Broadband current source output monitoring data processing method

Abstract

This invention belongs to the field of signal monitoring and processing technology, specifically relating to a method for processing broadband current source output monitoring data. The method includes: acquiring synchronous monitoring data of the broadband current source, dividing the output period and synchronously resampling to obtain a periodic monitoring data set; determining the current detail coefficient, the basic threshold, and the in-phase current mutation repetition rate; determining the abnormal current mutation discrimination value; determining the adaptive wavelet threshold and obtaining a denoised current monitoring sequence; obtaining the continuous overcurrent monitoring result and sending a protection command. This invention analyzes the in-phase repetition characteristics of current mutations within multiple output periods and their correspondence with the location of rapid changes in output voltage, adaptively adjusting the wavelet threshold to overcome the limitation of fixed thresholds in distinguishing between normal capacitive charging and discharging spikes and continuous overcurrent mutations, thereby improving the accuracy and timeliness of continuous overcurrent monitoring.

Description

Technical Field

[0001] This invention relates to the field of signal monitoring and processing technology. More specifically, this invention relates to a method for processing monitoring data from a broadband current source output. Background Technology

[0002] Broadband current sources can provide controlled current output to loads over a wide frequency range and can be used in scenarios such as piezoelectric ceramic driving, ultrasonic transduction, material testing, and electromagnetic field excitation. To determine the output status of a broadband current source, the device is usually equipped with a current monitoring terminal and a voltage monitoring terminal. The output current and output voltage are converted into monitoring voltages that can be received by the data acquisition device according to the device parameters, so that the protection module can perform overcurrent identification, current limiting, and shutdown control.

[0003] Loads such as piezoelectric ceramics and ultrasonic transducers have obvious capacitive characteristics. When the output voltage changes rapidly, the load will charge and discharge rapidly and form a short-time current spike. The instantaneous amplitude of the short-time current spike may be close to or exceed the instantaneous amplitude of the continuous overcurrent, but its duration is short and has a stable correspondence with the position of rapid output voltage change. It belongs to the capacitive transient that can be expected in the output process of the equipment. The continuous overcurrent will continue to exist in the position where the output voltage changes slowly and spans multiple output cycles. It is difficult to accurately distinguish between the two within the time required for protection action by simply comparing the current amplitude.

[0004] Existing processing methods typically employ wavelet denoising for current monitoring signals. The threshold for wavelet denoising is generally fixed and set based on the global noise level. While this can suppress some random noise, it does not utilize the repetitive pattern of in-phase current abrupt changes within adjacent output cycles, nor does it establish a correspondence between current abrupt changes and the location of rapid changes in output voltage. When the fixed threshold is too small, a significant amount of normal capacitive charging and discharging spikes remain, potentially triggering malfunctions in the protection module. When the fixed threshold is too large, the effective abrupt changes at the onset of continuous overcurrent are simultaneously weakened, causing a lag in continuous overcurrent identification. In particular, when normal capacitive charging and discharging spikes recur in every output cycle, the normal capacitive charging and discharging spikes may still be retained as abnormal current abrupt changes simply based on whether the current abrupt changes repeat across cycles.

[0005] In related technologies, for example, Chinese patent document CN101923139B, entitled "An Intelligent Method for Single-End Traveling Wave Fault Location in Transmission Lines," discloses a method for fault location based on current traveling waves. In the signal processing stage, it performs discrete wavelet decomposition on the collected current traveling wave data, uses a soft thresholding method to process the high-frequency coefficients at each decomposition scale, and extracts the polarity and time difference of the first three wavefronts from the processed current traveling wave. Then, it uses a neural network to determine the fault location of the transmission line. However, this patent document uses a method of determining the wavelet threshold based on the number of sampling points when denoising the current signal. Its drawback is that it does not consider the periodic characteristics of the broadband current source output signal, nor does it analyze the in-phase repetition of current abrupt changes within multiple output cycles, or the correspondence between current abrupt changes and the location of rapid changes in output voltage. When this method is applied to monitor the output of a broadband current source under capacitive load, both normal capacitive charging and discharging spikes and current mutations caused by continuous overcurrent may be manifested as high-frequency wavelet coefficients with large amplitudes. Simply processing based on a uniform threshold may easily retain normal capacitive charging and discharging spikes as abnormal signals or weaken the effective mutation characteristics at the beginning of continuous overcurrent, leading to misjudgment or delayed triggering of continuous overcurrent protection. Summary of the Invention

[0006] To address the technical problem that fixed-threshold wavelet denoising cannot simultaneously suppress normal capacitive charging and discharging peaks and preserve continuous overcurrent abrupt changes, this invention provides a broadband current source output monitoring data processing method, comprising: S001, acquiring synchronous monitoring data of the broadband current source, reading the current output frequency and output enable state, dividing the synchronous monitoring data into continuous output periods, and performing synchronous resampling to obtain a periodic current monitoring sequence and a periodic output voltage monitoring sequence; S002, calibrating the duration width of normal capacitive charging and discharging peaks, obtaining the current detail coefficients of each phase sampling point based on the duration width of normal capacitive charging and discharging peaks and the wavelet decomposition results of the periodic current monitoring sequence, and estimating the noise level of the detail layer corresponding to each output period based on the current detail coefficients. S003. Determine the basic threshold and all effective periods corresponding to the current output period. Count the number of times the in-phase current detail coefficient exceeds the basic threshold in the current output period and each effective period before it, and determine the in-phase current mutation repetition rate; S004. Determine the proportion of the voltage rapidly changing neighborhood based on the output voltage change rate of the phase sampling point, and obtain the abnormal current mutation discrimination value by combining it with the in-phase current mutation repetition rate; S005. Adjust the basic threshold according to the abnormal current mutation discrimination value to obtain an adaptive wavelet threshold, and use the adaptive wavelet threshold to obtain a denoised current monitoring sequence; S006. Obtain the continuous overcurrent monitoring result based on the duration for which the denoised current monitoring sequence continuously exceeds the overcurrent judgment current, and send a protection command to the broadband current source based on the continuous overcurrent monitoring result.

[0007] This invention establishes a correspondence between in-phase sampling points within different output cycles by dividing the output cycle. It then utilizes the in-phase current mutation repetition rate to reduce the retention requirement for random high-frequency disturbances that only occur in individual output cycles. Furthermore, it uses the proportion of the neighborhood of rapid voltage changes to handle the special case where normal capacitive charging and discharging spikes can also repeat across cycles. This allows the adaptive wavelet threshold to change the processing intensity based on the periodic repetition of current mutations and its relationship with the position of rapid changes in output voltage, thereby forming a complete data chain from synchronous monitoring data to protection commands.

[0008] Preferably, the calibration of the normal capacitive charge and discharge peak duration width includes: when the broadband current source is connected to a rated capacitive load, the output enable state is continuously effective, and the current output frequency and output amplitude remain unchanged, acquiring a periodic current monitoring sequence for several effective cycles; for each effective cycle, using the median of the current monitoring values ​​within that effective cycle as a reference value, determining the maximum absolute difference between each current monitoring value and the reference value and its corresponding peak time, determining the sampling time when the absolute difference between the current monitoring values ​​and the reference value before and after the peak time reaches half of the maximum absolute difference, using the time interval between the two sampling times as the normal capacitive charge and discharge peak duration width of that effective cycle, and using the median of the normal capacitive charge and discharge peak duration widths corresponding to several effective cycles as the normal capacitive charge and discharge peak duration width.

[0009] Preferably, the method for obtaining the current detail coefficients is as follows: performing discrete wavelet decomposition on each periodic current monitoring sequence, selecting the detail layer that covers the duration of the normal capacitive charging and discharging peak at the corresponding time scale from the multiple detail layers obtained by wavelet decomposition as the target detail layer, and using the coefficients of the target detail layer at each phase sampling point as the current detail coefficients.

[0010] Preferably, the method for obtaining the basic threshold is as follows: determine the median of all current detail coefficients in the current output cycle, divide the median of the absolute value of the difference between each current detail coefficient and the median by 0.6745, where 0.6745 is a constant for converting the absolute deviation of the median into the noise standard deviation, to obtain the noise level; multiply the noise level by the square root of twice the natural logarithm of the number of phase sampling points, and use the result as the basic threshold.

[0011] Preferably, the repetition rate of the in-phase current abrupt change satisfies the expression: In the formula, For the first The output cycle number The repetition rate of in-phase current abrupt change at each phase sampling point This includes the number of valid cycles, encompassing the current output cycle and all previous valid cycles. and The first The first effective period Current detail coefficients and basic thresholds at each phase sampling point It is an indicator function that takes the value 1 if the condition is true and 0 otherwise.

[0012] This invention establishes a phase comparison relationship among multiple effective cycles corresponding to the number of effective cycles, so that random high-frequency disturbances that only occur in individual output cycles obtain a small in-phase current mutation repetition rate, and current mutations that continuously occur in the same phase obtain a large in-phase current mutation repetition rate, thereby providing a direct basis for distinguishing between random disturbances and continuous overcurrent mutations.

[0013] Preferably, the method for obtaining the proportion of the voltage rapidly changing neighborhood is as follows: subtract the output voltage monitoring value of the previous phase sampling point from the output voltage monitoring value of a phase sampling point, and divide by the sampling time interval between the two phase sampling points to obtain the output voltage change rate of the phase sampling point. Calculate the output voltage change rate of a phase sampling point and its two adjacent phase sampling points. Count the number of phase sampling points whose absolute value of the output voltage change rate exceeds a preset voltage rapidly changing judgment value, and divide the number of phase sampling points by 3. Use the result as the proportion of the voltage rapidly changing neighborhood. The proportion of the voltage rapidly changing neighborhood is determined by connecting the first and last phase sampling points in the output cycle.

[0014] Preferably, the abnormal current sudden change discrimination value satisfies the expression: In the formula, , and The first The output cycle number The abnormal current mutation discrimination value, the repetition rate of the same phase current mutation, and the proportion of the neighborhood of rapid voltage change at each phase sampling point.

[0015] This invention corrects the repetition rate of in-phase current mutations by the proportion of the neighborhood of rapid voltage changes. When current mutations repeat across cycles but are concentrated at the location of rapid output voltage changes, the abnormal current mutation discrimination value decreases accordingly. When current mutations repeat across cycles and are located away from the location of rapid output voltage changes, the abnormal current mutation discrimination value remains large, thereby solving the special problem that normal capacitive charging and discharging peaks also have cross-cycle repetition.

[0016] Preferably, the adaptive wavelet threshold satisfies the expression: In the formula, and The first The output cycle number Adaptive wavelet threshold and abnormal current mutation discrimination value for each phase sampling point For the first The base threshold for each output cycle.

[0017] This invention converts the abnormal current mutation discrimination value into an adaptive wavelet threshold corresponding to each phase sampling point, so that the phase sampling points with larger abnormal current mutation discrimination values ​​adopt a processing intensity close to the basic threshold, and the phase sampling points with smaller abnormal current mutation discrimination values ​​adopt a larger processing intensity, thereby taking into account both the preservation of continuous overcurrent mutations and the suppression of normal capacitive charging and discharging peaks.

[0018] Preferably, obtaining the continuous overcurrent monitoring result includes: reading the continuous overcurrent setting current in the broadband current source protection parameters, and using the continuous overcurrent setting current as the overcurrent judgment current; when the absolute value of the denoised current monitoring sequence continuously exceeds the overcurrent judgment current, accumulating its duration; when the absolute value of the denoised current monitoring sequence does not exceed the overcurrent judgment current, clearing the accumulated duration to zero; in response to the accumulated duration being not less than the duration corresponding to the preset minimum number of duration cycles, generating the continuous overcurrent monitoring result, and sending a current limiting command, shutdown command, or alarm command to the protection module according to the protection configuration of the broadband current source.

[0019] This invention sets a continuous over-limit duration constraint on the noise-reducing current monitoring sequence, so that a single normal capacitive charge and discharge spike cannot directly trigger the protection module, and the output state that continuously exceeds the overcurrent judgment current and reaches the preset minimum duration period triggers the protection command, thereby reducing the probability of protection maloperation caused by normal capacitive transients.

[0020] Preferably, obtaining the periodic current monitoring sequence and the periodic output voltage monitoring sequence includes: acquiring the current monitoring voltage through the current monitoring terminal, converting the current monitoring voltage into a current monitoring sequence according to the current monitoring ratio corresponding to the broadband current source, and simultaneously acquiring the output voltage monitoring sequence through the voltage monitoring terminal, reading the current output frequency and output enable state, and when the output enable state is continuously effective and the current output frequency remains unchanged, dividing the output period into continuous output periods according to the current output frequency, and synchronously resampling the current monitoring sequence and the output voltage monitoring sequence with the starting point of each output period as a reference to obtain the periodic current monitoring sequence and the periodic output voltage monitoring sequence.

[0021] The beneficial effects of this invention are as follows: This invention uses the output cycle as the basic processing unit and maps the in-phase sampling points in different output cycles to the same output waveform position, so that there is a clear basis for comparison as to whether the current change is repeated across cycles, and reduces the influence of occasional electromagnetic interference and single-cycle random spikes on the wavelet threshold adjustment results.

[0022] This invention uses the proportion of the neighborhood of rapid voltage changes to correct the repetition rate of in-phase current abrupt changes, which can identify normal capacitive charging and discharging spikes that recur in each output cycle but are driven by rapid changes in output voltage. Without this correction, when the rated capacitive load is running stably, normal capacitive charging and discharging spikes will have a large retention demand due to cross-cycle repetition, which may still trigger the protection module to malfunction. Therefore, the correction can prevent normal capacitive charging and discharging spikes from being mistakenly retained due to cycle repetition.

[0023] This invention adjusts the wavelet threshold phase by phase based on the abnormal current mutation discrimination value. When the abnormal current mutation discrimination value is small, it enhances the suppression of normal capacitive charging and discharging peaks. When the abnormal current mutation discrimination value is large, it reduces the additional suppression intensity to retain continuous overcurrent mutations. Combined with the duration of continuous over-limit, it obtains the continuous overcurrent monitoring result, thus taking into account both the ability to suppress false alarms of normal capacitive transients and the timeliness of continuous overcurrent identification. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the broadband current source output monitoring data processing method of the present invention; Figure 2 This is a comparison chart of the noise reduction effects of existing fixed threshold processing and the adaptive threshold processing of this invention. Detailed Implementation

[0025] 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, not all, of the embodiments of the present invention. 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.

[0026] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] This invention discloses a method for processing broadband current source output monitoring data, referring to... Figure 1 This includes steps S001 to S005: S001, Obtain the periodic monitoring data set.

[0028] Specifically, the current monitoring voltage is acquired through the current monitoring terminal of the broadband current source, and the current monitoring ratio recorded in the parameters of the broadband current source device is read. The current monitoring ratio represents the current monitoring voltage corresponding to a unit current monitoring value. The current monitoring voltage at each sampling time is divided by the current monitoring ratio to obtain the current monitoring value at the corresponding sampling time. The current monitoring values ​​are arranged according to the sampling time to obtain the current monitoring sequence. At the same time, the output voltage monitoring sequence is acquired through the voltage monitoring terminal. The current monitoring terminal and the voltage monitoring terminal are connected to a data acquisition device that uses the same sampling clock, so that the corresponding sampling points in the current monitoring sequence and the output voltage monitoring sequence have the same sampling time.

[0029] Read the current output frequency and output enable state. When the output enable state is continuously effective and the current output frequency remains unchanged, take the reciprocal of the current output frequency as the duration of an output cycle. Divide the continuous output cycle starting from the first complete output cycle after the output enable state is continuously effective. Take the starting point of each output cycle as the phase zero point, and synchronously resample the current monitoring sequence and output voltage monitoring sequence in the corresponding output cycle to obtain a periodic current monitoring sequence and a periodic output voltage monitoring sequence with the same number of phase sampling points. Then, combine the periodic current monitoring sequence and the periodic output voltage monitoring sequence corresponding to the same output cycle into a periodic monitoring data group.

[0030] In this embodiment, the sampling frequency of the data acquisition device is set to 64 times the current output frequency, so that each output cycle contains 64 phase sampling points. When the current output frequency is 1kHz, one output cycle is 1ms. In other embodiments, the implementer can set the number of phase sampling points according to the shortest duration of the normal capacitive charge and discharge peak and the highest sampling frequency of the data acquisition device, and ensure that there are no less than 3 phase sampling points within the duration range of the normal capacitive charge and discharge peak, so as to avoid the normal capacitive charge and discharge peak falling into only one phase sampling point and thus being unable to determine its correspondence with the position of rapid change of output voltage.

[0031] The output period in which the output enable state is continuously valid, the current output frequency remains unchanged, and the data is complete is defined as the valid period. When the broadband current source is in DC output state, the output enable state is interrupted, the current output frequency changes, or the data does not constitute a complete output period, the corresponding data will not be included in the subsequent cross-cycle calculation. When the current output frequency changes, the output period is re-divided according to the changed current output frequency and the cross-cycle comparison buffer is cleared.

[0032] It should be noted that the same phase sampling points with the same phase number in different effective periods correspond to the same output waveform positions, thus establishing the basis for in-phase comparison; if the output period is not divided, frequency changes and sampling offsets will cause the same current mutation to fall into different sampling positions, resulting in distortion of the in-phase current mutation repetition rate.

[0033] In the subsequent parameter calibration of this invention, when no continuous overcurrent is applied, the broadband current source is connected to the rated capacitive load, and the current output frequency and output amplitude remain unchanged while keeping the output enable state continuously effective; when a continuous overcurrent is applied, the load is changed while keeping the current output frequency and output amplitude unchanged, so that the absolute value of the current monitoring value continuously exceeds the continuous overcurrent setting current in the broadband current source protection parameters, and the moment of changing the load is determined as the start moment of applying the continuous overcurrent.

[0034] S002. Determine the repetition rate of the same-phase current abrupt change at each phase sampling point.

[0035] It should be noted that fixed-threshold wavelet denoising typically sets a uniform threshold based solely on the noise level within the current output cycle, without utilizing the sustained characteristics of current abrupt changes across different output cycles. Therefore, it is difficult to distinguish between single-cycle high-frequency interference and effective abrupt changes caused by continuous overcurrent. Single-cycle high-frequency interference is usually sporadic, with its occurrence phase exhibiting strong randomness, and the probability of it repeating at the same phase position across multiple effective cycles is low. Continuous overcurrent, caused by load anomalies or abnormal output from broadband current sources, typically spans multiple output cycles, causing the current detail coefficients at the same phase position to continuously exceed the corresponding basic threshold. Therefore, this invention, based on output cycle division and synchronous resampling, counts the number of times the in-phase current detail coefficients exceed the basic threshold across multiple effective cycles to obtain the in-phase current abrupt change repetition rate.

[0036] Specifically, the method for obtaining the duration of normal capacitive charge and discharge peaks is as follows: when the broadband current source is connected to a rated capacitive load, the output enable state is continuously effective, and the current output frequency and output amplitude remain unchanged, a periodic current monitoring sequence of 10 effective cycles is collected; for each effective cycle, the median of the current monitoring values ​​within that effective cycle is used as a reference value, the maximum absolute difference between each current monitoring value and the reference value and its corresponding peak time are determined, and the sampling times when the absolute difference between the current monitoring values ​​and the reference value before and after the peak time reaches half of the maximum absolute difference are determined respectively, the time interval between the two sampling times is used as the duration of normal capacitive charge and discharge peaks for that effective cycle, and the median of the duration of normal capacitive charge and discharge peaks corresponding to the 10 effective cycles is used as the duration of normal capacitive charge and discharge peaks.

[0037] Specifically, the method for obtaining the target detail layer and current detail coefficients is as follows: Discrete wavelet decomposition is performed on each periodic current monitoring sequence. Among the multiple detail layers obtained from the wavelet decomposition, the detail layer that covers the duration of the normal capacitive charging and discharging peak at the corresponding time scale is selected as the target detail layer, and the coefficients of the target detail layer at each phase sampling point are used as the current detail coefficients. In this embodiment, a fourth-order vanishing moment wavelet basis is used. In other embodiments, other wavelet bases with tight support characteristics and capable of characterizing current abrupt changes can be used, and the target detail layer is determined according to the same time scale coverage principle.

[0038] Specifically, the method for obtaining the base threshold is as follows: Determine the median of all current detail coefficients within the current output cycle. Divide the median of the absolute values ​​of the differences between each current detail coefficient and the median by 0.6745 to obtain the noise level; 0.6745 is a constant for converting the absolute deviation of the median to the noise standard deviation. Multiply the noise level by the square root of twice the natural logarithm of the number of phase sampling points to obtain the base threshold. The larger the noise level or the number of phase sampling points, the larger the base threshold.

[0039] Specifically, the repetition rate of in-phase current abrupt changes satisfies the expression: ; In the formula, For the first The output cycle number The repetition rate of in-phase current abrupt change at each phase sampling point This includes the number of valid cycles, encompassing the current output cycle and all previous valid cycles. and The first The first effective period Current detail coefficients and basic thresholds at each phase sampling point It is an indicator function that takes the value 1 if the condition is true and 0 otherwise.

[0040] In the formula, the more times the absolute value of the in-phase current detail coefficient exceeds the basic threshold, the greater the repetition rate of the in-phase current mutation, indicating that the current mutation is more likely to continue across cycles; the larger the number of effective cycles, the less sensitive it is to single-cycle interference, but the longer the determination time of continuous overcurrent.

[0041] In this embodiment, the effective number of cycles is set to a range of 2 to 8. The number of false triggers when no continuous overcurrent is applied and the response time after applying continuous overcurrent are counted respectively. From the candidate values ​​whose response time does not exceed the maximum allowable continuous overcurrent time, the candidate value with the fewest false triggers is selected. If the number of false triggers is the same, the candidate value with the shortest response time is selected. In this embodiment, it is set to 4. The maximum allowable continuous overcurrent time is determined according to the longest protection response time and the overcurrent tolerance time allowed by the load in the broadband current source protection parameters. The smaller value between the longest protection response time and the overcurrent tolerance time is taken to ensure that the maximum allowable continuous overcurrent time does not exceed the safety limit of the broadband current source.

[0042] S003. Determine the abnormal current mutation discrimination value at each phase sampling point.

[0043] It should be noted that while the cross-cycle repetition of current surges can distinguish between single-cycle high-frequency interference and persistent current surges, when a broadband current source is connected to a rated capacitive load, rapid changes in output voltage drive the load to perform periodic charging and discharging, causing normal capacitive charging and discharging spikes to reappear at the same phase position in multiple output cycles. In this case, the normal capacitive charging and discharging spikes also exhibit a large in-phase current surge repetition rate. If this in-phase current surge repetition rate is directly used as the criterion for judging persistent overcurrent, normal capacitive charging and discharging spikes may still be retained as persistent anomalies. Considering that normal capacitive charging and discharging spikes usually concentrate near the location of rapid output voltage changes, while persistent overcurrents, in addition to appearing at the location of rapid output voltage changes, also extend to the location where the output voltage changes gradually. This invention further determines the proportion of the rapidly changing voltage neighborhood based on the output voltage change rate of the current sampling point and its adjacent phase sampling points, and uses this proportion of the rapidly changing voltage neighborhood to correct the in-phase current surge repetition rate, thus obtaining the abnormal current surge discrimination value.

[0044] Specifically, the method for obtaining the output voltage change rate of a phase sampling point is as follows: subtract the output voltage monitoring value of the previous phase sampling point from the output voltage monitoring value of a phase sampling point, and divide by the sampling time interval between the two phase sampling points to obtain the output voltage change rate of that phase sampling point.

[0045] Specifically, the method for obtaining the proportion of the voltage rapidly changing neighborhood is as follows: calculate the output voltage change rate of a phase sampling point and its two adjacent phase sampling points, count the number of phase sampling points whose absolute value of the output voltage change rate exceeds the preset voltage rapidly changing judgment value, and divide the number of phase sampling points by 3 to obtain the proportion of the voltage rapidly changing neighborhood. The proportion of the voltage rapidly changing neighborhood is determined by connecting the first and last phase sampling points in the order of the output cycle.

[0046] The preset voltage rapid change judgment value is calibrated based on the output voltage change rate when no continuous overcurrent is applied: At the same current output frequency and output amplitude, at least 100 complete output cycles are collected, and the median of the maximum absolute value of the output voltage change rate for each output cycle is taken. 50%, 60%, 70%, 80%, and 90% of this median are used as candidate values. From the candidate values ​​whose response time does not exceed the maximum allowable continuous overcurrent time, the candidate value with the fewest false triggers is selected. If there are multiple candidate values ​​with the fewest false triggers, the candidate value with the shortest response time is selected; in this embodiment, it is 70% of the median.

[0047] Specifically, the abnormal current sudden change discrimination value satisfies the expression: ; In the formula, , and The first The output cycle number The abnormal current mutation discrimination value, the repetition rate of the same phase current mutation, and the proportion of the neighborhood of rapid voltage change at each phase sampling point.

[0048] In the formula, the greater the repetition rate of in-phase current abrupt changes, the larger the abnormal current abrupt change discrimination value; the larger the proportion of the rapidly changing voltage neighborhood, the smaller the abnormal current abrupt change discrimination value. Therefore, current abrupt changes that repeat across cycles and are located away from the rapidly changing output voltage position will obtain a larger abnormal current abrupt change discrimination value.

[0049] It should be further explained that the proportion of the neighborhood of rapid voltage change is used to prevent periodic normal capacitive charging and discharging spikes from being mistakenly retained due to their large repetition rate of in-phase current abrupt changes. Its correction of the repetition rate of in-phase current abrupt changes does not change the screening result of single-cycle random interference.

[0050] S004. Determine the adaptive wavelet threshold for each phase sampling point and obtain the denoised current monitoring sequence.

[0051] It should be noted that the present invention adjusts the basic threshold according to the abnormal current mutation discrimination value, so that different phase sampling points are subjected to soft threshold processing adapted to the current mutation attributes, so as to take into account both the suppression of normal capacitive charging and discharging peaks and the retention of continuous overcurrent mutations.

[0052] Specifically, the adaptive wavelet threshold satisfies the expression: ; In the formula, and The first The output cycle number Adaptive wavelet threshold and abnormal current mutation discrimination value for each phase sampling point For the first The base threshold for each output cycle.

[0053] In the formula, the larger the abnormal current mutation discrimination value, the closer the adaptive wavelet threshold is to the basic threshold, and the weaker the suppression of continuous overcurrent mutation; the smaller the abnormal current mutation discrimination value, the larger the adaptive wavelet threshold, and the stronger the suppression of normal capacitive charging and discharging peaks.

[0054] For each phase sampling point, soft thresholding is performed on the current detail coefficients. Current detail coefficients whose absolute value is not greater than the corresponding adaptive wavelet threshold are set to zero, and current detail coefficients whose absolute value is greater than the corresponding adaptive wavelet threshold are shrunk towards zero to obtain the corresponding adaptive wavelet threshold value. The processed current detail coefficients are then used to perform wavelet reconstruction with the approximate coefficients that have not been thresholded to obtain the denoised current monitoring sequence of the current output period.

[0055] It should be noted that when the start of a continuous overcurrent coincides with a point of rapid change in output voltage, the abnormal current surge discrimination value at the start phase may decrease; however, continuous overcurrent will still generate recurring current surges across cycles at points where the output voltage changes more gradually, causing the corresponding phase sampling point to maintain a large abnormal current surge discrimination value. Therefore, the current surges generated by continuous overcurrent will not be completely eliminated in the soft thresholding process.

[0056] S005. Obtain the continuous overcurrent monitoring results and send protection commands.

[0057] Specifically, the continuous overcurrent setting current in the broadband current source protection parameters is read and used as the overcurrent judgment current. The absolute value of the denoised current monitoring sequence is compared with the overcurrent judgment current at each phase sampling point. The accumulated duration is calculated when the absolute value of the denoised current monitoring sequence continuously exceeds the overcurrent judgment current. The accumulated duration is reset to zero when the absolute value of the denoised current monitoring sequence at any phase sampling point does not exceed the overcurrent judgment current. The accumulated duration is the actual time and is not measured in fixed output cycle counts. After the current output frequency changes, the actual duration continues to be accumulated according to the changed sampling time.

[0058] The preset minimum duration period is set based on the maximum consecutive over-limit period of a normal capacitive charge / discharge spike and the protection response time. The maximum consecutive over-limit period of a normal capacitive charge / discharge spike refers to the number of output cycles covered by the longest duration during which the absolute value of the current monitoring value continuously exceeds the overcurrent judgment current when no continuous overcurrent is applied. In this embodiment, the single consecutive over-limit time of a normal capacitive charge / discharge spike is less than one output cycle; therefore, the maximum consecutive over-limit period of a normal capacitive charge / discharge spike is set to 1, and the preset minimum duration period is set to 3. Although normal capacitive charge / discharge spikes can recur in different output cycles, there are sampling intervals between adjacent spikes that do not exceed the overcurrent judgment current. The accumulated duration is reset to zero, thus preventing the duration corresponding to the preset minimum duration period from being reached. In other embodiments, implementers can set a preset minimum number of duration cycles based on the measured results under rated capacitive load. For example, when the load is sensitive to short-term overcurrent, the preset minimum number of duration cycles can be appropriately reduced, but the preset minimum number of duration cycles cannot be lower than the maximum number of consecutive over-limit cycles of normal capacitive charge and discharge peaks; when there are many occasional disturbances on site, the preset minimum number of duration cycles can be appropriately increased.

[0059] When the cumulative duration is not less than the duration corresponding to the preset minimum number of duration cycles, a continuous overcurrent monitoring result is generated, and a current limiting command, shutdown command, or alarm command is sent to the protection module according to the protection configuration of the broadband current source; when the cumulative duration is not greater than the duration corresponding to the preset minimum number of duration cycles, a monitoring result indicating that no continuous overcurrent was detected is generated, and continuous overcurrent protection is not triggered.

[0060] like Figure 2 As shown in the figure, this diagram compares the denoising effects of the existing fixed threshold processing and the adaptive threshold processing of this invention. The horizontal axis represents the monitoring time, the vertical axis represents the absolute value of the denoised current, and the horizontal dotted line represents the overcurrent judgment current. During the normal phase from 0 to 7.18 ms, the fixed threshold processing result still periodically shows residual peaks of approximately 8.1 to 8.6 A, repeatedly exceeding the judgment current, indicating that a uniform threshold is insufficient to eliminate normal capacitive charging and discharging peaks. The processing result of this invention suppresses most normal peaks below the judgment current. After entering the continuous overcurrent phase from 7.18 to 10.72 ms, the processing result of this invention rapidly rises to approximately 8.8 A and remains above the judgment current, while retaining periodic abrupt peaks. This is because this invention utilizes the repetition rate of in-phase current abrupt changes and the proportion of the neighborhood of rapid voltage changes to adjust the wavelet threshold, enhancing suppression at normal peak positions and reducing additional suppression intensity in the overcurrent region that spans multiple cycles and is detached from the position of rapid voltage changes. This demonstrates that this invention can both reduce the risk of false triggering caused by normal transients and retain continuous overcurrent characteristics, thereby improving the accuracy and timeliness of overcurrent monitoring.

Claims

1. A method for processing monitoring data from a broadband current source output, characterized in that, include: S001. Obtain the synchronous monitoring data of the broadband current source, read the current output frequency and output enable status, divide the synchronous monitoring data into continuous output cycles, and perform synchronous resampling to obtain the periodic current monitoring sequence and the periodic output voltage monitoring sequence. S002. Calibrate the duration of the normal capacitive charge and discharge peaks. Based on the duration of the normal capacitive charge and discharge peaks and the wavelet decomposition results of the periodic current monitoring sequence, obtain the current detail coefficients of each phase sampling point. Estimate the noise level of the detail layer corresponding to each output cycle based on the current detail coefficients and determine the basic threshold. Determine all effective cycles corresponding to the current output cycle. Count the number of times the in-phase current detail coefficients exceed the basic threshold in the current output cycle and each effective cycle before it, and determine the in-phase current abrupt repetition rate. S003. Based on the output voltage change rate of the phase sampling point, determine the proportion of the neighborhood of rapid voltage change, and combine it with the repetition rate of the current change in the same phase to obtain the abnormal current change discrimination value. S004. Adjust the basic threshold according to the abnormal current mutation discrimination value to obtain the adaptive wavelet threshold, and use the adaptive wavelet threshold to obtain the denoised current monitoring sequence. S005. Based on the duration of the continuous overcurrent judgment current exceeding the noise-reduced current monitoring sequence, obtain the continuous overcurrent monitoring result, and send a protection command to the broadband current source according to the continuous overcurrent monitoring result.

2. The broadband current source output monitoring data processing method according to claim 1, characterized in that, The calibration of the normal capacitive charge and discharge peak duration width includes: when the broadband current source is connected to the rated capacitive load, the output enable state is continuously effective, and the current output frequency and output amplitude remain unchanged, acquiring a periodic current monitoring sequence for several effective periods; for each effective period, the median of the current monitoring values ​​within the effective period is used as the reference value, the maximum absolute difference between each current monitoring value and the reference value and its corresponding peak time are determined, the sampling times before and after the peak time when the absolute difference between the current monitoring values ​​and the reference value reaches half of the maximum absolute difference are determined, the time interval between the two sampling times is used as the normal capacitive charge and discharge peak duration width of the effective period, and the median of the normal capacitive charge and discharge peak duration widths corresponding to several effective periods is used as the normal capacitive charge and discharge peak duration width.

3. The broadband current source output monitoring data processing method according to claim 1, characterized in that, The method for obtaining the current detail coefficients is as follows: perform discrete wavelet decomposition on each periodic current monitoring sequence, select the detail layer that covers the duration of the normal capacitive charging and discharging peak at the corresponding time scale from the multiple detail layers obtained by wavelet decomposition as the target detail layer, and use the coefficients of the target detail layer at each phase sampling point as the current detail coefficients.

4. The broadband current source output monitoring data processing method according to claim 1, characterized in that, The method for obtaining the basic threshold is as follows: determine the median of all current detail coefficients in the current output cycle, divide the median of the absolute value of the difference between each current detail coefficient and the median by 0.6745, where 0.6745 is a constant for converting the absolute deviation of the median into the noise standard deviation, and obtain the noise level; multiply the noise level by the square root of twice the natural logarithm of the number of phase sampling points, and use the result as the basic threshold.

5. The broadband current source output monitoring data processing method according to claim 1, characterized in that, The repetition rate of the in-phase current abrupt change satisfies the expression: ； In the formula, For the first The output cycle number The repetition rate of in-phase current abrupt change at each phase sampling point This includes the number of valid cycles, encompassing the current output cycle and all previous valid cycles. and The first The first effective period Current detail coefficients and basic thresholds at each phase sampling point It is an indicator function that takes the value 1 if the condition is true and 0 otherwise.

6. The broadband current source output monitoring data processing method according to claim 1, characterized in that, The method for obtaining the proportion of the voltage rapid change neighborhood is as follows: subtract the output voltage monitoring value of the previous phase sampling point from the output voltage monitoring value of a phase sampling point, and divide by the sampling time interval between the two phase sampling points to obtain the output voltage change rate of the phase sampling point. Calculate the output voltage change rate of a phase sampling point and its two adjacent phase sampling points. Count the number of phase sampling points whose absolute value of the output voltage change rate exceeds the preset voltage rapid change judgment value, and divide the number of phase sampling points by 3. Use the result as the proportion of the voltage rapid change neighborhood. The proportion of the voltage rapid change neighborhood is determined by connecting the first and last phase sampling points in the output cycle.

7. The broadband current source output monitoring data processing method according to claim 1, characterized in that, The abnormal current sudden change discrimination value satisfies the expression: ； In the formula, , and The first The output cycle number The abnormal current mutation discrimination value, the repetition rate of the same phase current mutation, and the proportion of the neighborhood of rapid voltage change at each phase sampling point.

8. The broadband current source output monitoring data processing method according to claim 1, characterized in that, The adaptive wavelet threshold satisfies the expression: ； In the formula, and The first The output cycle number Adaptive wavelet threshold and abnormal current mutation discrimination value for each phase sampling point For the first The base threshold for each output cycle.

9. The broadband current source output monitoring data processing method according to claim 1, characterized in that, Obtaining the continuous overcurrent monitoring result includes: reading the continuous overcurrent setting current in the broadband current source protection parameters and using the continuous overcurrent setting current as the overcurrent judgment current; when the absolute value of the denoised current monitoring sequence continuously exceeds the overcurrent judgment current, accumulating its duration; when the absolute value of the denoised current monitoring sequence does not exceed the overcurrent judgment current, clearing the accumulated duration to zero; in response to the accumulated duration being not less than the duration corresponding to the preset minimum number of duration cycles, generating the continuous overcurrent monitoring result, and sending a current limiting command, shutdown command, or alarm command to the protection module according to the protection configuration of the broadband current source.

10. The broadband current source output monitoring data processing method according to claim 1, characterized in that, The process of obtaining the periodic current monitoring sequence and the periodic output voltage monitoring sequence includes: acquiring the current monitoring voltage through the current monitoring terminal, converting the current monitoring voltage into a current monitoring sequence according to the current monitoring ratio corresponding to the broadband current source, acquiring the output voltage monitoring sequence through the voltage monitoring terminal, reading the current output frequency and output enable state, and when the output enable state is continuously effective and the current output frequency remains unchanged, dividing the output period into consecutive periods according to the current output frequency, and synchronously resampling the current monitoring sequence and the output voltage monitoring sequence with the starting point of each output period as a reference to obtain the periodic current monitoring sequence and the periodic output voltage monitoring sequence.

Images