A multi-path controllable power converter and a filtering method for implementing filtering

Abstract

The present application relates to the technical field of data processing, in particular to a kind of multi-way controllable power converter for realizing filter processing and filter processing method, comprising: obtaining the signal delay time and voltage signal data of multi-way controllable power converter in current monitoring time period and historical monitoring time period;According to the peak value distribution and frequency distribution of voltage signal data in each historical monitoring time period, analyze the noise interference degree, obtain synchronization characteristic index in combination with the difference of noise interference degree and pulse width parameter variation trend;According to the signal delay time of current monitoring time period and historical monitoring time period, obtain filter weight in combination with synchronization characteristic index;According to filter weight, the voltage signal data in current monitoring time period is decomposed, and filter is carried out in combination with the gauss characteristic of decomposed signal component, obtains the voltage signal data after filter processing.The filter processing result of the present application has good processing effect, guarantees the stability of power output.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a multi-path controllable power converter and a filtering processing method for realizing filtering processing. BACKGROUND

[0002] In the operation process of the multi-path controllable power converter, pulse width modulation (PWM) is used as the core modulation method to achieve accurate control of the output voltage and current by continuously adjusting the on and off time of the switching device. However, with multiple accumulations of modulation operations, the high-frequency noise, harmonic components and transient disturbances generated by each switching action will gradually superimpose, resulting in a complex cumulative effect of the mixed interference components in the power output signal. Not only can it exacerbate the current ripple, but also continuously affects the stability and cleanliness of the output. With the rapid development of electronic devices and power electronics technology, the quality and stability of the power supply become particularly important, especially in high-frequency and high-power application fields. Therefore, efficient filtering processing has become a key requirement to improve the performance of the power supply.

[0003] In the filtering process of the existing multi-path controllable power converter, fixed filtering parameters are used, which cannot effectively suppress high-frequency noise and transient disturbances in the circuit, resulting in a significant reduction in current ripple. This makes the filtering processing result poor, which further affects the stability of the power output. SUMMARY

[0004] In order to solve the technical problem that the existing method uses fixed filtering parameters in the filtering process, resulting in poor filtering processing results, the purpose of the present application is to provide a multi-path controllable power converter and a filtering processing method for realizing filtering processing, and the technical solution is as follows:

[0005] In the first aspect, the present application provides a filtering processing method of a multi-path controllable power converter for realizing filtering processing, comprising:

[0006] Obtaining signal delay time and voltage signal data of the multi-path controllable power converter in the current monitoring time period and the historical monitoring time period, each historical modulation operation of the multi-path controllable power converter corresponding to a historical monitoring time period and a pulse width parameter;

[0007] According to the peak value distribution and frequency distribution of the voltage signal data in each historical monitoring time period, analyzing the noise interference degree of each historical monitoring time period, combining the difference between the noise interference degree and the pulse width parameter change trend, obtaining the synchronization characteristic index of the modulation operation and the noise interference;

[0008] Based on the signal delay time of the current monitoring period and all historical monitoring periods, and combined with the aforementioned synchronization characteristic index, the filtering weights of the multi-channel controllable power converter are obtained.

[0009] Based on the filtering weights, the voltage signal data within the current monitoring time period is decomposed, and filtered by combining the Gaussian characteristics of the decomposed signal components to obtain the filtered voltage signal data.

[0010] Preferably, the step of analyzing the noise interference level of each historical monitoring period based on the peak and frequency distribution of voltage signal data within each historical monitoring period specifically includes:

[0011] Based on the difference in the degree of characteristic between adjacent peak signal segments within each historical monitoring period, the first signal noise index for each historical monitoring period is obtained.

[0012] Based on the frequency distribution of the spectral characteristics of the voltage signal data within each historical monitoring period, a second signal noise index is obtained for each historical monitoring period.

[0013] The product of the first signal noise index and the second signal noise index is used as the noise interference level for each historical monitoring period.

[0014] Preferably, obtaining the first signal noise index for each historical monitoring time period based on the difference in the degree of characteristic between adjacent peak signal segments within each historical monitoring time period specifically includes:

[0015] Within any historical monitoring period, the signal difference between any two adjacent peak signal points is obtained. The peak signal point corresponding to the maximum value of the signal difference is used as the segmentation signal point. All peak signal points whose peak signal points are greater than or equal to the segmentation signal points are selected to form a characteristic peak point sequence.

[0016] Based on the difference between the peak widths of every two adjacent characteristic peak points in the characteristic peak point sequence, the peak difference coefficient is determined; the maximum value of the time interval between adjacent characteristic peak points in the characteristic peak point sequence is taken as the maximum time interval.

[0017] The ratio of the peak difference coefficient to the maximum time interval is normalized to obtain the first signal noise index corresponding to any historical monitoring time period.

[0018] Preferably, obtaining the second signal noise index for each historical monitoring period based on the frequency distribution of the spectral characteristics of the voltage signal data within each historical monitoring period specifically includes:

[0019] A Fourier transform is performed on the voltage signal data within any historical monitoring period to obtain a spectrum, and the frequency value corresponding to the largest amplitude in the spectrum is taken as the main frequency.

[0020] The frequency value of each harmonic is determined based on integer multiples of the dominant frequency; curve fitting is performed on the frequency value and amplitude of each harmonic in the spectrum diagram, and the slope value of each harmonic is obtained on the fitted curve; based on the slope value of each harmonic, all harmonics are screened to obtain suspected anomalous harmonics.

[0021] The mean difference between the frequency value of each suspected abnormal harmonic and the maximum frequency value in the spectrum is negatively correlated to obtain the second signal noise index for any historical monitoring time period.

[0022] Preferably, the step of analyzing the noise interference level of each historical monitoring period based on the peak and frequency distribution of voltage signal data within each historical monitoring period, and combining the differences in noise interference level and pulse width parameter variation trends to obtain synchronization characteristic indicators of modulation operation and noise interference, specifically includes:

[0023] The first characteristic curve is obtained by curve fitting with the time point of the historical modulation operation corresponding to each historical monitoring period as the horizontal axis and the noise interference level of each historical monitoring period as the vertical axis. The second characteristic curve is obtained by curve fitting with the pulse width reference of the modulation operation corresponding to each historical monitoring period as the vertical axis.

[0024] Obtain the first slope value of each data point on the first feature curve and the second slope value of each data point on the second feature curve;

[0025] By performing negative correlation processing on the mean of the differences between the first slope value and the second slope value at all the same time points, the synchronization characteristic index of modulation operation and noise interference is obtained.

[0026] Preferably, the step of obtaining the filtering weights of the multi-channel controllable power converter based on the signal delay times of the current monitoring time period and all historical monitoring time periods, combined with the synchronization characteristic index, specifically includes:

[0027] Calculate the ratio of the signal delay time between the current monitoring time period and each historical monitoring time period, normalize the product of the mean of all ratios and the synchronization characteristic index, and obtain the filtering weight of the multi-channel controllable power converter.

[0028] Preferably, the step of decomposing the voltage signal data within the current monitoring time period according to the filtering weights, and filtering it by combining the Gaussian characteristics of the decomposed signal components to obtain the filtered voltage signal data specifically includes:

[0029] Using an objective function, the feature decomposition process of voltage signal data within the current monitoring time period is iteratively performed to obtain each feature component; wherein, the objective function is constructed based on the degree of non-Gaussianity of each feature component and the filtering weight; when the objective function converges to its maximum, the optimal component result of feature decomposition is obtained;

[0030] In the optimal decomposition result, the voltage signal data within the current monitoring time period is filtered based on the signal distribution of the last feature component in the decomposition order to obtain the filtered voltage signal data.

[0031] Preferably, the step of constructing the objective function based on the degree of non-Gaussianity of each feature component and the filtering weights specifically includes:

[0032] The absolute value of the difference between the kurtosis of each feature component and the kurtosis of the standard normal distribution is used as the non-Gaussianity index of each feature component;

[0033] The last feature component in the decomposition order of each iteration is taken as the target component, and the other feature components are taken as the control components.

[0034] The product of the ratio between the mean of the non-Gaussianity indices of all control components and the non-Gaussianity indices of the target component in each iteration, and the filtering weights, is used as the objective function for each iteration.

[0035] Preferably, in the optimal decomposition result, filtering the voltage signal data within the current monitoring time period based on the signal distribution of the last feature component in the decomposition order specifically includes:

[0036] In the optimal decomposition result, the standard deviation of the signal data of the last feature component in the decomposition order is negatively correlated and normalized to obtain the adjustment coefficient.

[0037] The product of the adjustment coefficient and the preset cutoff frequency of the low-pass filter is rounded down to obtain the adjusted cutoff frequency for filtering, and the voltage signal data within the current monitoring period is filtered.

[0038] In a second aspect, the present invention provides a multi-channel controllable power converter for implementing filtering processing, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the computer program is executed by the processor, it implements the steps of a filtering processing method for a multi-channel controllable power converter for implementing filtering processing.

[0039] The embodiments of the present invention have at least the following beneficial effects:

[0040] This invention first acquires the characteristic parameters corresponding to historical modulation operations and the parameters of the current monitoring time period, providing a data foundation for subsequent analysis of noise interference in voltage signal data within the current monitoring time period based on historical modulation operations. Then, firstly, it analyzes the time-domain peak distribution and frequency-domain frequency distribution corresponding to historical modulation operations to evaluate the noise interference level within the corresponding historical monitoring time period. Secondly, it analyzes the differences in the noise interference level and pulse width parameter variation trends to measure the synchronization characteristic indicators of modulation operations and noise interference, specifically quantifying whether the noise interference is caused by the modulation operation. Furthermore, based on the synchronization characteristic indicators, it further considers the signal delay time feature and adaptively determines the filtering weights of the multi-channel controllable power converter. Finally, combined with the filtering weights, it accurately separates high-frequency interference components from the voltage signal data while avoiding modulation signal distortion caused by over-separation. The filtering method of this invention can effectively remove interference components, resulting in better filtering performance and ensuring the stability of the power output. Attached Figure Description

[0041] To more clearly illustrate the technical solutions and advantages 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a flowchart of the filtering process of a multi-channel controllable power converter that implements filtering processing, provided by the present invention.

[0043] Figure 2 This is a schematic diagram of the distribution of voltage signal data during the current monitoring period provided by the present invention;

[0044] Figure 3 This is a flowchart of the steps of the method for obtaining the noise interference level provided by the present invention;

[0045] Figure 4 This is a schematic diagram of the power spectral density curve of voltage signal data within a historical monitoring period provided by the present invention;

[0046] Figure 5 This is a flowchart of the steps for obtaining synchronization feature indicators provided by the present invention;

[0047] Figure 6 This is a flowchart of the steps for filtering voltage signal data within the current monitoring time period provided by the present invention;

[0048] Figure 7This is a schematic diagram showing the distribution of filtered voltage signal data within the current monitoring time period provided by the present invention. Detailed Implementation

[0049] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a multi-channel controllable power converter and filtering method for implementing filtering according to the present invention.

[0050] Before introducing the specific solutions provided in the embodiments of this application, some terms used in this application will be explained to facilitate understanding by those skilled in the art, and will not be used to limit the scope of this application.

[0051] The basic working principle of a multi-channel controllable power converter is to control the input voltage conversion through one or more switching devices (such as MOSFETs) to achieve stable output of multiple output voltages. Its core lies in utilizing components such as transformers or inductors for energy transfer and adjusting the amplitude and stability of the output voltage through pulse width modulation (PWM) control technology. The ultimate goal of a multi-channel controllable power converter is to provide multiple outputs with different voltages and currents to meet the needs of different loads.

[0052] In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0054] The following description, in conjunction with the accompanying drawings, details a specific scheme for a multi-channel controllable power converter and filtering method for implementing filtering processing provided by the present invention.

[0055] Please see Figure 1 The diagram illustrates a flowchart of a filtering method for a multi-channel controllable power converter, according to an embodiment of the present invention. The method includes the following steps:

[0056] Step S100: Obtain the signal delay time and voltage signal data of the multi-channel controllable power converter in the current monitoring time period and the historical monitoring time period. Each historical modulation operation of the multi-channel controllable power converter corresponds to a historical monitoring time period and pulse width parameter.

[0057] First, the voltage signal of the power converter is monitored. An empirically determined monitoring time period is set. During the continuous monitoring of the voltage signals of the multiple controllable power converters, the filtering analysis of this invention is performed once after each monitoring time period.

[0058] Meanwhile, considering that with the accumulation of pulse width modulation operations, the high-frequency noise, harmonic components and transient interference generated by each switching action will gradually superimpose, resulting in a complex cumulative effect of the interference components mixed in the power output signal. This may not only exacerbate the current ripple, but also have a continuous impact on the stability and cleanliness of the output.

[0059] Therefore, to achieve optimal filtering under current operating conditions, it is necessary to fully leverage the patterns and characteristics accumulated from historical adjustments and accurately analyze the sources, intensity, and frequency distribution of the current accumulated noise. This requires identifying both inherent interference components generated by multiple modulation superpositions and new interference introduced by current load changes or the external environment. By correlating the correlation patterns in historical data with the real-time characteristics of the current noise, a reliable basis for optimizing filtering parameters can be provided. This effectively suppresses accumulated noise while avoiding over-filtering that could negatively impact the normal modulation response of the power supply, ultimately achieving a more targeted and efficient filtering effect.

[0060] Based on this, when implementing the filtering method of the present invention, the power converter has already performed a certain number of pulse width modulation operations, and can obtain various key modulation parameters under historical modulation operations. The historical modulation operations refer to pulse width modulation operations performed at historical times, and the number of historical modulation operations can be at least 5, which can be set by the implementer according to the specific implementation scenario.

[0061] Specifically, this embodiment uses the analysis process of filtering and adjusting data with a time interval of at least a preset time length after the current modulation operation as an example. The preset time length can be set to 10 minutes, and the time period after the current modulation operation is taken as the current monitoring time period. The filtering process of the voltage signal within the current monitoring time period is used as an example for explanation. Under each historical modulation operation, the time point and pulse width parameters of each historical modulation operation are recorded. The time interval between each historical modulation operation and the next adjacent historical modulation operation is taken as the historical monitoring time period corresponding to each historical modulation operation, which is used to monitor the data changes of the power converter after the modulation operation occurs.

[0062] It should be noted that the pulse width value after modulation operation, that is, the result of the adjustment of the on-time of the switching device in each pulse width modulation (PWM) operation, is the core parameter of the modulation operation, and the specific method of obtaining it is not restricted here.

[0063] Furthermore, the signal delay time and voltage signal data of the multi-channel controllable power converter during the current monitoring period and historical monitoring periods are acquired. The method for acquiring the voltage signal data is a well-known technique and will not be described in detail here. Figure 2 As shown, the voltage signal data for the current monitoring period is represented by a waveform graph.

[0064] In a multi-channel controllable power converter, the time required for voltage modulation can be predicted based on the actual pulse width modulation mechanism. Let's assume the predicted time for one modulation process is denoted as... Let t be the time required for the voltage to reach the target value after adjustment. Then, the signal delay time corresponding to one modulation operation is... In other words, the signal delay time for each historical modulation operation refers to the time difference between the actual time and the theoretical time for the voltage to reach the expected value under the corresponding historical modulation operation. Similarly, the signal delay time for the current monitoring period is obtained using the same method, referring to the time difference between the actual time and the expected time after the current modulation operation. In summary, the signal delay time characterizes the response delay of the modulation operation.

[0065] It should be noted that, This refers to the actual time required from the moment the modulation operation is triggered until the output voltage stabilizes and reaches the target value. It can be obtained by recording the time interval between the moment the modulation operation is triggered and the voltage reaching the target value. t This refers to the theoretical time required from the moment the modulation operation is triggered until the output voltage stabilizes and reaches the target value. It is derived based on an ideal circuit model (hardware parameters or simulation results) and is used to compare with the actual measured adjustment time t to calculate the delay deviation. This is a well-known technique in the field and will not be discussed further here.

[0066] Step S200: Based on the peak distribution and frequency distribution of voltage signal data within each historical monitoring period, analyze the noise interference level of each historical monitoring period. Combine the differences in noise interference level and pulse width parameter change trends to obtain the synchronization characteristic index of modulation operation and noise interference.

[0067] When performing feature analysis on the characteristic data corresponding to historical modulation operations, two main aspects are involved. First, it's necessary to analyze whether the voltage signal data within the historical monitoring period exhibits high-frequency noise interference characteristics after the historical modulation operation. This involves analyzing the harmonic variation characteristics of the voltage signal data within the time domain waveform and spectrum information to identify high-frequency noise interference characteristics within the historical monitoring period. Second, it's necessary to analyze whether the identified noise interference sources are related to the modulation operation behavior of the power converter, providing a solid theoretical foundation for subsequent filtering and circuit performance optimization measures. Based on this, and combining the analysis results from both aspects, the synchronization characteristic indicators of modulation operation and noise interference are quantified.

[0068] As a concrete example, such as Figure 3 As shown, in the first aspect, the method for obtaining the noise interference level can be implemented by steps S201 to S203.

[0069] Step S201: Based on the difference in the degree of characteristic between adjacent peak signal segments within each historical monitoring time period, obtain the first signal noise index for each historical monitoring time period.

[0070] The analysis process of the first signal noise index reflects the process of analyzing the time-domain waveform characteristics of voltage signal data within the historical monitoring period. When the waveform of the voltage signal data shows irregular spikes and rapid vibrations, it indicates that the voltage signal data may be a characteristic manifestation of high-frequency noise.

[0071] The first step is to obtain the signal difference between any two adjacent peak signal points within any historical monitoring period, and take the peak signal point corresponding to the maximum value of the signal difference as the segmentation signal point; then select all peak signal points that are greater than or equal to the segmentation signal point to form a characteristic peak point sequence.

[0072] This involves employing a peak point detection algorithm to detect voltage signal data over a historical monitoring period, acquiring each peak signal point within that period, and arranging all peak signal points in ascending order of voltage signal data value. A signal difference exists between every two adjacent peak signal points. It should be understood that the signal difference refers to the difference in voltage signal data between a peak signal point with a longer time sequence and one with a shorter time sequence; a larger signal difference indicates a greater variation between two adjacent peaks. It should be noted that the method for acquiring signal peak points is a known technique, and in other embodiments, implementers can use different methods depending on the specific implementation scenario.

[0073] All peak signal points whose peak signal points are greater than or equal to the dividing signal point are still arranged in ascending order of voltage signal data values ​​to form a characteristic peak point sequence. This can be understood as follows: using the dividing signal point as the dividing point, all peak signal points to the right of the dividing point belong to the more prominent peak portion, forming a characteristic peak point sequence for prominent signal analysis. It should be understood that the aforementioned peak signal point size refers to the value of the ordinate data of the peak signal point, which is also the value of the voltage signal data.

[0074] The second step is to determine the peak difference coefficient based on the difference between the peak widths of every two adjacent characteristic peak points in the characteristic peak point sequence; to take the maximum value of the time interval between adjacent characteristic peak points in the characteristic peak point sequence as the maximum time interval; and to normalize the ratio of the peak difference coefficient to the maximum time interval to obtain the first signal noise index corresponding to any historical monitoring time period.

[0075] As a concrete example, taking the nth historical monitoring period as an example, the calculation formula for the first signal-to-noise index corresponding to the nth historical monitoring period can be expressed as:

[0076] ;

[0077] in, This represents the first signal noise index during the nth historical monitoring period. This indicates the number of peak signal points in the characteristic peak point sequence. This represents the peak width corresponding to the i-th peak signal point in the characteristic peak point sequence of the nth historical monitoring time period. It represents the peak width corresponding to the (i+1)th peak signal point in the characteristic peak point sequence of the nth historical monitoring time period. This represents the maximum time interval between two adjacent peak signal points within the nth historical monitoring period. For example, the minimization normalization method can be used, and there are no restrictions here.

[0078] The peak width reflects the peak shape characteristics of each peak signal point. This reflects the width difference between the more prominent peak segments, that is, the difference in shape characteristics. This is achieved using ratios. This characterizes the waveform with irregular spikes and rapid vibrations. The larger the value, the greater the degree of high-frequency noise interference in the voltage signal data during the historical monitoring period. In this case, the value of the first signal noise is also larger.

[0079] It should be noted that in this embodiment, the time interval between the valley points adjacent to each peak signal point is used as the peak width of the corresponding peak signal point. In other embodiments, the implementer may also use the half-height peak width as the peak width of each peak signal point, which will not be discussed further here.

[0080] Step S202: Based on the frequency distribution of the spectral characteristics of the voltage signal data within each historical monitoring period, obtain the second signal noise index for each historical monitoring period.

[0081] Spikes and noise fluctuations in the time-domain waveform can be caused by a variety of factors, not just high-frequency noise. For example, transient response of the power converter, load changes, switching frequency, and mutual interference between systems can all lead to waveform variations. The second signal noise index reflects the process of analyzing the frequency domain characteristics of voltage signal data over a historical monitoring period.

[0082] The first step is to perform a Fourier transform on the voltage signal data within any historical monitoring period to obtain a spectrum. The frequency value corresponding to the largest amplitude in the spectrum is then taken as the dominant frequency. Based on integer multiples of the dominant frequency, the frequency value of each harmonic is determined.

[0083] Specifically, the output voltage during the historical monitoring period is sampled using Fast Fourier Transform (FFT) to generate a spectrum graph. The horizontal axis of the spectrum graph represents frequency, and the vertical axis represents amplitude. Then, the maximum amplitude value among all amplitude values ​​is counted, and the frequency value corresponding to the maximum amplitude value is recorded as the dominant frequency of the voltage signal data during the historical monitoring period. The dominant frequency represents the core operating frequency of the power converter output voltage signal during the historical monitoring period. It is the fundamental frequency of the signal, and other harmonic components are generated based on it.

[0084] It should be understood that harmonics are integer multiples of the dominant frequency, for example, when the dominant frequency is... The frequencies corresponding to the harmonics are respectively , , Harmonics are the main frequency components in a signal besides the fundamental frequency. High-frequency noise often exists in the form of harmonic superposition. From the spectrum, all data points that satisfy the condition of having frequencies that are integer multiples of the fundamental frequency are selected; these data points are the harmonic components. Simultaneously, the frequency of each harmonic is recorded (e.g., ...). , ) and the corresponding amplitude (vertical axis of the spectrum).

[0085] The second step is to perform curve fitting on the spectrum graph based on the frequency value and amplitude of each harmonic, and obtain the slope value of each harmonic on the fitted curve; based on the slope value of each harmonic, all harmonics are screened to obtain suspected abnormal harmonics.

[0086] Power converters also generate harmonics (such as integer multiples of the fundamental frequency) during normal operation, but high-frequency noise can cause abnormal growth (non-natural distribution) of harmonic amplitude in specific high-frequency bands. A single peak frequency may be a random fluctuation, but the trend of a rapid increase in harmonic amplitude with increasing frequency is more indicative of continuous high-frequency noise interference. In other words, noise often manifests as continuous interference in the high-frequency band.

[0087] Based on this characteristic, by plotting the frequency value and corresponding amplitude of each harmonic on the x-axis and the amplitude value on the y-axis, a linear fit using the least squares method can most directly reflect the trend of amplitude change with harmonic frequency. Specifically, the slope value of each harmonic is calculated on the fitted curve, reflecting the trend of frequency and amplitude change at each harmonic position. It should be noted that the method for calculating the slope of the data points on the curve is a well-known technique and will not be elaborated upon here. Curve fitting can directly reflect the overall trend of change.

[0088] Suspected anomalous harmonics are those harmonics whose amplitude increases rapidly with increasing frequency; these harmonics may be related to noise interference. The specific screening method is the same as the method for obtaining the characteristic peak point sequence in step S201.

[0089] Specifically, harmonics with slope values ​​greater than 0 are selected from all harmonics and arranged in ascending order of slope value. Following this arrangement, the difference between each slope value and its preceding neighbor is calculated. The slope value corresponding to the maximum difference is used as a dividing point. Harmonics corresponding to all slope values ​​to the right of this dividing point are considered suspected anomalous harmonics. This selection process reflects the process of identifying harmonics with significant variation trends from all harmonics for subsequent frequency difference characteristic analysis.

[0090] like Figure 4 The power spectral density curve of voltage signal data within a historical monitoring period is shown. Figure 4 There are obvious power spectrum peaks around 2000Hz, 3000Hz, and 4000Hz, which correspond to the concentrated frequency band of high-frequency noise, and it is highly likely that they are abnormal harmonics.

[0091] The third step is to perform negative correlation processing on the mean difference between the frequency value of each suspected abnormal harmonic and the maximum frequency value in the spectrum to obtain the second signal noise index for any historical monitoring time period.

[0092] Furthermore, by combining the frequencies corresponding to the selected high positive slopes with the maximum frequency value, it is determined whether the frequencies of these harmonics are concentrated in the high-frequency band, thereby quantifying the interference intensity of high-frequency noise.

[0093] Specifically, for the nth historical monitoring period, the absolute value of the difference between the frequency value of each suspected anomalous harmonic and the maximum frequency value in the corresponding spectrum is calculated. The mean of the absolute values ​​of the differences corresponding to all suspected anomalous harmonics is then negatively correlated to obtain the second signal-noise index for the nth historical monitoring period. A negative exponential function can be used in this process. The negative correlation is processed in the form of [formula / method]. This indicates the parameter to be processed.

[0094] The smaller the difference between the frequency value of each suspected anomalous harmonic and the maximum frequency value in the corresponding spectrum, the closer the frequency value of the suspected anomalous harmonic is to the maximum frequency value. Furthermore, when the average value of the difference is larger, it indicates that the screened suspected anomalous harmonics are all in the high-frequency band, and the harmonics increase significantly in the high-frequency band. This phenomenon may be related to high-frequency noise interference, and the corresponding value of the second signal noise index is larger.

[0095] Step S203: The product between the first signal noise index and the second signal noise index is used as the noise interference level for each historical monitoring period.

[0096] The first signal noise index reflects the potential noise interference from the time-domain waveform characteristics of the voltage signal data within the historical monitoring period, while the second signal noise index reflects the potential noise interference from the frequency-domain variation trend characteristics of the voltage signal data within the historical monitoring period. Combining the results of the two characteristic analyses can provide a relatively comprehensive characterization of the noise interference level within each historical monitoring period.

[0097] Furthermore, the second aspect involves analyzing whether the noise interference in the voltage signal data corresponds to the modulation operation. If significant differences in noise characteristics appear under different modulation operations, it indicates that the noise interference may originate from the modulation operation behavior of the power converter. This provides a data basis for subsequent filtering analysis; that is, if there is a strong correlation between modulation operation behavior and noise interference, attention should be paid to the filtering scale in subsequent filtering operations.

[0098] Based on this characteristic, such as Figure 5 As shown, the method for obtaining synchronization feature indicators can be implemented by steps S204 to S207.

[0099] Step S204: Using the time point of the historical modulation operation corresponding to each historical monitoring period as the horizontal axis and the noise interference level of each historical monitoring period as the vertical axis, curve fitting is performed to obtain the first feature curve.

[0100] Step S205: Using the time point of the historical modulation operation corresponding to each historical monitoring time period as the horizontal axis and the pulse width reference of the modulation operation corresponding to each historical monitoring time period as the vertical axis, curve fitting is performed to obtain the second feature curve.

[0101] It should be noted that both the first and second characteristic curves can be linearly fitted using the least squares method.

[0102] Step S206: Obtain the first slope value of each data point on the first feature curve and the second slope value of each data point on the second feature curve.

[0103] It should be noted that the first slope value refers to the slope value corresponding to each time point on the first characteristic curve, reflecting the changing trend of the degree of noise interference on the voltage signal data corresponding to each historical adjustment operation over time. The second slope value refers to the slope value corresponding to each time point on the second characteristic curve, reflecting the changing trend of the pulse width parameter for each historical modulation operation over time. The calculation method of the slope value is a well-known technique and will not be elaborated further here.

[0104] Step S207: Perform negative correlation processing on the mean of the differences between the first slope value and the second slope value at all the same time points to obtain the synchronization characteristic index of modulation operation and noise interference.

[0105] As a concrete example, synchronization characteristics of modulation operation and noise interference. The calculation formula can be expressed as: ,in T This indicates the number of time points contained in the first and second characteristic curves. This represents the first slope value of the first characteristic curve at time point t. This represents the second slope value of the second characteristic curve at time point t. This represents an exponential function with the natural constant e as its base.

[0106] mean The smaller the value of , the more similar the changing trends of the two fitted curves are, that is, the higher the synchronization effect between the change in noise interference and the change in modulation strategy, and the larger the corresponding value of the synchronization characteristic index of modulation operation and noise interference.

[0107] Step S300: Based on the signal delay times of the current monitoring time period and all historical monitoring time periods, and in conjunction with the synchronization characteristic index, obtain the filtering weights of the multi-channel controllable power converter.

[0108] The response relationship between noise interference and modulation strategies is also affected by signal response delay. Delay can affect not only the timing and waveform of the signal but also the relative relationship between the signal and noise. Observing the distribution of voltage signal data under different delay levels can help design more effective filtering strategies to filter specific noise types. The distribution characteristics under different delay levels can reveal the dynamic behavior of the system in response to different input signals. By analyzing these characteristics, filter parameters can be optimized to better adapt to the dynamic response requirements of the system and reduce the adverse effects of delay.

[0109] Therefore, by comparing the output voltage distribution at different delays, the root cause of the noise can be identified. For example, if the output voltage distribution at a specific delay exhibits strong interference characteristics, it may indicate that this delay is directly related to a modulation behavior of the power converter.

[0110] More specifically, in determining the filter weight, this embodiment comprehensively considers the delay level of historical modulation operations and the delay level within the current monitoring period, and takes into account the synchronization characteristics of modulation operations and noise interference. By using the value of the filter weight, the correlation between noise interference, signal delay, and the modulation operation behavior of the power converter can be characterized.

[0111] Specifically, the ratio of the signal delay time between the current monitoring time period and each historical monitoring time period is calculated. The product of the mean of all ratios and the synchronization characteristic index is then normalized to obtain the filter weights of the multi-channel controllable power converter. As a concrete example, the formula for calculating the filter weights of the multi-channel controllable power converter can be expressed as:

[0112] ;

[0113] in, The filter weights for a multi-channel controllable power converter. Synchronization characteristics of modulation operation and noise interference; This refers to the number of historical modulation operations, which is also the number of historical monitoring periods. Signal delay time during the current monitoring period Let u be the signal delay time for the u-th historical monitoring period; This is the normalization function.

[0114] When the mean A larger value indicates a higher time delay caused by the power converter's modulation operation during the current monitoring period. When the product... The larger the value of , the higher the correlation between noise interference and delay behavior and the modulation operation behavior of the power converter in the output voltage signal data. This value is used as the weight value in the subsequent interference component decomposition process to ensure that the influence of modulation behavior is retained while removing filtering.

[0115] Step S400: Based on the filtering weights, the voltage signal data within the current monitoring time period is decomposed, and the Gaussian characteristics of the decomposed signal components are combined for filtering to obtain the filtered voltage signal data.

[0116] The main purpose of this step is to analyze the interference components in the voltage signal data of the power converter during the current monitoring period. By combining this with the aforementioned filtering weights, the interference components in the voltage signal data can be separated and removed in a relatively reasonable manner. The purpose of using the filtering weights is to ensure that the filtering effect does not have a reverse effect. More specifically, the core is to accurately separate high-frequency interference components from the voltage signal data based on the improved Independent Component Analysis (ICA) method, combined with the filtering weights. This provides a data foundation for subsequent filter design while avoiding modulation signal distortion caused by over-separation.

[0117] As a concrete example, such as Figure 6 As shown, the process of filtering the voltage signal data within the current monitoring time period can be implemented by steps S401 to S403.

[0118] Step S401: Using the objective function, iterate through the feature decomposition process of the voltage signal data within the current monitoring time period to obtain each feature component.

[0119] Specifically, in this embodiment, independent component analysis is used to perform feature analysis on the voltage signal data within the current monitoring time period to obtain each feature component in the decomposition result. The basic principle is to decompose the mixed signal into statistically independent components, and it is assumed that the independent components have the greatest non-Gaussianity; that is, the stronger the non-Gaussianity of each feature component in the decomposition result, the higher its independence.

[0120] Furthermore, in order to separate the noise interference components, this embodiment adopts an iterative scheme to determine the number of independent components to be decomposed in the independent component analysis method, that is, the number of decomposition layers, in order to obtain the best feature decomposition results.

[0121] First, the initial iteration parameters need to be set, namely, the preset number of decomposition layers is 2 and the iteration step size is 1. Each iteration process refers to a complete decomposition process. The number of decomposition layers refers to the number of feature components obtained by decomposition. Then, the number of decomposition layers is adjusted through iteration to optimize the separation effect.

[0122] Step S402: Construct an objective function based on the degree of non-Gaussianity of each feature component and the filtering weights; when the objective function converges to its maximum, the optimal component result of feature decomposition is obtained.

[0123] The specific method for constructing the objective function is as follows:

[0124] The first step is to use the absolute value of the difference between the kurtosis of each feature component and the kurtosis of the standard normal distribution as the non-Gaussianity index of each feature component.

[0125] It should be noted that kurtosis is a characteristic parameter describing the steepness or sharpness of the peaks in a data distribution, representing the degree to which the corresponding data distribution deviates from a normal distribution. The calculation method for kurtosis is a well-known technique and will not be described in detail here. In this embodiment, the theoretical value of kurtosis for a standard normal distribution is 0, representing pure Gaussianity.

[0126] The greater the difference between the kurtosis of each feature component and the kurtosis of the standard normal distribution, the further the data distribution of the feature component deviates from the normal distribution. In this case, the non-Gaussianity of the corresponding feature component is stronger. Conversely, the smaller the difference, the closer the data distribution of the feature component is to the normal distribution. The non-Gaussianity of the corresponding feature component is weaker, and the Gaussianity is stronger.

[0127] It should be understood that non-Gaussianity refers to the degree to which the data distribution of the feature components does not conform to the characteristics of a Gaussian distribution, while Gaussianity refers to the degree to which the data distribution of the feature components conforms to the characteristics of a Gaussian distribution.

[0128] The non-Gaussianity index of each characteristic component provides the data basis for subsequently distinguishing between effective signals (strong non-Gaussianity) and interference components (strong Gaussianity). The core idea of ​​independent component analysis is that effective signals have strong non-Gaussianity, such as the stable voltage component of a power supply output. However, since noise usually exhibits a random and irregular distribution, high-frequency noise interference components have strong Gaussianity. Therefore, the difference in non-Gaussianity can effectively distinguish between effective signals and interference components.

[0129] The second step is to take the last feature component in the decomposition order of each iteration as the target component and take the other feature components as the control components.

[0130] It should be understood that in the process of eigenvalue decomposition, the decomposition order is usually a process of separating from the strongest signal to the weakest signal. The last feature component in the decomposition order of each iteration refers to the last independent component generated in the decomposition process, which is determined by the number of decomposition layers in the iteration process.

[0131] Meanwhile, the decomposition order is related to the strength of the non-Gaussianity of the signal. That is, the stronger the non-Gaussianity of the feature component, the earlier it is decomposed. Therefore, the target component represents the feature component with the weakest non-Gaussianity, which may usually represent noise or interference components. The other feature components that are decomposed before the target component, that is, the control components, represent the effective signal.

[0132] The third step is to use the product of the ratio between the mean of the non-Gaussianity index of all control components and the non-Gaussianity index of the target component in each iteration and the filtering weight as the objective function in each iteration.

[0133] The ratio between the mean of the non-Gaussianity indices of all control components and the non-Gaussianity indices of the target component represents the difference in non-Gaussianity between the control components (effective components) and the target components (interference components) during independent component analysis. The larger the ratio, the stronger the Gaussianity in the target component. The filter weight is introduced when constructing the objective function to avoid over-separation. If the noise is strongly correlated with the modulation operation behavior, the filter weight will limit the amplification of the ratio and prevent the modulation-related signal from being mistakenly classified as interference.

[0134] Finally, when the objective function converges to its maximum, the optimal component result of the eigenvalue decomposition is obtained. That is, the objective function is used to determine the optimal number of decomposition layers in the iterative process. When the objective function converges to its maximum, the number of decomposition layers at this point is optimal. The larger the ratio between the mean of the non-Gaussianity indices of all control components and the non-Gaussianity indices of the target component, the more significant the difference in non-Gaussianity between the target component and the control components, meaning the stronger the Gaussianity of the target component, and the more likely the target component is to be an interfering component.

[0135] Step S403: In the optimal decomposition result, the voltage signal data in the current monitoring time period is filtered according to the signal distribution of the last feature component in the decomposition order to obtain the filtered voltage signal data.

[0136] The optimal decomposition result has the optimal number of decomposition layers. The target component in the optimal decomposition result represents the Gaussian components (i.e., high-frequency interference components) in the original voltage signal data, while the control component in the optimal decomposition result is the effective signal with non-Gaussian characteristics (such as the stable voltage component of the power supply output).

[0137] In the optimal decomposition result, the standard deviation of the signal data of the last feature component in the decomposition order is negatively correlated and normalized to obtain the adjustment coefficient. The method of negative correlation normalization is a well-known technique, for example, it can be achieved by using... Negative correlation normalization is performed in the following manner, where For normalization function, This represents the parameter to be processed, without specific limitations. The adjustment coefficient characterizes the noise intensity of the high-frequency interference component in the optimal decomposition result.

[0138] Furthermore, the product of the adjustment coefficient and the preset cutoff frequency of the low-pass filter is rounded down to obtain the adjusted cutoff frequency for filtering. This adjusted cutoff frequency is then set as the new cutoff frequency of the low-pass filter. The voltage signal data within the current monitoring time period is then filtered to obtain the filtered voltage signal data for that period. The preset cutoff frequency of the low-pass filter is a value set based on experience.

[0139] When the standard deviation of high-frequency interference components is small, the fluctuation of noise interference components is small, indicating that the noise in the voltage signal data is weak. In this case, there is no need to over-filter high-frequency components, and the cutoff frequency should be larger to retain high-frequency details in the effective signal. When the standard deviation of high-frequency interference components is large, the fluctuation of noise interference components is drastic. In this case, high-frequency noise needs to be suppressed. The cutoff frequency of the low-pass filter should be smaller to prioritize the stability of the output signal and avoid signal distortion caused by noise.

[0140] This embodiment uses adaptive setting of the low-pass filter cutoff frequency to filter the voltage signal data within the current monitoring time period, effectively removing high-frequency interference and retaining the valid signal. Figure 7 The data shown is the filtered voltage signal data for the current monitoring period.

[0141] In other embodiments, to facilitate subsequent real-time monitoring of signal data, feature extraction can be performed on the filtered signal to extract useful information, such as the effective values ​​and peak values ​​of voltage and current waveforms, for subsequent control and monitoring. The filtered output voltage signal can be used as a feedback signal and integrated into a closed-loop control system to adjust the output parameters of the multi-channel controllable power converter, ensuring that the system can respond quickly to changes. Detailed operation logs can be established to record voltage signal data, control decisions, and their effects before and after filtering, facilitating subsequent analysis and improvement. This part is not the focus of this application and is only briefly introduced here.

[0142] In summary, this invention first evaluates the waveform and rapid vibration behavior of voltage signal data during historical monitoring periods. Then, it comprehensively assesses the noise interference level in the voltage signal data during historical monitoring periods by combining the changes in harmonic components in the voltage signal data spectrum. Next, it analyzes the modulation behavior of the power converter and the resulting changes in noise interference, obtaining the synchronization effect between changes in noise interference and changes in the power converter's modulation behavior. Based on the possible time delay during modulation, a weight value is obtained for the voltage signal data during filtering. This weight value is then introduced into an independent component analysis method to remove interference components from the voltage signal data, which serves as the basis for designing filter data. This enables filtering processing of multi-channel controllable power converters, ensuring efficient filtering while reducing system complexity, size, and cost, and improving real-time response to current ripple.

[0143] This invention also provides a multi-channel controllable power converter for implementing filtering processing, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the computer program is executed by the processor, it implements the steps of a filtering processing method for a multi-channel controllable power converter for implementing filtering processing.

[0144] Since a detailed embodiment of a filtering method for a multi-channel controllable power converter has already been described, it will not be repeated here.

[0145] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A filtering method for a multi-channel controllable power converter, characterized in that, The method includes the following steps: Acquire signal delay time and voltage signal data of the multi-channel controllable power converter in the current monitoring time period and historical monitoring time period. Each historical modulation operation of the multi-channel controllable power converter corresponds to a historical monitoring time period and pulse width parameter. Based on the peak and frequency distribution of voltage signal data in each historical monitoring period, the noise interference level in each historical monitoring period is analyzed. Combining the differences in noise interference level and pulse width parameter change trends, the synchronization characteristic index of modulation operation and noise interference is obtained. Based on the signal delay time of the current monitoring period and all historical monitoring periods, and combined with the aforementioned synchronization characteristic index, the filtering weights of the multi-channel controllable power converter are obtained. Based on the filtering weights, the voltage signal data within the current monitoring time period is decomposed, and filtered by combining the Gaussian characteristics of the decomposed signal components to obtain the filtered voltage signal data.

2. The filtering method for a multi-channel controllable power converter according to claim 1, characterized in that, The analysis of noise interference levels for each historical monitoring period, based on the peak and frequency distributions of voltage signal data, specifically includes: Based on the difference in the degree of characteristic between adjacent peak signal segments within each historical monitoring period, the first signal noise index for each historical monitoring period is obtained. Based on the frequency distribution of the spectral characteristics of the voltage signal data within each historical monitoring period, a second signal noise index is obtained for each historical monitoring period. The product of the first signal noise index and the second signal noise index is used as the noise interference level for each historical monitoring period.

3. The filtering method for a multi-channel controllable power converter according to claim 2, characterized in that, The method of obtaining the first signal noise index for each historical monitoring time period based on the difference in the degree of characteristic between adjacent peak signal segments within each historical monitoring time period specifically includes: Within any historical monitoring period, the signal difference between any two adjacent peak signal points is obtained. The peak signal point corresponding to the maximum value of the signal difference is used as the segmentation signal point. All peak signal points whose peak signal points are greater than or equal to the segmentation signal points are selected to form a characteristic peak point sequence. Based on the difference between the peak widths of every two adjacent characteristic peak points in the characteristic peak point sequence, the peak difference coefficient is determined; the maximum value of the time interval between adjacent characteristic peak points in the characteristic peak point sequence is taken as the maximum time interval. The ratio of the peak difference coefficient to the maximum time interval is normalized to obtain the first signal noise index corresponding to any historical monitoring time period.

4. The filtering method for a multi-channel controllable power converter according to claim 3, characterized in that, The second signal noise index for each historical monitoring period is obtained based on the frequency distribution of the spectral characteristics of the voltage signal data within each historical monitoring period, specifically including: A Fourier transform is performed on the voltage signal data within any historical monitoring period to obtain a spectrum, and the frequency value corresponding to the largest amplitude in the spectrum is taken as the main frequency. The frequency value of each harmonic is determined based on integer multiples of the dominant frequency; curve fitting is performed on the frequency value and amplitude of each harmonic in the spectrum diagram, and the slope value of each harmonic is obtained on the fitted curve; based on the slope value of each harmonic, all harmonics are screened to obtain suspected anomalous harmonics. The mean difference between the frequency value of each suspected abnormal harmonic and the maximum frequency value in the spectrum is negatively correlated to obtain the second signal noise index for any historical monitoring time period.

5. The filtering method for a multi-channel controllable power converter according to claim 1, characterized in that, The method analyzes the noise interference level of each historical monitoring period based on the peak and frequency distribution of voltage signal data within each historical monitoring period. Combining the differences in noise interference level and pulse width parameter variation trends, synchronization characteristic indicators of modulation operation and noise interference are obtained, specifically including: The first characteristic curve is obtained by curve fitting with the time point of the historical modulation operation corresponding to each historical monitoring period as the horizontal axis and the noise interference level of each historical monitoring period as the vertical axis. The second characteristic curve is obtained by curve fitting with the pulse width reference of the modulation operation corresponding to each historical monitoring period as the vertical axis. Obtain the first slope value of each data point on the first feature curve and the second slope value of each data point on the second feature curve; By performing negative correlation processing on the mean of the differences between the first slope value and the second slope value at all the same time points, the synchronization characteristic index of modulation operation and noise interference is obtained.

6. The filtering method for a multi-channel controllable power converter according to claim 1, characterized in that, The step of obtaining the filtering weights of the multi-channel controllable power converter based on the signal delay times of the current monitoring time period and all historical monitoring time periods, combined with the synchronization characteristic index, specifically includes: Calculate the ratio of the signal delay time between the current monitoring time period and each historical monitoring time period, normalize the product of the mean of all ratios and the synchronization characteristic index, and obtain the filtering weight of the multi-channel controllable power converter.

7. The filtering method for a multi-channel controllable power converter according to claim 1, characterized in that, The process of decomposing the voltage signal data within the current monitoring time period according to the filtering weights, and then filtering the decomposed signal components based on their Gaussian characteristics to obtain filtered voltage signal data specifically includes: Using an objective function, the feature decomposition process of voltage signal data within the current monitoring time period is iteratively performed to obtain each feature component; wherein, the objective function is constructed based on the degree of non-Gaussianity of each feature component and the filtering weight; when the objective function converges to its maximum, the optimal component result of feature decomposition is obtained; In the optimal decomposition result, the voltage signal data within the current monitoring time period is filtered based on the signal distribution of the last feature component in the decomposition order to obtain the filtered voltage signal data.

8. The filtering method for a multi-channel controllable power converter according to claim 7, characterized in that, The construction of the objective function based on the degree of non-Gaussianity of each feature component and the filtering weights specifically includes: The absolute value of the difference between the kurtosis of each feature component and the kurtosis of the standard normal distribution is used as the non-Gaussianity index of each feature component; The last feature component in the decomposition order of each iteration is taken as the target component, and the other feature components are taken as the control components. The product of the ratio between the mean of the non-Gaussianity indices of all control components and the non-Gaussianity indices of the target component in each iteration, and the filtering weights, is used as the objective function for each iteration.

9. The filtering method for a multi-channel controllable power converter according to claim 7, characterized in that, In the optimal decomposition result, the voltage signal data within the current monitoring time period is filtered based on the signal distribution of the last feature component in the decomposition order, specifically including: In the optimal decomposition result, the standard deviation of the signal data of the last feature component in the decomposition order is negatively correlated and normalized to obtain the adjustment coefficient. The product of the adjustment coefficient and the preset cutoff frequency of the low-pass filter is rounded down to obtain the adjusted cutoff frequency for filtering, and the voltage signal data within the current monitoring period is filtered.

10. A multi-channel controllable power converter for implementing filtering processing, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of a filtering method for a multi-channel controllable power converter as described in any one of claims 1-9.

Images