A method and system for identifying main sound source noise for a flexible HVDC converter station

Abstract

The present application relates to the technical field of noise identification, and particularly relates to a main sound source noise identification method and system for a flexible direct current converter station. Current of the converter station and audio time sequence data of each monitoring point are acquired; time-frequency diagrams are obtained by processing the audio time sequence data, amplitude variation differences of different monitoring points under the same frequency point are analyzed, amplitude variation of the target point frequency point is adjusted accordingly, an amplitude variation index is obtained, and whether the noise is from the converter transformer is analyzed conveniently; for the noise caused by the magnetostriction of the converter transformer core, the vibration frequency is stable, and the amplitude varies synchronously with the current, so in the target point time-frequency diagram, the correlation between the amplitude variation index and the current time sequence data and the difference of the time sequence characteristics under each frequency are analyzed, a multi-frequency synchronization index is determined, and the possibility that the frequency component is from the magnetostriction effect of the core is reflected; finally, the amplitude variation and the multi-frequency synchronization index are fused, noise characteristics are comprehensively analyzed, and noise source identification accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of noise identification, in particular to a main sound source noise identification method and system for a flexible DC converter station. BACKGROUND

[0002] With the rapid development of the power industry, flexible DC power transmission technology has been widely used in many fields due to its advantages of flexibility, controllability, and realization of multi-terminal DC power transmission. The flexible DC converter station, as the core equipment of the flexible DC power transmission system, its operating state is directly related to the stability and reliability of the entire power system. However, the flexible DC converter station will produce complex noise during operation, which not only pollutes the surrounding environment, but also may mask the potential fault characteristics inside the device, so accurately identifying the main sound source noise of the flexible DC converter station is of great significance for evaluating the operating state of the device, timely discovering potential faults and taking effective noise reduction measures.

[0003] The prior art usually identifies noise based on peak frequency or fixed characteristic spectrum, but due to the multiple noise sources and dense spatial distribution in the converter station, in addition to the noise generated by the converter transformer core due to the magnetostriction effect, the noise radiated by the bridge arm reactor, cooling fan and other equipment is also superimposed and reflected in space, which will show dynamic mixing characteristics of multiple frequency components, so it is difficult to adapt to this complex and variable noise environment if only relying on peak frequency or fixed characteristic spectrum, which will greatly reduce the accuracy of main sound source noise identification. SUMMARY

[0004] In order to solve the technical problem that the noise source in the converter station is multiple and densely distributed in space, in addition to the noise generated by the converter transformer core due to the magnetostriction effect, the noise radiated by the bridge arm reactor, cooling fan and other equipment is also superimposed and reflected in space, which will show dynamic mixing characteristics of multiple frequency components, so it is difficult to adapt to this complex and variable noise environment if only relying on peak frequency or fixed characteristic spectrum, which will greatly reduce the accuracy of main sound source noise identification, the purpose of the present application is to provide a main sound source noise identification method and system for a flexible DC converter station, the technical scheme adopted is as follows:

[0005] A main sound source noise identification method for a flexible DC converter station, comprising:

[0006] Obtaining current time series data of the converter station and audio time series data at each monitoring point;

[0007] The audio time sequence data of each monitoring point is processed to obtain a time-frequency graph; a target point is determined based on the positional relationship of the monitoring points; in the time-frequency graphs corresponding to different monitoring points, the amplitude change difference characteristics of different monitoring points at the same frequency point are analyzed, the amplitude change of the target point at the time-frequency point is adjusted, and the amplitude change index of the target point at the time-frequency point is obtained;

[0008] In the time-frequency graph of the target point, the correlation between the amplitude change index of the time-frequency point in the local time range and the current time sequence data is analyzed, and the time sequence characteristics of the local time range are combined to determine the time-frequency stability index corresponding to each time-frequency point; the difference characteristics of the time-frequency stability index at the same time and at different frequencies are analyzed, and the sorting characteristics of the frequency are determined to determine the multi-frequency synchronization index corresponding to each time;

[0009] At the target point, the amplitude change index and the multi-frequency synchronization index corresponding to each time are fused to identify the noise source of the target point.

[0010] Further, the amplitude change index acquisition method comprises:

[0011] In the time-frequency graph of each monitoring point, the amplitude difference value of each time-frequency point and the previous time-frequency point adjacent in time is normalized as the amplitude change factor corresponding to each time-frequency point at the same frequency;

[0012] The difference characteristics of the amplitude change factor of different monitoring points at the same frequency point are analyzed to determine the transformer dominant index corresponding to each time-frequency point;

[0013] In the time-frequency graph of the target point, the product of the transformer dominant index and the amplitude change factor corresponding to each time-frequency point is normalized as the amplitude change index corresponding to each time-frequency point.

[0014] Further, the transformer dominant index acquisition method comprises:

[0015] In the time-frequency graphs of all monitoring points, the same time-frequency points are taken as time-frequency point groups, and in each time-frequency point group, all time-frequency points are sorted according to the preset order of the monitoring points to obtain a sorting sequence;

[0016] In the sorting sequence, the difference value of the amplitude change factor of the first time-frequency point and the last time-frequency point is taken as the overall change factor; a first-order difference sequence of the amplitude change factor of all time-frequency points in the sorting sequence is obtained, and the sum of the absolute values of all values in the first-order difference sequence is taken as the local change factor;

[0017] Map the absolute value of the difference between the overall change factor and the local change factor to a negative correlation and normalize the value, as the transformer dominant indicator corresponding to all time-frequency points in each time-frequency point group.

[0018] Further, the acquisition method of the time-frequency stability indicator comprises:

[0019] In the time-frequency graph of the target point, the time sequence is uniformly divided into multiple time interval;

[0020] For any time, based on the time sequence relationship between the time and each time interval, the distance weight between the time and each time interval is determined;

[0021] In each time interval, the correlation between the amplitude change indicator and the current time sequence data is analyzed to determine the stability factor of each time interval at each frequency;

[0022] The product of the distance weight between each time and each time interval and the stability factor of each time interval at each frequency is taken as the weighted stability factor between each time and each time interval at each frequency;

[0023] The normalized value of the mean of all weighted stability factors of each time at each frequency is taken as the time-frequency stability indicator of each time at each frequency.

[0024] Further, the acquisition method of the distance weight comprises:

[0025] For any time, the distance weight between the time and the time interval in which the time is located is set to 1, and the minimum value of the absolute value of the difference between the time and each remaining time interval is mapped to a negative correlation and normalized, as the distance weight between the time and the time interval.

[0026] Further, the acquisition method of the stability factor comprises:

[0027] In each time interval, the amplitude change indicators of all times at the same frequency are sorted in time sequence to obtain an amplitude change sequence, and the absolute value of the Pearson correlation coefficient between the amplitude change sequence and the current subsequence corresponding to the time interval in the current time sequence data is normalized to obtain the stability factor of each time interval at each frequency.

[0028] Further, the acquisition method of the multi-frequency synchronization indicator comprises:

[0029] In the time-frequency graph of the target point, the difference between each frequency and the preset frequency is mapped to a negative correlation and normalized, as the attention weight corresponding to each frequency;

[0030] At the same moment, the absolute value of the difference between each frequency and the preset frequency is negatively correlated and normalized, and the value is taken as a similarity factor between each frequency and the preset frequency;

[0031] At each moment, the product of the attention weight corresponding to each frequency and the similarity factor between each frequency and the preset frequency is taken as a weighted similarity value, and the normalized value of the average of the weighted similarity values between the preset frequency and all frequencies at each moment is taken as a multi-frequency synchronization indicator corresponding to each moment.

[0032] Further, the amplitude variation indicator and the multi-frequency synchronization indicator corresponding to each moment at the target point are fused to identify the noise source of the target point, comprising:

[0033] At the target point, the product of the amplitude variation indicator and the multi-frequency synchronization indicator corresponding to each moment is normalized to obtain a core noise indicator at each moment.

[0034] In the time-frequency graph of the target point, when the core noise indicator corresponding to a certain moment is greater than or equal to a preset core noise threshold, the moment is marked as a core noise moment.

[0035] Further, the time-frequency graph acquisition method comprises:

[0036] The audio time series data of each monitoring point is processed based on short-time Fourier transform, thereby obtaining a time-frequency graph of each monitoring point.

[0037] A main sound source noise identification system for a flexible HVDC converter station, comprising:

[0038] A data acquisition module is configured to acquire current time series data of the converter station and audio time series data at each monitoring point.

[0039] An amplitude variation analysis module is configured to process the audio time series data of each monitoring point to obtain a time-frequency graph, determine a target point based on the positional relationship of the monitoring points, analyze the amplitude variation difference characteristics of different monitoring points at the same frequency point in the time-frequency graphs corresponding to the different monitoring points, adjust the amplitude variation at the time-frequency point corresponding to the target point, and obtain an amplitude variation indicator at the time-frequency point corresponding to the target point.

[0040] A frequency domain synchronization analysis module is configured to analyze the correlation between the amplitude variation indicator of the time-frequency point and the current time series data within a local time range in the time-frequency graph of the target point, and combine the time series characteristics of the local time range to determine a time-frequency stability indicator corresponding to each time-frequency point, analyze the difference characteristics of the time-frequency stability indicators at the same moment at different frequencies and the sorting characteristics of the frequencies, and determine a multi-frequency synchronization indicator corresponding to each moment.

[0041] The noise identification module is used to fuse the amplitude change index and multi-frequency synchronization index at each time point to identify the noise source at the target point.

[0042] The present invention has the following beneficial effects:

[0043] The process involves acquiring current timing data and audio timing data from various monitoring points at the converter station. Current timing data reflects the equipment's operating status, while audio timing data contains noise information. The audio timing data from each monitoring point is processed to obtain a time-frequency diagram, converting the time-domain audio signal into a time-frequency representation. This clearly shows the distribution of sound across different times and frequencies. Furthermore, the amplitude variation characteristics at the same time and frequency point are analyzed within the time-frequency diagrams corresponding to different monitoring points. Based on this analysis, the amplitude variation at the target point's corresponding time and frequency point is adjusted to obtain an amplitude variation index at the target point's corresponding time and frequency point. This facilitates subsequent analysis to determine whether the noise primarily originates from the converter transformer, thereby mitigating the influence of other noise factors. Furthermore, for the magnetostrictive noise of the converter transformer core, its excitation originates from the periodic changes in the power grid's frequency flux. The frequency of this vibration does not drift over time; only the amplitude changes synchronously with the current. Therefore, in the time-frequency diagram of the target location, the correlation between the amplitude variation index and the current time-series data at each time point, as well as the differences in time-series characteristics at different frequencies, are analyzed to determine the multi-frequency synchronization index corresponding to each moment. This index reflects the possibility that the frequency components originate from the magnetostrictive effect of the core. Finally, at the target location, the amplitude variation index and the multi-frequency synchronization index corresponding to each moment are fused, integrating the noise's characteristics from different aspects. This fusion provides a more comprehensive and accurate description of the noise's features. Therefore, using the fused index to identify the noise source at the target location can improve the accuracy and reliability of noise source identification. Attached Figure Description

[0044] 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.

[0045] Figure 1 This is a flowchart of a method for identifying the main sound source noise in a flexible DC converter station according to an embodiment of the present invention;

[0046] Figure 2A flowchart illustrating a method for obtaining an amplitude change index according to an embodiment of the present invention;

[0047] Figure 3 A flowchart illustrating a method for obtaining a time-frequency stability index according to an embodiment of the present invention;

[0048] Figure 4 This is a system block diagram of a main sound source noise identification system for a flexible DC converter station provided in one embodiment of the present invention;

[0049] Figure 5 This is a schematic diagram of an electronic device structure provided in one embodiment of the present invention. Detailed Implementation

[0050] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method and system for identifying main sound source noise in a flexible DC converter station according to the present invention. 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 can be combined in any suitable form.

[0051] 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.

[0052] The following description, in conjunction with the accompanying drawings, details a specific scheme for a main sound source noise identification method and system for a flexible DC converter station provided by the present invention.

[0053] Please see Figure 1 The diagram illustrates a method flowchart for identifying the main sound source noise in a flexible DC converter station according to an embodiment of the present invention. The method includes the following steps:

[0054] Step S1: Obtain the current timing data of the converter station and the audio timing data at each monitoring point.

[0055] There are various noise sources in flexible DC converter stations, but the core of the converter transformer will generate periodic structural vibrations under electromagnetic excitation due to the magnetostriction effect, and form sound radiation with significant fundamental frequency and integer multiple harmonics in the air. This type of noise not only has a high energy proportion and a wide coverage, but also shows obvious temporal fluctuations with changes in load current and magnetic flux density. Compared with aerodynamic noise such as fans and valve cooling systems, core magnetostriction noise has stronger electromagnetic acoustic coupling characteristics and traceability. Therefore, in this embodiment of the invention, the core vibration noise caused by the magnetostriction effect is mainly identified.

[0056] Five monitoring points are set up around the converter transformer in the converter station. Each monitoring point is equipped with an acoustic sensor (such as a condenser microphone). The acoustic sensors are arranged at equal intervals from the converter transformer, from closest to furthest. For example, the distance between every two acoustic sensors is 10m. The sampling frequency is 4kHz, thereby obtaining the audio timing data of each monitoring point in the converter station. At the same time, according to the voltage level and current of the converter station, a current transformer with an appropriate range and accuracy is selected to obtain the current timing data of the converter station. The sampling frequency is also set to 4kHz.

[0057] It should be noted that in this embodiment of the invention, current timing data and audio timing data need to be collected synchronously, and the length of the timing data is set to 5 minutes. The collection frequency and the length of the timing data can be adjusted according to the implementation scenario, and are not limited here. Similarly, the setting of monitoring points can also be set according to the actual situation, and are not limited here.

[0058] Step S2: Process the audio time-series data of each monitoring point to obtain a time-frequency diagram; determine the target point based on the positional relationship of the monitoring points; analyze the amplitude variation characteristics of different monitoring points at the same time-frequency point in the time-frequency diagrams corresponding to different monitoring points, adjust the amplitude variation at the time-frequency point corresponding to the target point, and obtain the amplitude variation index at the time-frequency point corresponding to the target point.

[0059] Time-domain waveforms (such as the amplitude change of an audio signal over time) can only reflect the overall energy change of the signal and cannot distinguish the dynamic characteristics of different frequency components. However, the noise characteristics of the environment in which the equipment in the converter station is located, such as the sound radiation generated by the magnetostriction effect of the converter transformer core, often give the audio data significant fundamental frequency and its integer multiples of harmonics. Furthermore, there are various devices such as transformers, converter valves, and reactors in the converter station, and their acoustic signals may superimpose to form a complex mixed sound field. Therefore, in this embodiment of the invention, the audio time-series data of each monitoring point can be processed to obtain a time-frequency diagram, which is used to establish a mapping between the time domain and the frequency domain, showing the relationship between the frequency components of the signal and time, which facilitates better capture of the frequency characteristics and time-frequency distribution of the signal.

[0060] Preferably, in one embodiment of the present invention, the method for obtaining a time-frequency diagram includes:

[0061] The audio time-series data of each monitoring point is processed based on the short-time Fourier transform to obtain the time-frequency diagram of each monitoring point. In the video diagram, the horizontal axis is time and the vertical axis is frequency. Each intersection point (time t, frequency f) corresponds to a time-frequency point, and the color intensity of each time-frequency point indicates the amplitude.

[0062] It should be noted that the process of obtaining the time-frequency graph using short-time Fourier transform is a well-known technique, and the specific process will not be described in detail here. In other embodiments of the present invention, other methods may also be used to obtain the time-frequency graph, and no limitation is made here.

[0063] Around the converter transformer in the converter station, the near-end point mainly captures the transformer's own noise (magnetostriction, electromagnetic vibration); while the far-end point, because it is farther from the noise source, may be more significantly affected by the ambient background noise. Therefore, if the influence of the converter transformer's own noise is still greater in the audio timing data at the far-end point, then the converter transformer can be considered more likely to be the main noise source. Therefore, preferably, in this embodiment of the invention, the monitoring point farthest from the converter transformer in the converter station can be taken as the target point.

[0064] Converter transformers are centralized sound sources, and their sound radiation energy decreases exponentially with distance (e.g., 6 dB / time). If the amplitude change of a certain frequency (e.g., 100 Hz) is significant at monitoring points close to the transformer (e.g., at 10 m) (e.g., amplitude fluctuation ±5 dB), while at more distant points (e.g., at 50 m) the amplitude fluctuation weakens to ±1 dB, it indicates that the energy at that frequency mainly originates from the transformer. Therefore, analyzing the amplitude change differences in the time-frequency graphs corresponding to different monitoring points facilitates a preliminary analysis of whether the noise mainly originates from the converter transformer, thereby mitigating the influence of other noise factors. Thus, in the time-frequency graphs corresponding to different monitoring points, the amplitude change differences at the same time-frequency point are analyzed, and the amplitude change at the time-frequency point corresponding to the target point is adjusted to obtain the amplitude change index at the time-frequency point corresponding to the target point.

[0065] Preferably, in one embodiment of the present invention, the method for obtaining the amplitude change index includes:

[0066] Please see Figure 2 The diagram illustrates a method flowchart for obtaining an amplitude change index according to an embodiment of the present invention. The method includes the following steps:

[0067] Step S201: In the time-frequency graph of each monitoring point, analyze the amplitude changes at the same frequency at adjacent times, and determine the amplitude change factor corresponding to each time-frequency point.

[0068] In the time-frequency graph of each monitoring point, at the same frequency, the normalized value of the amplitude difference between each time-frequency point and the preceding time-frequency point is used as the amplitude variation factor for each time-frequency point. The amplitude variation factor can characterize the amplitude variation characteristics between adjacent times at the same frequency in each monitoring point. Since the amplitude difference here may be positive or negative, the normalization process here can be performed using... function.

[0069] It should be noted that, since the magnetostrictive noise of the iron core has a significant fundamental frequency (100Hz) and its integer multiple harmonics, the frequencies in the embodiments of the present invention specifically refer to 100Hz, 200Hz, 300Hz, 400Hz, ..., 700Hz; for the time-frequency point corresponding to the first moment at each frequency, since there is no adjacent previous time-frequency point, its corresponding amplitude variation factor can be set to be consistent with the second time-frequency point.

[0070] Step S202: Analyze the differences in amplitude variation factors at different monitoring points under the same time and frequency, and determine the dominant transformer index corresponding to each time and frequency.

[0071] When the amplitude change of a certain frequency is only obvious at the monitoring point close to the converter transformer, and the amplitude change is weaker at other monitoring points farther away from the transformer, it indicates that the energy of that frequency is mainly generated by the converter transformer. If the interference is caused by other sound sources, it will affect the law of amplitude change as the distance increases and the noise propagation becomes weaker. Therefore, the difference in spatial distribution can be used to determine whether the noise from the converter transformer is the main noise source at each time and each frequency.

[0072] First, in the time-frequency graphs of all monitoring points, points with the same time frequency are grouped together as time-frequency point groups, for example, monitoring points Monitoring points ... monitoring points A time-frequency point group is formed. In each time-frequency point group, all time-frequency points are sorted according to the preset order of the monitoring points (from near to far) to obtain the sorting sequence.

[0073] In the sorted sequence, the difference between the amplitude variation factors of the first and last time-frequency points is taken as the overall variation factor. The overall variation factor characterizes the amplitude variation features of the first and last time-frequency points of the sequence. A value less than 0 indicates that the amplitude variation at the far end is stronger and more likely to be an interference source, while a value greater than 0 indicates that the amplitude variation at the near end is stronger and more likely to be a transformer noise source. Then, the first-order difference sequence of the amplitude variation factors of all time-frequency points in the sorted sequence is obtained. The values ​​in the first-order difference sequence reflect the intensity of the fluctuation in the spatial distribution. A positive value indicates that the amplitude variation at the far end between two adjacent monitoring points is stronger. The sum of the absolute values ​​of all values ​​in the first-order difference sequence is taken as the local variation factor.

[0074] When the noise source is transformer noise, the value of the local variation factor should be consistent with the value of the overall variation factor. Therefore, if the difference between the two is too large, it indicates the presence of other noise sources, causing a deviation between the local amplitude variation characteristics and the overall variation characteristics. Thus, the absolute value of the difference between the overall and local variation factors is calculated. A larger absolute value indicates a greater difference, suggesting a lower probability that the main noise source is caused by the iron core magnetostriction effect. Conversely, a smaller difference indicates a higher probability that the main noise source is caused by the iron core magnetostriction effect. Therefore, the absolute value of this difference is negatively correlated and normalized to correct the logical relationship, thereby obtaining the transformer-dominant index for all time-frequency points in each time-frequency point group. A larger transformer-dominant index indicates that at that time-frequency point, the main factor affecting amplitude variation is the converter transformer, with less interference from other noise sources. This negative correlation mapping and normalization can be performed using the formula... ,in, Let x represent an exponential function with the natural constant e as the base, and let x represent the independent variable.

[0075] Step S203: In the time-frequency diagram of the target location, adjust the amplitude change factor corresponding to the time-frequency point using the transformer dominant index corresponding to the time-frequency point, thereby obtaining the amplitude change index corresponding to each time-frequency point.

[0076] Based on the aforementioned steps, the transformer dominance index corresponding to each time-frequency point can be determined. This index characterizes the probability that the noise source is dominated by the transformer. Therefore, in the time-frequency diagram of the target location, the normalized value of the product of the transformer dominance index and the amplitude variation factor corresponding to each time-frequency point can be used as the amplitude variation index corresponding to each time-frequency point. This amplitude variation index has been adjusted by the transformer dominance index, thus making the amplitude variation factor affected by the transformer magnetostriction effect more prominent and effectively reducing the components of other sound sources in the amplitude variation. Normalization is a technique well-known to those skilled in the art. The choice of normalization function can be linear normalization or standard normalization, etc., and the specific normalization method is not limited here.

[0077] Step S3: In the time-frequency diagram of the target location, analyze the correlation between the amplitude change index of the time-frequency point and the current time-series data within the local time range, and combine it with the time-series characteristics of the local time range to determine the time-frequency stability index corresponding to each time-frequency point; analyze the difference characteristics of the time-frequency stability index at different frequencies at the same time, as well as the frequency ranking characteristics, to determine the multi-frequency synchronization index corresponding to each time.

[0078] For the magnetostrictive noise of the converter transformer core, its excitation originates from the periodic changes in the power grid's frequency magnetic flux. Therefore, it continuously generates mechanical vibrations at the 100Hz fundamental frequency and its harmonics. The frequency of this vibration does not drift with time; only the amplitude changes synchronously with the current. Thus, on the time-frequency diagram, it appears as a narrow band of energy fringes with a fixed position and varying brightness over time. Furthermore, since the magnetostrictive effect is essentially caused by changes in the magnetic flux density within the core, and magnetic flux density is proportional to the current, the change in core magnetic flux density occurs as the alternating current increases. Increased amplitude leads to greater magnetostrictive deformation, resulting in a significant increase in vibration amplitude and sound pressure level. Therefore, when the brightness of the stripes at a fixed frequency on the time-frequency plot changes synchronously with the current, it indicates that the noise amplitude is dynamically affected by the current. Furthermore, when performing correlation analysis on the amplitude and current at a fixed frequency, if the fan start-up / shutdown, cooling tower startup, or sudden load change of the bridge arm reactor happens to occur within the analysis time window, their acoustic energy changes will generate additional energy in the same frequency band, reducing the correlation between the acoustic signal and the current signal caused by the magnetostrictive effect. Therefore, in the time-frequency plot of the target location, the correlation between the amplitude change index of the time-frequency point within a local time range and the current time-series data can be analyzed, and combined with the time-series characteristics of the local time range, to obtain the time-frequency stability index corresponding to each time-frequency point, which can be used to further characterize the possibility that the noise source is a converter transformer.

[0079] Preferably, in one embodiment of the present invention, the method for obtaining the time-frequency stability index includes:

[0080] Please see Figure 3 The diagram illustrates a method flowchart for obtaining time-frequency stability indicators according to an embodiment of the present invention. The method includes the following steps:

[0081] Step S301: Divide the time series into multiple time intervals and analyze the time series relationship between each time interval to obtain the distance weight.

[0082] In the time-frequency diagram of the target location, the time series is evenly divided into multiple time intervals. For example, 5 seconds can be set as the length of a time interval to divide the time series.

[0083] For any given moment, the distance weight between that moment and its corresponding time interval is set to 1, indicating a strong correlation. Within each of the remaining time intervals, the minimum absolute value of the difference between that moment and all moments within that interval is obtained, and this value is then subjected to negative correlation mapping and normalization. This normalized value is used as the distance weight between the moment and the time interval. The smaller the absolute value of the difference, the larger the distance weight, indicating a closer temporal sequence and higher reference value for the data features within that time interval, i.e., a stronger correlation. The negative correlation mapping and normalization can be performed using the following formula: ,in, Let x represent an exponential function with the natural constant e as the base, and let x represent the independent variable.

[0084] Step S302: In each time interval, analyze the correlation between the amplitude change index and the current time series data, and determine the stability factor of each time interval at each frequency.

[0085] Within each time interval, the amplitude change indicators at the same frequency at all times are sorted according to time sequence to obtain the amplitude change sequence at each frequency. Then, based on the time interval, the corresponding current subsequence is extracted from the current time series data.

[0086] Based on the aforementioned logic, for the magnetostrictive noise of the converter transformer core, the amplitude change of frequency should exhibit a synchronous trend with the current change. Therefore, the Pearson correlation coefficient between the amplitude change sequence and the corresponding current subsequence at each frequency for each time interval was calculated. The Pearson correlation coefficient was then normalized to obtain the stability factor at each frequency for each time interval. A larger stability factor indicates a stronger consistency with the persistent characteristics of magnetostrictive noise of the converter transformer core caused by current changes, and a higher likelihood that it is indeed magnetostrictive noise. Since the Pearson correlation coefficient can be positive or negative, the normalization process here can employ... function.

[0087] Step S303: At each frequency, the stability factor and distance weight between each time point and time interval are fused to obtain the time-frequency stability index at each frequency for each time point.

[0088] Based on the analysis in step S301, it is known that the greater the distance weight between time points and time intervals, the higher the reference value of the data change characteristics within the time intervals. Based on the analysis in step S302, it is known that the greater the stability factor of a time interval at a given frequency, the more likely it is core magnetostrictive noise. Therefore, the product of the distance weight between each time point and each time interval and the stability factor of each time interval at each frequency is used as the weighted stability factor between each time point and each time interval at each frequency. The normalized value of the mean of the weighted stability factors between each time point and all time intervals at each frequency is used as the time-frequency stability index of each time point at each frequency. The larger the time-frequency stability index, the better the time-frequency stability at that time point and the lower the probability of drift, and the greater the likelihood of being affected by converter transformer core magnetostrictive noise. Normalization is a technique well-known to those skilled in the art. The normalization function can be linear normalization or standard normalization, etc., and the specific normalization method is not limited here.

[0089] At this point, we can obtain the time-frequency stability index at each frequency at each moment, where the frequency specifically refers to 100Hz, 200Hz, 300Hz, 400Hz, ..., 700Hz.

[0090] Because core vibrations caused by magnetostriction generate energy simultaneously at the fundamental frequency and harmonics, while external interference noise typically only varies at individual frequencies or random frequency bands, the synchronous variation characteristics of the fundamental frequency and harmonics can be considered a typical logical basis for determining the existence of core vibration noise. When the time-frequency stability characteristics of the fundamental frequency and its multiple integer multiples of harmonics exhibit stable temporal simultaneity changes in time-frequency analysis, it indicates that these frequency components are more likely to originate from the core magnetostriction effect. Therefore, further analysis of the differences in time-frequency stability indices at different frequencies at the same time, as well as the frequency ranking characteristics, determined the multi-frequency synchronization indices corresponding to each time point.

[0091] Preferably, in one embodiment of the present invention, the method for obtaining multi-frequency synchronization indicators includes:

[0092] In the time-frequency graph of the target location, the difference between each frequency and the preset frequency (referring to the fundamental frequency of 100Hz) is negatively correlated and normalized. This normalized value is used as the attention weight for each frequency. The larger the attention weight, the closer the frequency is to the fundamental frequency, and the more attention it needs to receive. This negative correlation mapping and normalization can be performed using the formula... ,in, Let x represent an exponential function with the natural constant e as the base, and let x represent the independent variable.

[0093] At the same time, the absolute value of the difference between the time-frequency stability index of each frequency and the preset frequency (fundamental frequency 100Hz) is calculated. The smaller the absolute value of the difference, the more consistent the time-frequency stability index between that frequency and the fundamental frequency, and the higher the synchronization. Therefore, the absolute value of the difference is negatively correlated and normalized to correct the logical relationship, obtaining the similarity factor between each frequency and the fundamental frequency. The larger the similarity factor, the more consistent the time-frequency stability between the two frequencies at that time. The negative correlation mapping and normalization process here can be performed using the formula... ,in, Let x represent an exponential function with the natural constant e as the base, and let x represent the independent variable.

[0094] Finally, at each time step, the product of the attention weight corresponding to each frequency and the similarity factor between each frequency and the preset frequency is used as the weighted similarity value. The mean of the weighted similarity values ​​between the preset frequency and all frequencies at each time step is normalized and used as the multi-frequency synchronization index for each time step. The larger the multi-frequency synchronization index, the more similar the time-frequency stability of the fundamental frequency is to other frequencies, and the more likely it is to be affected by core magnetostriction noise. Normalization is a technique well known to those skilled in the art. The normalization function can be linear normalization or standard normalization, etc., and the specific normalization method is not limited here.

[0095] Step S4: At the target location, the amplitude change index and multi-frequency synchronization index corresponding to each moment are fused together to identify the noise source at the target location.

[0096] The amplitude variation index, after being adjusted by the transformer-dominant index, makes the amplitude variation factor affected by the transformer magnetostriction effect more prominent, effectively reducing the components of other sound sources in the amplitude variation degree; while the multi-frequency synchronization index characterizes the possibility of being affected by iron core magnetostriction noise at each moment. Therefore, in this step, these two indices can be fused to identify the noise source at the target location.

[0097] Preferably, in one embodiment of the present invention, at the target location, the amplitude change index and multi-frequency synchronization index corresponding to each moment are fused to identify the noise source at the target location, including:

[0098] At the target location, the normalized product of the amplitude change index and the multi-frequency synchronization index at each time moment is used as the core noise index at each time moment. The larger the core noise index, the more likely the audio data at that time moment is noise from the DC transformer. Normalization is a technique well-known to those skilled in the art, and the normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0099] Finally, in the time-frequency graph of the target location, when the core noise index corresponding to a certain moment is greater than or equal to the preset core noise threshold, that moment is marked as the core noise moment.

[0100] It should be noted that in this embodiment of the present invention, the preset core noise threshold is 0.6. The specific value can be adjusted according to the implementation scenario and is not limited here.

[0101] In summary, acquiring current timing data and audio timing data at various monitoring points of the converter station is crucial. Current timing data reflects the equipment's operating status, while audio timing data contains noise information. Processing the audio timing data at each monitoring point yields a time-frequency diagram, converting the time-domain audio signal into a time-frequency representation. This clearly shows the sound distribution across different times and frequencies. Furthermore, analyzing the amplitude variation characteristics at the same time and frequency point across different monitoring points in the time-frequency diagrams allows for adjustments to the amplitude variation at the target point's corresponding time and frequency point. This yields an amplitude variation index at the target point's corresponding time and frequency point, facilitating subsequent analysis to determine if the noise primarily originates from the converter transformer, thereby mitigating the influence of other noise factors. Furthermore, for the magnetostrictive noise of the converter transformer core, its excitation originates from the periodic changes in the power grid's frequency flux. The frequency of this vibration does not drift over time; only the amplitude changes synchronously with the current. Therefore, in the time-frequency diagram of the target location, the correlation between the amplitude variation index and the current time-series data at each time point, as well as the differences in time-series characteristics at different frequencies, are analyzed to determine the multi-frequency synchronization index corresponding to each moment. This index reflects the possibility that the frequency components originate from the magnetostrictive effect of the core. Finally, at the target location, the amplitude variation index and the multi-frequency synchronization index corresponding to each moment are fused, integrating the noise's characteristics from different aspects. This fusion provides a more comprehensive and accurate description of the noise's features. Therefore, using the fused index to identify the noise source at the target location can improve the accuracy and reliability of noise source identification.

[0102] This invention also provides a main sound source noise identification system for flexible DC converter stations. Please refer to [link to relevant documentation]. Figure 4 The diagram shows a system block diagram, including a data acquisition module 401 for implementing step S1 in the above method embodiment; an amplitude change analysis module 402 for implementing step S2 in the above method embodiment; a frequency domain synchronization analysis module 403 for implementing step S3 in the above method embodiment; and a noise identification module 404 for implementing step S4 in the above method embodiment.

[0103] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the main source noise identification system for flexible DC converter stations and the main source noise identification method for flexible DC converter stations provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiments and will not be repeated here.

[0104] Please seeFigure 5 The diagram illustrates the structure of an electronic device according to an embodiment of the present invention, including a processor 500, a memory 501, a bus 502, and a communication interface 503. The processor 500, the communication interface 503, and the memory 501 are connected via the bus 502. The memory 501 may include a high-speed random access memory, and the bus 502 may be an ISA bus, a PCI bus, or an EISA bus, etc. The processor 500 may be an integrated circuit chip with signal processing capabilities. The memory 501 stores at least one instruction, at least one program, a code set, or an instruction set. When the processor loads and executes the at least one instruction, at least one program, a code set, or an instruction set, it implements the steps in a method for identifying the main sound source noise in a flexible DC converter station.

[0105] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0106] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for identifying the main sound source noise in a flexible DC converter station, characterized in that, The method includes: Acquire the current timing data of the converter station and the audio timing data at each monitoring point; The audio time-series data of each monitoring point is processed to obtain a time-frequency diagram; the target point is determined based on the positional relationship of the monitoring points; the amplitude variation characteristics of different monitoring points at the same time-frequency point are analyzed in the time-frequency diagrams corresponding to different monitoring points, and the amplitude variation at the time-frequency point corresponding to the target point is adjusted to obtain the amplitude variation index at the time-frequency point corresponding to the target point. In the time-frequency plot of the target location, analyze the correlation between the amplitude change index of the time-frequency point and the current time-series data within the local time range, and combine it with the time-series characteristics of the local time range to determine the time-frequency stability index corresponding to each time-frequency point; analyze the difference characteristics of the time-frequency stability index at different frequencies at the same time, as well as the frequency ranking characteristics, to determine the multi-frequency synchronization index corresponding to each time. At the target location, the amplitude change index and multi-frequency synchronization index corresponding to each moment are fused together to identify the noise source at the target location. The method for obtaining the amplitude change index includes: In the time-frequency graph of each monitoring point, at the same frequency, the normalized value of the amplitude difference between each time-frequency point and the previous time-frequency point adjacent to the time is used as the amplitude change factor corresponding to each time-frequency point. Analyze the differences in amplitude variation factors at different monitoring points under the same time and frequency points, and determine the dominant transformer indicators corresponding to each time and frequency point; In the time-frequency diagram of the target location, the normalized value of the product of the transformer dominant index and the amplitude change factor corresponding to each time-frequency point is taken as the amplitude change index corresponding to each time-frequency point. The method for obtaining the multi-frequency synchronization index includes: In the time-frequency graph of the target location, the difference between each frequency and the preset frequency is negatively correlated and normalized, and the result is used as the attention weight for each frequency. At the same time, the absolute value of the difference between the time-frequency stability index of each frequency and the preset frequency is negatively correlated and normalized, and used as the similarity factor between each frequency and the preset frequency. At each time point, the product of the attention weight corresponding to each frequency and the similarity factor between each frequency and the preset frequency is used as the weighted similarity value. The mean of the weighted similarity values ​​between the preset frequency and all frequencies at each time point is normalized and used as the multi-frequency synchronization index corresponding to each time point.

2. The method for identifying the main sound source noise in a flexible DC converter station according to claim 1, characterized in that, The methods for obtaining the main indicators of the transformer include: In the time-frequency diagrams of all monitoring points, the same time-frequency points are grouped together. In each time-frequency point group, all time-frequency points are sorted according to the preset order of the monitoring points to obtain the sorting sequence. In the sorting sequence, the difference between the amplitude change factor of the first time-frequency point and the last time-frequency point is taken as the overall change factor; the first-order difference sequence of the amplitude change factors of all time-frequency points in the sorting sequence is obtained, and the sum of the absolute values ​​of all values ​​in the first-order difference sequence is taken as the local change factor. The absolute value of the difference between the overall change factor and the local change factor is negatively correlated and normalized, and then used as the transformer dominant index for all time-frequency points in each time-frequency point group.

3. The method for identifying the main sound source noise in a flexible DC converter station according to claim 1, characterized in that, The method for obtaining the time-frequency stability index includes: In the time-frequency diagram of the target location, the time sequence is evenly divided into multiple time intervals; For any given moment, the distance weight between each moment and each time interval is determined based on the temporal relationship between that moment and each time interval. Within each time interval, the correlation between amplitude change index and current time series data is analyzed to determine the stability factor at each frequency for each time interval. The product of the distance weight between each time point and each time interval and the stability factor of each time interval at each frequency is used as the weighted stability factor between each time point and each time interval at each frequency. The normalized value of the mean of all weighted stability factors at each frequency at each time point is used as the time-frequency stability index at each frequency at each time point.

4. The method for identifying the main sound source noise in a flexible DC converter station according to claim 3, characterized in that, The method for obtaining the distance weight includes: For any given moment, the distance weight between that moment and the time interval in which that moment is located is set to 1. In each of the remaining time intervals, the minimum absolute value of the difference between the moments is negatively correlated and normalized, and this value is used as the distance weight between the moment and the time interval.

5. The method for identifying the main sound source noise in a flexible DC converter station according to claim 3, characterized in that, The method for obtaining the stability factor includes: In each time interval, the amplitude change indicators at the same frequency at all times are sorted according to time sequence to obtain the amplitude change sequence. The absolute value of the Pearson correlation coefficient between the amplitude change sequence and the corresponding current subsequence in the current time series data of the time interval is normalized to obtain the stability factor of each time interval at each frequency.

6. The method for identifying the main sound source noise in a flexible DC converter station according to claim 1, characterized in that, At the target location, the amplitude change index and multi-frequency synchronization index at each time moment are fused to identify the noise source at the target location, including: At the target location, the normalized value of the product of the amplitude change index and the multi-frequency synchronization index at each time moment is used as the core noise index at each time moment. In the time-frequency graph of the target location, when the core noise index corresponding to a certain moment is greater than or equal to the preset core noise threshold, that moment is marked as the core noise moment.

7. The method for identifying the main sound source noise in a flexible DC converter station according to claim 1, characterized in that, The method for obtaining the time-frequency diagram includes: The audio time-series data of each monitoring point are processed based on the short-time Fourier transform to obtain the time-frequency diagram of each monitoring point.

8. A main source noise identification system for a flexible DC converter station, characterized in that, A method for implementing the main sound source noise identification method according to any one of claims 1-7 includes: The data acquisition module is used to acquire the current timing data of the converter station and the audio timing data at each monitoring point. The amplitude change analysis module is used to process the audio time-series data of each monitoring point to obtain a time-frequency diagram; determine the target point based on the positional relationship of the monitoring points; analyze the amplitude change differences of different monitoring points at the same time-frequency point in the time-frequency diagrams corresponding to different monitoring points, adjust the amplitude change at the time-frequency point corresponding to the target point, and obtain the amplitude change index at the time-frequency point corresponding to the target point. The frequency domain synchronization analysis module is used to analyze the correlation between the amplitude change index of time and frequency points and the current time series data within a local time range in the time-frequency graph of the target point, and combine it with the time series characteristics of the local time range to determine the time-frequency stability index corresponding to each time-frequency point; analyze the difference characteristics of time-frequency stability index at different frequencies at the same time, as well as the frequency ranking characteristics, to determine the multi-frequency synchronization index corresponding to each time. The noise identification module is used to fuse the amplitude change index and multi-frequency synchronization index at each time point to identify the noise source at the target point.

Images