scale data anomalies occur across multiple frequency bands over a prolonged period, thus preventing the output target feature matrix from having excessively low data authenticity and implementing a security locking mechanism.
[0051] Correspondingly, the method provided in this embodiment may also include subsequent processing of the target feature matrix. For example, the edge terminal can input the target feature matrix into the error calculation module, or compress and package it before sending it to the CNN (Convolutional Neural Network) multi-channel error prediction model on the server. Figure 2 The state assessment model in the system ensures that the system maps the dynamic operating state of the current transformer without delay.
[0052] In this embodiment, the present invention determines the transient anomaly features corresponding to each target frequency in the broadband impedance spectrum data according to the cleaning threshold. By utilizing the cleaning threshold determined based on robust statistics, it can quickly and accurately identify robust transient anomaly features in the broadband impedance spectrum data, thereby repairing and removing transient anomaly features. This can prevent extreme outliers from affecting the overall judgment baseline, achieve precise rejection of transient anomaly features, improve the anti-pulse interference capability under harsh operating conditions, and ensure the accuracy of data cleaning.
[0053] Corresponding to the above method embodiments, this invention also provides an impedance spectrum data processing device for a current transformer. The impedance spectrum data processing device for a current transformer described below and the impedance spectrum data processing method for a current transformer described above can be referred to in correspondence.
[0054] Please refer to Figure 3 , Figure 3 This is a schematic diagram of the structure of an impedance spectrum data processing device for a current transformer provided in an embodiment of the present invention. The device may include: Memory D1 is used to store computer programs; Processor D2 is used to execute a computer program to implement the steps of the impedance spectrum data processing method for current transformers provided in the above method embodiments.
[0055] In this embodiment, the impedance spectrum data processing device for the current transformer can be specifically an edge terminal located near the current transformer; or it can be a host device or server that is communicatively connected to the edge terminal.
[0056] Corresponding to the above method embodiments, this invention also provides an impedance spectrum data processing system for a current transformer. The impedance spectrum data processing system for a current transformer described below and the impedance spectrum data processing method for a current transformer described above can be referred to in correspondence.
[0057] An impedance spectrum data processing system for a current transformer includes: a sensing and acquisition unit and an edge device; The edge device is an impedance spectrum data processing device for a current transformer as provided in the above embodiment; the sensing and acquisition unit is used to collect and send broadband impedance spectrum data of the current transformer to the edge device.
[0058] In some embodiments, the system provided in this embodiment may further include a server or host device that is communicatively connected to an edge device, for online evaluation or error compensation using a target feature matrix sent by the edge device.
[0059] Corresponding to the above method embodiments, this invention also provides a computer program product. The computer program product described below and the impedance spectrum data processing method for a current transformer described above can be referred to in correspondence.
[0060] A computer program product includes a computer program / instructions that, when executed by a processor, implement the steps of the impedance spectrum data processing method for a current transformer provided in the above-described method embodiments.
[0061] Corresponding to the above method embodiments, this invention also provides a computer-readable storage medium. The computer-readable storage medium described below and the impedance spectrum data processing method for a current transformer described above can be referred to in correspondence.
[0062] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the impedance spectrum data processing method for a current transformer as provided in the above-described method embodiments.
[0063] The computer-readable storage medium can specifically be a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other readable storage medium capable of storing program code.
[0064] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the devices, systems, computer-readable storage media, and computer program products disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple, and relevant details can be found in the method section.
[0065] The impedance spectrum data processing method, device, system, and computer-readable storage medium for a current transformer provided by this invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make various improvements and modifications to this invention without departing from its principles, and these improvements and modifications also fall within the protection scope of this invention.
Claims
1. A method for processing impedance spectrum data of a current transformer, characterized in that, include: Acquire and store broadband impedance spectrum data of current transformers; wherein, the broadband impedance spectrum data includes impedance characteristics corresponding to each frequency in a preset frequency set; Based on the robust statistics corresponding to each impedance feature sequence in the target impedance spectrum data, the cleaning threshold corresponding to each target frequency is determined; wherein, the target impedance spectrum data includes the impedance feature sequence within a preset time period corresponding to each target frequency in the broadband impedance spectrum data, and the target frequency is the frequency in the preset frequency set. Based on the cleaning threshold, determine the transient anomaly characteristics corresponding to each of the target frequencies in the broadband impedance spectrum data; Based on the target impedance characteristics in the broadband impedance spectrum data, the transient anomaly characteristics are repaired to obtain the target feature matrix; wherein, the target impedance characteristics include impedance characteristics adjacent to the transient anomaly characteristics.
2. The impedance spectrum data processing method for a current transformer according to claim 1, characterized in that, The acquisition and storage of broadband impedance spectrum data of the current transformer includes: Receive broadband impedance spectrum data of the current transformer collected by the sensing and acquisition unit according to a preset sampling period; The broadband impedance spectrum data is stored using a circular buffer; wherein the target impedance spectrum data is the impedance characteristic in a multidimensional sliding data window in the circular buffer, the time window length of the multidimensional sliding data window is a preset number of preset sampling periods, and the multidimensional sliding data window is updated internally in a first-in-first-out manner.
3. The impedance spectrum data processing method for a current transformer according to claim 1, characterized in that, The robust statistic is the median absolute deviation; the step of determining the cleaning threshold corresponding to each target frequency based on the robust statistic corresponding to each impedance characteristic sequence in the target impedance spectrum data includes: Extract the median from the current impedance feature sequence; wherein the current impedance feature sequence is any of the impedance feature sequences described above; Based on the median, calculate the absolute deviation between each impedance feature in the current impedance feature sequence and the median; Extract the median from the absolute deviations to obtain the median absolute deviation corresponding to the current impedance characteristic sequence; The cleaning threshold corresponding to the current impedance characteristic sequence is determined based on the median absolute deviation of the current impedance characteristic sequence.
4. The impedance spectrum data processing method for a current transformer according to claim 3, characterized in that, The step of determining the cleaning threshold corresponding to the current impedance feature sequence based on the median absolute deviation of the current impedance feature sequence includes: The cleaning threshold corresponding to the current impedance feature sequence is determined based on the median absolute deviation, the preset penalty coefficient, and the preset normal distribution scaling constant.
5. The impedance spectrum data processing method for a current transformer according to claim 3, characterized in that, The step of determining the transient anomaly characteristics corresponding to each of the target frequencies in the broadband impedance spectrum data based on the cleaning threshold includes: If the absolute value of the difference between the current impedance feature and the target median is greater than the target cleaning threshold, then the current impedance feature is determined to be the transient anomaly feature; wherein, the current impedance feature is any impedance feature corresponding to the target frequency in the broadband impedance spectrum data; the target median is the median of the impedance feature sequence corresponding to the target frequency of the current impedance feature; and the target cleaning threshold is the cleaning threshold corresponding to the target frequency of the current impedance feature.
6. The impedance spectrum data processing method for a current transformer according to any one of claims 1 to 5, characterized in that, The step of repairing the transient anomaly features based on the target impedance features in the broadband impedance spectrum data to obtain the target feature matrix includes: If the number of consecutive anomalous features corresponding to the current transient anomalous feature is less than or equal to the continuous interference threshold, then the current transient anomalous feature is repaired using first-order linear interpolation based on the target impedance feature corresponding to the current transient anomalous feature in the target impedance feature sequence; wherein, the current transient anomalous feature is any unrepaired transient anomalous feature, and the number of consecutive anomalous features is the length of the transient anomalous feature sequence including the current transient anomalous feature in the target impedance feature sequence; the target impedance feature sequence is the sequence of impedance features in the broadband impedance spectrum data corresponding to the frequency of the current transient anomalous feature; If the number of consecutive abnormal features corresponding to the current transient abnormal feature is greater than the continuous interference threshold, then the current transient abnormal feature is repaired by cross-channel mapping based on the target impedance feature corresponding to the current transient abnormal feature in the adjacent impedance feature sequence; wherein, the adjacent impedance feature sequence is at least one impedance feature sequence in the broadband impedance spectrum data that is adjacent to the target impedance feature sequence.
7. The impedance spectrum data processing method for a current transformer according to claim 6, characterized in that, The step of repairing the current transient anomaly feature using first-order linear interpolation based on the target impedance feature corresponding to the current transient anomaly feature in the target impedance feature sequence includes: pass Calculate the repaired impedance characteristics corresponding to the current transient anomaly characteristics; where, This refers to the sampling time of the current transient anomaly feature; For the target impedance characteristic sequence The impedance feature that is not the transient anomaly feature at the latest sampling time; For the target impedance characteristic sequence The impedance feature that is not the transient anomaly feature at the earliest sampling time thereafter; for The sampling time; for The sampling time.
8. An impedance spectrum data processing device for a current transformer, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of the impedance spectrum data processing method for a current transformer as described in any one of claims 1 to 7.
9. A system for processing impedance spectrum data of a current transformer, characterized in that, include: Sensing and data acquisition units and edge devices; Wherein, the edge device is the impedance spectrum data processing device for the current transformer as described in claim 8; The sensing and acquisition unit is used to collect and send broadband impedance spectrum data of the current transformer to the edge device.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the impedance spectrum data processing method for a current transformer as described in any one of claims 1 to 7.