Pressure sensitive device damage diagnostic system in conjunction with pulse testing

Abstract

The present application relates to the technical field of electric fault detection, in particular to a pressure sensitive device damage diagnosis system combined with pulse test, comprising the following modules: an original signal acquisition module, inputting pulse sequences to the pressure sensitive device, recording voltage response signals and current response signals under each pulse sequence respectively, and generating a response signal set; based on the response signal set, wavelet decomposition is performed on the voltage response signals and the current response signals. In the present application, the pulse sequences are inputted and the voltage response signals and the current response signals under each pulse sequence are recorded, so that the data acquisition covers different working states and reduces the feature loss caused by a single test point. The voltage response signals and the current response signals are divided into different frequency intervals by using the wavelet decomposition method, which is helpful to separate the high-frequency and low-frequency features, so that the local abnormal signals are not covered by the global features, and the analysis ability for short-time fault features is enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrical fault detection, in particular to a pressure-sensitive device damage diagnosis system combined with pulse testing. BACKGROUND

[0002] The pressure-sensitive device damage diagnosis method is a fault detection technology for pressure-sensitive resistors, aiming to evaluate the damage type, failure mode and performance degradation of the pressure-sensitive device through analysis of current and voltage response signals.

[0003] The prior art based on simple analysis of current and voltage response signals makes it difficult to fully extract fault characteristics, especially in high-frequency signal disturbance or transient fault conditions, which is easily disturbed by noise, making it difficult to distinguish damage signs. At the same time, different frequency components of the signal cannot be analyzed separately, and local abnormal signals are easily covered by overall trends, especially in the early stage of device damage, small characteristic changes are easily ignored, reducing the sensitivity of fault detection. Therefore, improvement is needed. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art and to provide a pressure-sensitive device damage diagnosis system combined with pulse testing.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme, a pressure-sensitive device damage diagnosis system combined with pulse testing, comprising the following modules:

[0006] The original signal acquisition module inputs a pulse sequence to the pressure-sensitive device, records the voltage response signal and the current response signal under each pulse sequence respectively, and generates a response signal set; based on the response signal set, wavelet decomposition is performed on the voltage response signal and the current response signal to obtain a time-frequency feature matrix;

[0007] The signal wavelet decomposition module calculates the instantaneous frequency and phase value of the high-frequency coefficient array and the low-frequency coefficient array based on the time-frequency feature matrix to generate a frequency-phase vector; based on the frequency-phase vector, the instantaneous frequency and phase value are reconstructed into a frequency sequence according to the time sequence, and the frequency sequence is subjected to Hilbert transform to generate a fault feature spectrum;

[0008] The fault feature spectrum generation module segments the amplitude spectrum and the phase spectrum on the time axis based on the fault feature spectrum, calculates the linear fitting parameters of each segment of the amplitude spectrum and the phase spectrum, and generates a time-domain fitting parameter set; based on the time-domain fitting parameter set, the linear fitting parameters are combined with the voltage-to-current ratio in the current time segment to establish a characteristic drift curve;

[0009] The fault diagnosis and mapping module, based on the feature drift curve, combines the threshold drift amount with the voltage response signal and the current response signal of the current time period to generate a combined feature value, calculates the difference between the combined value and the standard threshold value based on the combined feature value, discriminates and classifies the damage type of the pressure sensitive device according to the difference, and establishes a damage type mapping table.

[0010] Preferably, the obtaining step of the response signal set is:

[0011] The pressure sensitive device is connected with the pulse generator, the pulse sequence is started to emit, the voltage response signal and the current response signal caused by each pulse sequence are recorded to obtain preliminary response data;

[0012] Based on the preliminary response data, the voltage response signal and the current response signal are classified and arranged to obtain comprehensive response data;

[0013] Based on the comprehensive response data, all voltage response signals and current response signals are integrated to form a response signal set.

[0014] Preferably, the obtaining step of the time-frequency feature matrix is:

[0015] Based on the response signal set, the voltage response signal and the current response signal are respectively wavelet decomposed, the wavelet coefficients of each signal at different scales are extracted, the wavelet decomposition coefficients are classified according to the preset frequency interval, and the high-frequency coefficient group and the low-frequency coefficient group are obtained;

[0016] According to the high-frequency coefficient group and the low-frequency coefficient group, the energy distribution value of each group is calculated, and the calculation formula is:

[0017]

[0018] Wherein, E k represents the energy distribution value of the kth group, C k,m represents the absolute value of the mth wavelet decomposition coefficient in the kth group, D k,m represents the deviation term of the mth coefficient in the kth group, and QM is the number of wavelet coefficients.

[0019] Based on the energy distribution value, a time-frequency feature matrix is constructed, and the energy distribution value of each group is filled in the time-frequency feature matrix.

[0020] Preferably, the obtaining step of the frequency phase vector is:

[0021] Based on the time-frequency feature matrix, the components of the high-frequency coefficient group and the low-frequency coefficient group are extracted, the instantaneous amplitude change rate of each component is calculated, and the instantaneous change feature is obtained by combining the time information to obtain the instantaneous change feature data;

[0022] Based on the instantaneous change characteristic data, instantaneous frequency and phase values are calculated, time sequence information of each component is sorted, and instantaneous frequency and phase change conditions of each component are associated to obtain instantaneous frequency data and phase data;

[0023] Based on the instantaneous frequency data and phase data, data is matched and classified according to corresponding frequency intervals, and corresponding instantaneous frequency and phase values are combined into a frequency phase vector.

[0024] Preferably, the acquisition step of the fault characteristic spectrum is:

[0025] Based on the frequency phase vector, instantaneous frequency and phase values corresponding to each time point are extracted, sorted in time sequence order, and interpolated to complete, and the frequency points with abnormal changes are smoothed to obtain a reconstructed frequency sequence.

[0026] The reconstructed frequency sequence is subjected to Hilbert transform to calculate the transformed instantaneous amplitude spectrum value, and the calculation formula is:

[0027]

[0028] Wherein, A j is the instantaneous amplitude spectrum value corresponding to the jth frequency point, P j is the original component value in the frequency sequence, is the component after applying Hilbert transform to P j

[0029] Based on the instantaneous amplitude spectrum value, a feature matrix including amplitude information is constructed, the amplitude value corresponding to each time point is classified according to frequency distribution, and the classified data is arranged in time sequence to form a fault characteristic spectrum.

[0030] Preferably, the acquisition step of the time domain fitting parameter set is:

[0031] Based on the fault characteristic spectrum, the time axis is divided into multiple time periods, the amplitude data and phase data in each time period are extracted to form segmented amplitude spectrum and phase spectrum.

[0032] According to the segmented amplitude spectrum and phase spectrum, the linear fitting parameters of each time period are calculated, and the calculation formula is:

[0033]

[0034] Wherein, L k represents the linear fitting parameter of the kth period, p k,m represents the mth amplitude data point in the kth period, q k,m represents the mth phase data point in the kth period.

[0035] ​Based on the linear fitting parameters, the fitting parameters of all time periods are arranged in time sequence, and a complete time-domain fitting parameter set is constructed, and the fitting parameters of each time period are associated according to the distribution relationship of amplitude spectrum and phase spectrum to generate the time-domain fitting parameter set.

[0036] Preferably, the feature drift curve acquisition step is:

[0037] Based on the time-domain fitting parameter set, the linear fitting parameters of each time period are extracted, and the voltage-current ratio of the corresponding time period is obtained, and a sliding window method is used for smoothing processing to form time period matching data;

[0038] According to the time period matching data, the threshold drift of each time period is calculated, and the calculation formula is:

[0039]

[0040] Wherein, T k represents the threshold drift of the kth time period, P k,m represents the mth linear fitting parameter value in the kth time period, R k,m represents the mth voltage-current ratio in the kth time period, and M represents the total number of data points in the kth time period.

[0041] Based on the threshold drift, the drift data of all time periods are arranged in time sequence, and a feature drift curve including the drift features of all time periods is constructed.

[0042] Preferably, the acquisition step of the combined feature value is:

[0043] According to the feature drift curve, the combined feature value is calculated, and the calculation formula is:

[0044]

[0045] Wherein, TC k represents the combined feature value of the kth time period, U k (x) represents the voltage response signal function of the kth time period, I k (x) represents the current response signal function of the kth time period, U k,j represents the jth voltage response signal value of the kth time period, I k,j represents the jth current response signal value of the kth time period, T k,j represents the jth threshold drift of the kth time period, J represents the total number of data points in the kth time period, and Q is the total variation range of the signal in the time period.

[0046] Preferably, the acquisition step of the damage type mapping table is:

[0047] Obtaining standard threshold data, arranging the combined feature values in time sequence to form the combined feature values and the standard threshold data;

[0048] Based on the combined feature values and the standard threshold data, the difference amount of each time period is calculated, the calculated difference amount is matched according to the time axis, and all the difference amounts are arranged in time sequence of data points to obtain the difference amount between the combined value and the standard threshold value;

[0049] Based on the difference amount between the combined value and the standard threshold value, a damage type discrimination rule is defined, all data points corresponding to the damage type are classified according to a preset classification threshold, and data of different damage types are mapped to corresponding categories to establish a damage type mapping table.

[0050] Compared with the prior art, the advantages and positive effects of the present application are that:

[0051] In the present application, the input pulse sequence and the voltage response signal and the current response signal under each pulse sequence are recorded, so that the data acquisition covers different working states, and the feature loss caused by a single test point is reduced. The voltage response signal and the current response signal are divided into different frequency intervals by using wavelet decomposition, which is helpful to separate high-frequency and low-frequency features, so that local abnormal signals are not covered by global features, and the analysis ability of short-time fault features is enhanced. The instantaneous frequency and phase value are calculated and the frequency phase vector is constructed, so that the dynamic evolution process of the signal is easier to analyze, and the time series reconstruction is combined to make the instantaneous features form a continuous change trajectory, and the traceability of the fault features is improved. The fault feature spectrum is obtained by using Hilbert transform, so that the change of signal energy distribution is more intuitive, and the discrimination ability of damage types is enhanced. The amplitude spectrum and the phase spectrum are segmented on the time axis, the linear fitting parameters of each segment are calculated, so that the local trend change of the signal is clear and visible, and the analysis ability of complex fault modes is improved. The feature drift curve is established by combining the voltage-current ratio, so that the fault features can form a trend in the time dimension, and the interference of single-time abnormal points on the discrimination result is avoided. The threshold drift amount, the voltage response signal and the current response signal are fused by using numerical combination, so that the fault features are more comprehensive, and the misjudgment caused by a single feature is reduced. The difference amount between the combined value and the standard threshold value is calculated, so that the damage classification has a clearer quantitative basis, and the damage type mapping table is constructed, so that different damage modes form a traceable standard classification, the automation degree of damage diagnosis is improved, and the demand for manual intervention is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 The present application is a step schematic diagram. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0054] Please refer to Figure 1 The present application provides a technical scheme, a pressure-sensitive device damage diagnosis system combined with pulse test, comprising the following modules:

[0055] The original signal acquisition module inputs a pulse sequence to the pressure-sensitive device, records the voltage response signal and the current response signal under each pulse sequence respectively, and generates a response signal set; based on the response signal set, the voltage response signal and the current response signal are wavelet decomposed to obtain a time-frequency feature matrix;

[0056] The signal wavelet decomposition module calculates the instantaneous frequency and the phase value based on the time-frequency feature matrix, generates a frequency-phase vector, reconstructs the instantaneous frequency and the phase value into a frequency sequence according to the time sequence based on the frequency-phase vector, performs Hilbert transform on the frequency sequence, and generates a fault feature spectrum;

[0057] The fault feature spectrum generation module segments the amplitude spectrum and the phase spectrum on the time axis based on the fault feature spectrum, calculates the linear fitting parameters of each segment of the amplitude spectrum and the phase spectrum, and generates a time-domain fitting parameter set; based on the time-domain fitting parameter set, the linear fitting parameters are combined with the voltage-to-current ratio in the current time segment to establish a characteristic drift curve;

[0058] The fault diagnosis and mapping module combines the threshold drift amount with the voltage response signal and the current response signal of the current time segment based on the characteristic drift curve, generates a combined feature value, calculates the difference between the combined value and the standard threshold value based on the combined feature value, discriminates and classifies the damage type of the pressure-sensitive device according to the difference, and establishes a damage type mapping table.

[0059] The acquisition step of the response signal set is:

[0060] The pressure-sensitive device is connected with the pulse generator, the pulse sequence emission is started, the voltage response signal and the current response signal caused by each pulse sequence are recorded, and preliminary response data are obtained;

[0061] Based on the preliminary response data, the voltage response signal and the current response signal are classified and arranged to obtain comprehensive response data;

[0062] Based on the comprehensive response data, all voltage response signals and current response signals are integrated to form a response signal set.

[0063] Specifically, refer to the use specification of the pressure sensitive device and combine the technical data of the pulse generator to first confirm the rated bearing range of the connecting cable and the rated voltage and current limit of the pressure sensitive device, for example, control the voltage in the range of 0V to 24V and the current in the range of 0A to 5A, and after observing that the connecting terminal meets the device specification, set the actual number of pulses to 20 and the emission interval of each pulse to 1 second, so as to facilitate the comparison of the response condition at each emission time through such fixed time interval, and then check the corresponding use manual of the pressure sensitive device and the pulse generator before starting the pulse emission, and make clear that the safety protection value stated therein is based on multiple actual measurements and industry experience, for example, the pulse voltage threshold value is generally taken as 80% of the device calibration range as the initial set value, if the voltage measured at any time during the emission process exceeds the value, it is determined that the pulse intensity is too high, and the detection needs to be paused and the output position is corrected again, at the same time, the voltage response signal and the current response signal caused by each pulse are recorded in real time at the collection end, the specific method is to read the voltage and current curves generated by each pulse in segments at a sampling frequency of 10kHz, and compare with the intervals set in the front, for example, judge whether there is an oscillation overrun in the range of 0V to 24V, and whether there is a current surge in the range of 0A to 5A, if it is found that the single pulse peak value or average value collected is abnormal, it is prompted in a marked way for further investigation, and finally the signal generated by each pulse is saved in a temporary database, and the preliminary response data is obtained after being summarized.

[0064] Compare the preliminary response data, and confirm how the rated range of each voltage response signal and current response signal is established through multiple experiments and industry experience when referring to the relevant device data, for example, compare the voltage data with the range of 0V to 24V one by one, and compare the current data with the range of 0A to 5A one by one, if it is found that any record is in the suspicious interval or exceeds the nominal upper limit set by experience in advance, mark the record as an object that needs to be observed, then sort and arrange all records according to voltage and current, among which the voltage signal classification is mainly based on interval division, and the high frequency interval or low frequency interval is determined according to the number of records, the current signal is operated in the same way, but at the same time, the current transient over-shoot phenomenon is evaluated by combining the maximum deviation value of the current, these specific numerical ranges can be referred to the actual measured average value plus or minus the corresponding error correction amount, all sorting processes are sorted and grouped by fixed method, to avoid missing potential abnormalities and maintain the comparability of the overall data, after all classification results are completed, the classified voltage response signal and current response signal are summarized and recorded one by one, to obtain the comprehensive response data.

[0065] After obtaining the comprehensive response data, the voltage response signals and the current response signals in each category are rearranged and integrated. The specific steps are as follows: first, all records exceeding the device experience setting range are screened out and listed separately, for example, if a voltage record is higher than 24V or lower than 0V, it is placed in the abnormal list separately, if the current record exceeds 5A or is lower than 0A, it is also placed in the abnormal list, then the records within the range of 0V to 24V and 0A to 5A are paired in time sequence to ensure that the voltage and current collected at the same time can accurately correspond and form a comparable integrated data set, then the voltage data and the current data at the corresponding time point are verified together, for example, the ratio of each record is calculated and the interval is divided according to the experience threshold, if the ratio exceeds the mean value plus or minus a certain deviation, it is marked as a possible abnormal point, after completing this batch of processing, all normal interval data and abnormal list are placed in the same table for use, and finally the response signal set is formed.

[0066] The acquisition step of the time-frequency feature matrix is:

[0067] Based on the response signal set, the voltage response signal and the current response signal are respectively wavelet decomposed, the wavelet coefficients of each signal at different scales are extracted, the wavelet decomposition coefficients are classified according to the preset frequency interval, and the high-frequency coefficient group and the low-frequency coefficient group are obtained;

[0068] According to the high-frequency coefficient group and the low-frequency coefficient group, the energy distribution value of each group is calculated, and the calculation formula is:

[0069]

[0070] Among them, E k represents the energy distribution value of the kth group, C k,m represents the absolute value of the mth wavelet decomposition coefficient in the kth group, D k,m represents the deviation term of the mth coefficient in the kth group, and QM is the number of wavelet coefficients.

[0071] Based on the energy distribution value, the time-frequency feature matrix is constructed, and the energy distribution value of each group is filled in the time-frequency feature matrix.

[0072] Specifically, based on the previously obtained response signal set, the corresponding characteristics of the voltage response signal and the current response signal are compared by referring to the recorded pulse test conditions, the amplitude change of each signal in the frequency range of 0 Hz to 100 kHz is read, and then the wavelet function type of each signal at different scales is determined by combining the segmented sequence after frequency domain sampling. Then, referring to the frequency interval division, for example, defining 20 Hz and below as the low frequency band, 20 Hz to 500 Hz as the sub-low frequency band, 500 Hz to 5 kHz as the medium frequency band, and 5 kHz to 100 kHz as the high frequency band, the wavelet coefficients corresponding to each signal are extracted in each interval according to this interval, and the occurrence time and amplitude absolute value are recorded. When the amplitude value in any paragraph exceeds the empirical threshold value statistically obtained in the previous experiment, the paragraph is marked as an abnormal segment. The empirical threshold value is obtained through multiple pulse tests and on-site monitoring. Specifically, based on the pulse response record, the average upper limit of the peak amplitude of each paragraph in nearly ten groups of experimental data is summed up and a safety margin is added to define the comparison standard. Then, the marked abnormal segment is compared again to confirm whether the amplitude deviation is further increased. If the value still exceeds the set range of the empirical threshold value, the paragraph is recorded to distinguish potential failure factors. Finally, among all the decomposition results obtained through the above operations, all the wavelet coefficient sets with frequency distribution falling in the high frequency band and all the wavelet coefficient sets with frequency distribution falling in the low frequency band are extracted and combined to form the high frequency coefficient group and the low frequency coefficient group, respectively.

[0073] The formula has the beneficial effect of combining the comprehensive characteristic quantities of and , and introducing for proportional measurement in the denominator, so that the energy distribution value takes into account both the absolute amplitude and the deviation, avoiding the local imbalance problem caused by simply using the square or absolute value.

[0074] C k,m The parameter acquisition step is C k,m , which refers to the absolute value of the mth wavelet decomposition coefficient in the kth group. The coefficient comes from the data component at the corresponding hierarchical scale after wavelet decomposition of the response signal. First, the wavelet scale is determined according to the number of decomposition layers, and then the coefficients at each time point at this scale are read in turn, and their absolute values are obtained to obtain C k,m . If the original value of the coefficient at a certain time is -2.3 in actual measurement, the value of C k,m at that time is 2.3. The specific acquisition process depends on the multi-resolution decomposition means of the signal in the previous stage and the determination of the wavelet basis function. For example, use the recorded basis function morphology under different boundary extensions, read the transformed coefficient matrix with a self-defined program, then extract the kth value of the mth value in time sequence and take the absolute value form, and finally obtain C k,m, for example, a certain pulse signal measured in the field contains 8 coefficient values at 500 Hz layering, and after taking the absolute value in turn, it constitutes C k,m The specific sequence of C

[0075] D k,m The acquisition step of the parameter is D k,m , which represents the deviation term of the mth wavelet coefficient in the kth group. It is necessary to subtract the average coefficient value at this layering scale or the previous cumulative coefficient reference value from the actual observation value to obtain the deviation, which is updated by continuously accumulating the average value of the coefficients in the same group in the pulse experiment. k,m k,m With the change, the specific method is to update the average value MM k of the waveform recorded in the previous round based on the same group, then calculate the difference between the original value of the measured coefficient and MM k , for example, the original value of the coefficient at a certain time is 2.5, and the average value of the previous round is 2.2, then D k,m is 0.3, and repeating this process for all times can form the sequence of D k,m , if someone wants to obtain the deviation term under other conditions, it can also be achieved by corresponding the same pulse test and the same layering scale, and then subtracting the average value of the historical statistics in the same group from the measured coefficient value, which can obtain the value of D k,m .

[0076] The acquisition step of the QM parameter is that QM represents the total number of wavelet coefficients in the kth group, which is usually determined by the number of decomposition layers and the number of time sequence sampling points, for example, when measuring the pulse, 400 sampling points are set for each signal, and through 5-layer wavelet decomposition, a certain number of coefficients will be generated at each layer. If 12 coefficients are obtained by merging at the kth group at this level, QM can be accumulated to 12. This accumulation process may include the merging of multiple sub-bands to include the total number of coefficients in the same frequency band. In addition, when some layers have repeated frequency bands, the repeated items need to be removed. Finally, the number of effective wavelet coefficients calculated is QM. In specific operation, the number of available coefficient points of each layer can be recorded in advance and added, or the total number of coefficients contained in this group can be directly counted after the transformation is completed. These methods can determine the value of QM.

[0077] ​The k parameter is obtained by grouping the frequency bands or decomposition levels, and if five-layer wavelet decomposition is performed, k is usually selected from 1 to 5. However, in applications requiring secondary subdivision of high frequency bands, the range of k is expanded to 10 or even more, and the value of k is determined according to the frequency division strategy. In field measurement, the frequency is first divided into several intervals, and then wavelet decomposition is performed on each interval. Each interval is recorded as a group, and k corresponds to the interval number. For example, 0-500 Hz is set as the first group, 500-5,000 Hz is set as the second group, and so on.

[0078] The m parameter is obtained by grouping the frequency bands or decomposition levels, and if five-layer wavelet decomposition is performed, k is usually selected from 1 to 5. However, in applications requiring secondary subdivision of high frequency bands, the range of k is expanded to 10 or even more, and the value of k is determined according to the frequency division strategy. In field measurement, the frequency is first divided into several intervals, and then wavelet decomposition is performed on each interval. Each interval is recorded as a group, and k corresponds to the interval number. For example, 0-500 Hz is set as the first group, 500-5,000 Hz is set as the second group, and so on.

[0079] Substituting the parameters into the formula, 9.24 is obtained. This result indicates that when E2 is close to 9.24, the energy distribution in the second frequency interval is relatively concentrated under the current pulse test environment. If E2 further increases to above 30, the absolute coefficient cube sum and deviation accumulation of this group are abnormally high. The E2 of other groups can also be calculated in the same way. k Thus, the overall energy distribution is further determined.

[0080] Based on the energy distribution values obtained above, a blank structure is created in the predetermined time-frequency feature matrix area to accommodate the distribution records of each group. Then, the calculation results of each group are traversed one by one, and the corresponding E k The numerical values are mapped to the row and column positions of the time-frequency feature matrix according to the group number index and time sequence, and are compared and analyzed. In order to ensure that each row and each column in the matrix is aligned with the actual segmentation order, the index corresponding to different time segments needs to be assigned in the row direction of the matrix, such as the segmentation number between 1-2 seconds is recorded as the first row, the segmentation number between 2-3 seconds is recorded as the second row, and so on. The column direction is expanded according to the order of the first group and the second group. If the E k value calculated by any group in a certain time exceeds the upper limit of the range obtained in the previous pulse test, such as the upper limit of the range is defined as 50, a special mark is made at the corresponding position of the matrix during filling, and these marks are recorded in a separate warning list in the subsequent summary step. The determination of the upper limit of the range is mainly through statistical analysis of the maximum E kAnd combined with a certain degree of fluctuation margin, so as to determine a proper distribution boundary, for example, in 10 consecutive trials to select each group E k The highest value plus 10% compensation, the final value will be set as the upper limit of the energy distribution, when confirming that all rows and columns in the matrix have completed the corresponding E k Value writing, it can be regarded as the time-frequency feature matrix has been constructed, so that the energy distribution value of each group is correctly filled in the corresponding position.

[0081] The acquisition step of the frequency phase vector is:

[0082] Based on the time-frequency feature matrix, the high-frequency coefficient group and the low-frequency coefficient group are extracted respectively, the instantaneous amplitude change rate is calculated for each component, and the instantaneous change characteristic is obtained by combining the time information, to obtain the instantaneous change characteristic data;

[0083] Based on the instantaneous change characteristic data, the instantaneous frequency and phase values are calculated, the time sequence information of each component is arranged, and the instantaneous frequency and phase change of each component is associated, to obtain the instantaneous frequency data and phase data;

[0084] Based on the instantaneous frequency data and phase data, the data is matched and classified according to the corresponding frequency interval, and the corresponding instantaneous frequency and phase values are combined into a frequency phase vector.

[0085] Specifically, based on the time-frequency feature matrix, the components of the high-frequency coefficient array and the low-frequency coefficient array are extracted respectively, the instantaneous amplitude change rate of each component is calculated, and the instantaneous change feature is obtained by combining the time information. First, the values of each time slice in the high-frequency coefficient array and the low-frequency coefficient array need to be compared point by point, the amplitude difference of each component at adjacent time points is recorded, and then the ratio of the difference value to the current amplitude is used to determine the instantaneous amplitude change rate. For example, if the amplitude of a component is 3.2 at t1 and 3.5 at t2, the instantaneous increment of the component can be calculated as 3.5 minus 3.2 equals 0.3, and then 0.3 divided by 3.2 is about 0.0937. If this instantaneous amplitude change rate exceeds the empirical threshold obtained from previous experiments, for example, take 0.1 as the judgment standard and determine it by comparing the results of multiple rounds of detection, then the component will be marked as a rate deviation interval. After the instantaneous amplitude change rate of each component is calculated, the specific change trend of the component in the adjacent time period is also determined according to the corresponding time label. This part of the value can be accumulated and compared with the total sum of each time, and whether the total sum exceeds the predetermined upper limit. For example, if it is found through continuous observation that the cumulative change rate has reached 2.5 in a certain period of time, and the upper limit of the stable interval obtained through repeated experiments is usually not more than 1.5, then further investigation of the component is needed. In this way, the instantaneous amplitude change of each component can be effectively collected and recorded and distinguished with time information. Finally, the time sequence change data obtained is sorted to obtain the instantaneous change feature data.

[0086] Based on the instantaneous change characteristic data, the instantaneous frequency and phase values are calculated, the time sequence information of each component is arranged, and the instantaneous frequency and phase change of each component is associated. The instantaneous change characteristic data needs to be split into independent sequences according to the component attribution, then the amplitude and time position of each component at each time are read, the interval between adjacent peaks or adjacent zero-crossing points is determined by measurable means such as zero-crossing or differentiation, the interval is divided by the corresponding time length to obtain the instantaneous frequency, for example, when the time interval between two peaks of a component is 0.002 seconds, the instantaneous frequency can be about 500 Hz, and the specific value is recorded for comparison with similar components. If the instantaneous frequency exceeds the upper limit of the frequency range obtained in the previous field measurement, for example, according to the peak distribution of the components collected by ten pulse tests, most of them fall between 100 Hz and 3 kHz, if the current component appears above 5000 Hz, further analysis is needed to determine whether abnormal oscillation occurs by analyzing the frequency trend of the component on the entire time axis and combining the amplitude and time change. As for the calculation of the phase value, the relative position relationship between the adjacent two points can be used to achieve it. When the time sequence corresponding to each peak is determined, the reference point of the component is compared with the reference phase. If the difference is 180 degrees, it means that there is a large misalignment between the current interval and the reference period of the component. After arranging these information continuously, the instantaneous phase change data of each component is formed. Finally, the instantaneous frequency data and phase data are matched and merged in the same record by corresponding to the sequence index, and the instantaneous frequency data and phase data are obtained.

[0087] Based on the instantaneous frequency data and phase data, the data is matched and classified according to the corresponding frequency interval. First, the specific division standard of frequency is confirmed, for example, 0 Hz to 500 Hz is recorded as low frequency, 500 Hz to 5 kHz is recorded as medium frequency, and 5 kHz to 100 kHz is recorded as high frequency. The actual division can refer to the observation results of the previous multiple pulse tests and the rated parameters of the equipment. When matching the phase data for each instantaneous frequency data, the time index and amplitude correlation degree of the two are checked. If the instantaneous frequency is 150 Hz and the phase is 45 degrees at the same time, the data is classified into the low frequency segment list and matched with the subsequent time. If the instantaneous frequency of the same component rises to 3 kHz and the phase changes to 120 degrees at the next time, it is classified into the medium frequency record. After traversing all the time and completing the matching, the corresponding instantaneous frequency and phase value in each record are combined to form a binary vector, and then the time marker is marked in the segmented index table and sequentially sorted. During this period, it is also necessary to check whether there is repeated or missing time segment information. If there is a missing, the average frequency and phase of the adjacent time can be used to fill in the record. After confirming that there is no obvious error by comparison again, the matched instantaneous frequency and phase value are combined into a frequency phase vector.

[0088] The acquisition steps of the fault characteristic spectrum are as follows:

[0089] Based on the frequency phase vector, the instantaneous frequency and phase values corresponding to each time point are extracted, sorted in time sequence order, and interpolated to complete, and the frequency points with abnormal changes are smoothed to obtain the reconstructed frequency sequence;

[0090] The reconstructed frequency sequence is subjected to Hilbert transform, and the transformed instantaneous amplitude spectrum value is calculated, and the calculation formula is:

[0091]

[0092] Wherein, A j is the instantaneous amplitude spectrum value corresponding to the jth frequency point, P j is the original component value in the frequency sequence, is the component after applying Hilbert transform to P j

[0093] Based on the instantaneous amplitude spectrum value, a feature matrix including amplitude information is constructed, the amplitude value corresponding to each time point is classified according to frequency distribution, and the classified data is arranged in time sequence to form a fault feature spectrum.

[0094] Specifically, based on the frequency phase vector obtained in the foregoing, the instantaneous frequency and phase values corresponding to each time point are extracted and preliminarily sorted, and then all records are arranged one by one in time sequence order, and the missing time data is interpolated and completed by the average value of adjacent time slices. The interpolation process needs to determine the left and right boundary time points of the interpolation interval and analyze the existing instantaneous frequency and phase data at both ends, and then the values of the two points are taken for linear or polynomial point-by-point calculation. For example, a plurality of step points are set in the interval of 0.01 seconds of adjacent observation time difference, and the instantaneous frequency and phase of these points are interpolated and filled. If the frequency of some time point is detected to increase by more than the range of the pre-established empirical threshold, for example, the threshold range is 50Hz to 5000Hz and is obtained by analyzing the distribution of the sampling data of multiple pulse experiments, when the frequency of any time point exceeds 5000Hz, it is marked as an abnormal record. The specific increase ratio of the frequency increment of the record is calculated by comparing the frequency increment of the record with that of the time before and after the record. If the increment is greater than 50% of the original value and lasts for more than twice sampling, it can be determined as a local abnormal change point. Then, the instantaneous frequency is smoothed in this time period. The smoothing method can adopt the median value processing of adjacent 3 or 5 time points, and the value deviating obviously is corrected to the value close to the surrounding value. If the phase value also jumps at the same time point, it can be checked synchronously and processed similarly, and the time point where the outlying amplitude appears is marked in the record. After the interpolation and smoothing of all time data are completed, the finally corrected frequency record is merged into a continuous sequence in time sequence to obtain the reconstructed frequency sequence.

[0095] The beneficial effect of the formula is that​ The original component P j After performing the Hilbert transform, the real part of the original frequency component is combined with its corresponding imaginary part, thereby obtaining a more intuitive instantaneous intensity at the amplitude spectrum level.

[0096] P j The parameter acquisition step is:

[0097] This parameter is the value of the original component in the frequency sequence, representing the frequency observed at the jth time index or the amplitude record related to the frequency, which is usually obtained by subdividing the segmented data during the pulse test. Each segment of data is accompanied by a specific time stamp, which can be used to accurately locate the actual sampling time corresponding to the jth observation value. The frequency value or amplitude detected at this time is marked as P j If higher precision is required, random noise can be eliminated by averaging or multiple repeated tests on the original sampling, and finally a stable and reliable P j sequence is formed. For example, when j = 3, the original record at this observation time can be read as 2.0.

[0098] The parameter acquisition step is:

[0099] This parameter refers to the component obtained by applying the Hilbert transform to P j , which is implemented by performing a fast transform operation on the continuous frequency sequence. It requires Fourier domain processing of the entire frequency sequence and introduction of a phase shift before converting back to the time domain. The result obtained is the component value corresponding to P j after the Hilbert transform. The numerical size also depends on the sampling accuracy and frequency range division. If is to be obtained, a Fourier transform can be performed on the record for j = 3 at the sampling time and combined with the existing frequency band index for operation. Finally, the imaginary part corresponding to the index is extracted during the inverse transform and taken as In the following example, the value of will be listed as 1.5.

[0100] The j parameter acquisition step is:

[0101] This parameter is an index number used to identify the position of each record point in the entire frequency sequence. j starts from 1 and increases sequentially in time order. For example, in the observation of the pulse signal in the interval from 0 to 2 seconds, a total of 10 points are sampled, and the value of j ranges from 1 to 10. Each j corresponds to an independent P j and observation value.

[0102] Substituting the parameters into the formula gives 2.5, which indicates that when the original component P3 = 2.0 and the Hilbert transform component When the instantaneous amplitude spectrum value A3 of the 3rd frequency point is 2.5, if A j Significant increase or decrease, that is, the corresponding change trend can be recorded in the subsequent feature matrix and summarized in the next analysis step.

[0103] Based on the instantaneous amplitude spectrum value, combined with the amplitude value and frequency distribution information corresponding to each time point, a continuous time range needs to be selected in the pulse signal sampling process, the instantaneous amplitude spectrum values at different times in this time period are read one by one, and compared with the predetermined frequency classification list, for example, 0Hz to 500Hz is divided into low frequency area, 500Hz to 5kHz is divided into medium frequency area, and 5kHz above is divided into high frequency area, then the instantaneous amplitude spectrum values are classified and placed in the corresponding data paragraph according to the frequency interval of each time point, and the corresponding frequency and amplitude are matched one by one. If it is found that the amplitude at a certain time exceeds the upper limit of the range previously obtained by cumulative observation statistics, for example, the amplitude peak obtained after several rounds of detection is 100, which is set to 110 as the upper limit by adding a 10% buffer, if the amplitude reaches 120, it is marked as exceeding the boundary value, and the identification information is saved in the subsequent comparable record summary. Then, according to the time sequence, the whole amplitude spectrum value is arranged time by time or sample by sample, so that each time node can find the corresponding amplitude and frequency combination, and the matched record is generated in a row or a column in the matrix, and the complete time sequence distribution structure is expanded row by row or column by column. The structure can be located by row and column using the index j mentioned above. Finally, each record is summarized to form a feature matrix containing amplitude information. In the subsequent matrix, the combination of different time periods and frequencies is traversed, and the distribution data associated with the fault is derived to form a fault feature spectrum.

[0104] The acquisition step of the time domain fitting parameter set is:

[0105] Based on the fault feature spectrum, the amplitude data and phase data in each time period are extracted to form the segmented amplitude spectrum and phase spectrum.

[0106] According to the segmented amplitude spectrum and phase spectrum, the linear fitting parameters of each time period are calculated, and the calculation formula is:

[0107]

[0108] Wherein, L k represents the linear fitting parameter of the kth segment, p k,m represents the mth amplitude data point in the kth segment, q k,m represents the mth phase data point in the kth segment.

[0109] Based on the linear fitting parameters, the fitting parameters of all time periods are arranged in chronological order, and a complete time-domain fitting parameter set is constructed. The fitting parameters of each time period are associated according to the distribution relationship of the amplitude spectrum and the phase spectrum to generate the time-domain fitting parameter set.

[0110] Specifically, based on the fault characteristic spectrum, first, the amplitude data and phase data obtained are split according to the chronological order of the time axis. The number of specific time periods is determined according to the number of records contained in each time in the fault characteristic spectrum obtained in the foregoing. Each period can be set to between 1 second and 2 seconds or longer. When actually dividing, it is necessary to combine the field pulse sampling rate and the continuous observation range of the device to determine whether enough amplitude data and phase data can be contained in the time period. After each period is determined, the amplitude values in the period are grouped and registered. All amplitude values belonging to the same time period are aggregated together, and the corresponding phase values are read to form a one-to-one correspondence record. Then, the amplitude and phase information in the same paragraph are compared one by one with the pre-established effective range. For example, the amplitude is compared with the interval of 0 to 100, and the phase is compared with the range of 180 degrees to 180 degrees. All parts exceeding the upper limit of the device statistics are identified as abnormal records and marked in the abnormal list of the time period. If the abnormal records exceed a certain empirical threshold, such as 10, the threshold is obtained by adding two standard deviations to the average abnormal value of each period in multiple industrial field tests, it is necessary to further check the period. Otherwise, the period is identified as a normal segment that can enter the fitting calculation process. According to this idea, the amplitude data and phase data of all time periods are processed, and the amplitude spectrum and phase spectrum in each time period are summarized as the amplitude spectrum and phase spectrum after segmentation.

[0111] The formula has the benefit of combining amplitude data and phase data in the form of product, difference cube, etc., and simultaneously incorporating absolute value and square root operation, which can measure the multiple differences between amplitude and phase in the same expression, thereby providing more abundant references for subsequent time-domain analysis.

[0112] p k,m The parameter acquisition step is:

[0113] The parameter represents the mth amplitude data point in the kth segment, which usually appears in the post-processing stage of pulse measurement, and its value is usually between 0 and 200, depending on the quantization method of the waveform amplitude of the measuring device and affected by the field environment or test configuration. By collecting the maximum amplitude or effective value of each pulse in the device operation and recording the corresponding time stamp, the amplitude values falling within the kth time range are uniformly marked as p k,m , m starts from 1 and increments to the total number of records in the paragraph. For example, multiple field measurements can find that the amplitude peak value of some pulses is 150 or higher. Whenever it is confirmed that the amplitude peak value belongs to the kth segment, it is written into pk,m , then the comparable p k,m is obtained. Taking the following example, when k = 3, m takes 1 to 4 respectively, p 3,1 = 80, p 3,2 = 120, p 3,3 = 95, p 3,4 = 110.

[0114] The acquisition step of the parameter q k,m is:

[0115] This parameter refers to the mth phase data point in the kth segment, which is usually between 180 degrees and 180 degrees. For the convenience of operation, the degrees can be converted into radians or the original angle unit can be maintained, but the unit conflict in the operation relationship with p k,m must be ensured. In field measurement, the phase difference of the pulse waveform relative to a certain reference zero point or the maximum amplitude moment is often given. If it is confirmed that this phase difference appears in the kth segment recording time, it is identified as q k,m . One pulse may correspond to multiple phase detection points, and after all are collected, the complete sequence of q k,m is formed. The acquisition must be completed in the time domain alignment in the previous step to ensure that the amplitude peak and phase information of the same pulse can be paired under the same index. Taking the following example, q 3,1 = 30, q 3,2 = 45, q 3,3 = 60, q 3,4 = 10.

[0116] The acquisition step of the parameter k is:

[0117] This parameter is used to represent the sequence number of the time segment, which is usually determined by the time axis division prepared by the researcher in advance. For example, each segment is set to 2 seconds. If the entire observation time is 10 seconds, then k is between 1 and 5. k = 1 represents the first segment 0 to 2 seconds, k = 2 represents the second segment 2 to 4 seconds, and so on. In practice, other lengths can also be used for division, as long as the same sampling strategy is maintained in each segment and the start and end boundaries of the amplitude and phase data can be distinguished, and then the k value is obtained.

[0118] The acquisition step of the parameter m is:

[0119] This parameter is used to distinguish the sequence number of each data point in the kth segment. The value starts from 1 and increases to the total number of amplitude or phase data in the segment. If a total of 50 records are obtained in the same segment, then m ranges from 1 to 50. Different m corresponds to p k,m and q k,m, other environment if keep sampling resolution same can get comparable m value sequence, m in subsequent operation plays the role of index, each amplitude and phase data with its time period k bundled together, convenient to summary to the same formula in the whole calculation.

[0120] Substitute the parameters into the formula to calculate L3=0.51, which shows that when the amplitude and phase data in the third segment are obtained according to the above operation process, L3=0.51 can be regarded as a linear fitting parameter for measuring the coupling degree between the amplitude and phase in this period, if L k The numerical value increases by more than 1.0 or 2.0 in each segment, which is an empirical upper limit, which can be identified as a potential anomaly and further investigated according to the needs in the subsequent steps.

[0121] Based on the linear fitting parameter, arrange all the fitting parameters of the time period in time sequence, and calculate L k Then sort them in order of k, for example, k=1 corresponds to the data in the first 2 seconds, k=2 corresponds to the data from 2 seconds to 4 seconds, and so on, and all L k Values are sorted on the time axis and checked for missing data. If any segment of L k Is empty, you need to check again whether the amplitude or phase information of this segment is complete. If it is confirmed to be missing, you can use the average value of the surrounding segments to fill it in the same time period, so as to ensure the continuity of the mapping between the time period and L k After that, combine the previous amplitude spectrum and phase spectrum distribution relationship to perform correlation operation. The specific method is to maintain an index record of amplitude and phase for each k segment, and the index is one-to-one corresponding to L k If it is found that the average of the corresponding amplitude spectrum reaches 60 but some records in the phase spectrum are greatly offset when k=1, record the L1 value of this segment for comparison with other segments, and then observe the continuous change of multiple segments in the time domain. Each comparison will link the L k Value with the amplitude and phase of the same segment, mark the time point and integrate these values into a unified sequence in order. For specific inspection inside the device, other parameters can also be stored together. Finally, after arranging all the fitting parameters of the time period, they form a complete set of time domain fitting parameters.

[0122] The steps for obtaining the feature drift curve are:

[0123] Based on the time domain fitting parameter set, extract the linear fitting parameter of each time period, and obtain the voltage-current ratio of the corresponding time period, and use the sliding window method for smoothing to form time period matching data;

[0124] According to the time period matching data, calculate the threshold drift of each time period, and the calculation formula is:

[0125]

[0126] wherein T k represents the threshold shift amount of the kth time period, P k,m represents the mth linear fitting parameter value in the kth time period, R k,m represents the mth voltage-to-current ratio value in the kth time period, and M represents the total number of data points in the kth time period.

[0127] Based on the threshold shift amount, the shift data of all time periods are arranged in chronological order, and a feature shift curve including the shift features of all time periods is constructed.

[0128] Specifically, based on the set of time-domain fitting parameters, the linear fitting parameters of each time period are extracted and the voltage-to-current ratio of the corresponding time period is obtained. First, the start and end positions of each time period are searched in the set of time-domain fitting parameters. Then, the linear fitting parameter data points and the voltage-to-current ratio records in each period are found according to these position ranges, and the same period data is smoothed in a sliding window manner. Specifically, the mean or median of adjacent data points is calculated as the smoothing result of the current window according to the determined window size. For example, if the window size is set to 5 data points, the average value of the 5 data points is calculated after reading them, and then the average value is written back to the corresponding index in the current paragraph. Then, the data point is moved forward by 1 and the operation is repeated. If a period has 50 data points, 46 smoothing results can be obtained. Once it is found that the linear fitting parameter value or the voltage-to-current ratio of a certain data point is far beyond the pre-set allowable range during the smoothing process, for example, the ratio peak value is 2.5 in multiple pulse experiments, and a safety margin of 0.3 is added to obtain 2.8 as an upper limit threshold. When the voltage-to-current ratio of a certain record reaches 3.2, the data point is additionally marked after smoothing. When the final summary is performed, the marked points can be excluded from subsequent calculations or saved separately. The specific processing procedures are determined according to the site requirements. After all the smoothing operations are completed, the records of the entire period are rearranged in chronological order, and the abnormal points that have been marked are explained. The smoothed data is summarized as time period matching data.

[0129] The formula has the benefits of considering the square proportion of the deviation between the linear fitting parameter value and the voltage-to-current ratio, and incorporating multiple-level indicators such as absolute product and fourth-order difference, so that the comprehensive measurement of difference and coupling strength is integrated in the formula.

[0130] P k,m The parameter acquisition steps are as follows:

[0131] This parameter represents the value of the mth linear fitting parameter in the kth time period, the range of which can be known from the statistics above, for example, previous test results show that the typical value of the linear fitting parameter falls within the interval of 0 to 2, and it is observed that in some extreme cases, it can reach 3 to 4. In order to accurately collect P k,m All fitting results need to be recorded in chronological order within the kth period, and each value is recorded as P k,m , m is accumulated from 1 to M, if the number of data points in the kth period is 50, then m = 1 to 50, that is, the comparable P k,m value sequence can be obtained. The example below will give specific values to illustrate how to bring them into the formula.

[0132] R k,m The acquisition steps of the parameter are as follows:

[0133] This parameter refers to the mth voltage-to-current ratio in the kth time period. During on-site measurement, the voltage and current values are often collected separately and the ratio is formed under the same timestamp. If the voltage is 48V and the current is 2A, the ratio is 24. In order to maintain consistency, this type of ratio needs to be made into a sequence (R k,1 ,R k,2 ,…,R k,M ) within the same period k. The range usually varies with the type of equipment. Some devices specify that the ratio in the working interval can be as high as 50 or 60. During collection, the actual values obtained during previous operation can be referred to and the upper and lower limits can be statistically determined, and then they can be included in this period record. Afterwards, matching R k,m can be obtained in the same equipment environment or similar working conditions.

[0134] The acquisition steps of the M parameter are as follows:

[0135] This parameter is used to count the total number of data points in the kth period, which is an integer related to the length of the time period and the sampling frequency. For example, if 10 seconds of observation is divided into 5 time periods, each period is 2 seconds, and if 25 records are obtained per second, there are about 50 records per period. M = 50. When the sampling frequency needs to be adjusted slightly or when a short fault period needs to be handled, M may also fluctuate slightly. Only the actual number of valid records remaining after the end of the segmentation needs to be counted to determine M, and then this integer is used to calculate the range of each summation and product expression in the formula.

[0136] Substituting the parameters into the formula gives 0.928. This result shows that when the linear fitting parameter and the voltage-to-current ratio in this example period are as listed above, the threshold drift is about 0.928. If the same period is detected multiple times and it is found that T3 is close to 2 or 3 or more, it can be recorded as a potential abnormal area related to large-scale drift. Subsequently, T k in different k periods can be identified and compared in the time axis and further observed in depth.

[0137] Based on the threshold drift amount, in the order of time period k from 1 to the last period, read the previously calculated T k , form a continuous drift data sequence, and splice the sequence according to the order of k and arrange them in the same chart or the same data structure. If T1 is about equal to 0.5 when k = 1, T2 is about equal to 0.7 when k = 2, and T3 is equal to the previously calculated 0.928 when k = 3 and increases gradually in subsequent periods, it indicates that the index shows a gradual upward trend along the time axis. The corresponding peak value needs to be viewed together with the previously obtained linear fitting parameters and the voltage-to-current ratio. If the drift characteristics of some periods increase excessively, they are recorded and included in the subsequent judgment of fault conditions. Finally, the drift amounts of all periods are summarized by this method to obtain a complete characteristic drift curve.

[0138] The acquisition step of the combined characteristic value is:

[0139] According to the characteristic drift curve, the combined characteristic value is calculated, and the calculation formula is:

[0140]

[0141] TC k represents the combined characteristic value of the kth time period, U k (x) represents the voltage response signal function of the kth time period, I k (x) represents the current response signal function of the kth time period, U k,j represents the jth voltage response signal value of the kth time period, I k,j represents the jth current response signal value of the kth time period, T k,j represents the jth threshold drift amount of the kth time period, J represents the total number of data points in the kth time period, and Q is the total variation range of the signal in the time period.

[0142] Specifically, the formula has the beneficial effect of integrating the signal function form and the discrete data form in the same expression. By performing integration and difference operations on the voltage response signal function U k (x) and the current response signal function I k (x), and performing multiple measurements such as multiplication and square sum on the discrete voltage and current data U k,j , I k,j , and the threshold drift amount T k,j , the overall description of the multi-layer relationship between voltage and current in the same time period is obtained in a comprehensive formula.

[0143] The acquisition step of the U k (x) parameter is:

[0144] The parameter represents the voltage response signal function of the kth time period, which is often recorded in a continuous measurement system. The voltage signal changes over time within the time period, and the voltage curve is tracked in real time by a sensor and a sampling card. The original discrete data is obtained by interpolation or curve fitting, and the function argument x can take the zero point of the time period as the starting point. When the time is pushed back, U k (x) can be output. To ensure accuracy, the periodic waveform is usually processed by mean or filtering to eliminate noise and transient fluctuations, and a reliable fitting method is used to construct U k (x) within the range from the start to the end of the time period. If the voltage range in industrial equipment may be between 0V and 24V, or even higher, the current equipment rating and the upper limit of sampling should be combined in actual application. If the data curve obtained at k=3 is relatively stable, a continuous function can be formed by interpolation between adjacent time points. For example, 100 data points are selected within a certain time period of 0.01 seconds, and a smooth voltage curve is fitted to represent the overall voltage function of the kth period.

[0145] I k The parameter acquisition steps are as follows:

[0146] The parameter refers to the current response signal function corresponding to U k (x), which represents the current value changing over time within the kth time period. Similar to the voltage acquisition process, current sampling should be performed at the same frequency during the operation of the field device, and then the obtained discrete points are denoised or curve fitted to obtain I k (x). In industrial equipment, the current range may be between 0A and 5A, or it may change in a wider range. By using a suitable power meter or sensor to finely sample the current and marking the time of each sampling point with the same time reference, I k (x) can be obtained. If the current signal has a mutation section, it can be checked one by one in the discrete data and excluded or down-weighted during fitting to ensure that the finally formed I k (x) function is representative. For example, when the instantaneous current measured at a certain time reaches 10A, and the system design allows a maximum of 5A, it needs to be judged whether it is a sensor failure or a transient jump in combination with the actual field. After establishing the corrected current curve, it can be mapped to I k (x) for subsequent calculation.

[0147] U k,j The parameter acquisition steps are as follows:

[0148] The parameter represents the jth voltage response signal value within the kth time period, which is the U k(x) Discretized results, usually directly from the voltage sampling of field measurement, in the time slice k distinguished in advance, each voltage data record is given an index j and arranged according to the sampling time sequence, for example, if 200 sampling points are collected in k = 1 segment, j is from 1 to 200, and the corresponding voltage value U k,1 , k,2 , k,200 , the numerical value may be between 0V and 24V or within the rated range of other devices, in order to better combine with integral and difference operations, some analysis methods will normalize these discrete voltage values again, and some people will directly use the original value, which depends on the accuracy requirements of field application. If other observers measure with the same time segmentation strategy and sampling frequency, the U k,j sequence under the same index j can also be obtained, the following example part will show how to bring U k,j into operation.

[0149] I k,j The acquisition steps of the parameter are:

[0150] The parameter corresponds to U k,j , which represents the current response signal value at the jth sampling time in the kth time period, and the acquisition method is the same as the current function I k (x) The same as the discrete sampling before construction, record the current value at each time in k segments, and mark the order with j index, if 250 current points are collected in this segment, j is from 1 to 250, each record is uniquely corresponding to its sampling time, and the time reference with the voltage sampling is consistent, ensuring that it can be subtracted or multiplied with U k,j , if the current has a sharp peak or zero drift phenomenon, it needs to be identified in the sampling process and reasonably modified or confirmed before forming the I k,j sequence, so as to obtain the matching I k,j sequence.

[0151] T k,j The acquisition steps of the parameter are:

[0152] The parameter represents the jth threshold drift in the kth time period, which is usually the calculation result in the formula or discrimination process, in the pulse measurement or industrial device sampling period, by recording the linear fitting parameters and ratio information, then bring it into the threshold drift formula introduced before to calculate T k , combined with the sampling index j to split the local offset value of each data point in the same time period, the sequence of T k,j is formed, for example, for a 2-second observation, if the local drift change is calculated every 0.02 seconds, 100 T k,jRecord, j from 1 to 100, each reflecting the threshold drift at that time point.

[0153] The J parameter acquisition step is:

[0154] The parameter is the total number of data points in the k time period, which is used to limit the upper and lower indices of operations such as and The specific size is directly related to the sampling frequency and the length of the time period, for example, if 1000 samples are taken per second and the kth segment is 0.5 seconds long, then J is equal to 500. The sampling rate or segment size will be set according to the required resolution and device bandwidth, and all subsequent operations on the same segment can use the same J value as the upper index. As long as the actual number of data points retained is checked before recording, J can be confirmed.

[0155] The Q parameter acquisition step is:

[0156] The parameter represents the total variation range of the signal in the time period, which is used to limit the integral between U k (x) and I k (x). From 0 to Q, it can be understood as the complete coverage of the voltage and current functions in this segment in the order of time or sample size. If the segment lasts for 0.01 seconds and the interpolation is expected to smooth x from 0 to 0.01, then Q is 0.01. If x represents the total number of sampling points, Q can also be taken as the maximum value according to the number of sampling points. The key is to ensure that

[0157] can correctly reflect the voltage and current difference in this segment.

[0158] Substituting the parameters into the formula gives 0.171, which indicates that when the voltage and current functions in the second time period are in the interval x = 0 to 0.02 seconds, and the corresponding discrete data and threshold drift are as above, the combined characteristic value is about 0.171. If the same process is applied to other time periods and it is found that TC k is greater than 1.0, it can be identified as an interval with significant voltage and current difference, and can be further used for difference comparison with standard threshold data and subsequent classification and discrimination.

[0159] The acquisition step of the damage type mapping table is:

[0160] Obtain the standard threshold data, arrange the combined characteristic values in chronological order to form the combined characteristic values and standard threshold data.

[0161] Based on the combined characteristic values and standard threshold data, calculate the difference between each time period, match the calculated difference along the time axis, and arrange all the differences in chronological order of data points to obtain the difference between the combined value and the standard threshold.

[0162] Based on the difference between the combination value and the standard threshold value, define the damage type discrimination rule, classify the damage type corresponding to all data points according to the pre-set classification threshold, and map the data of different damage types to the corresponding category to establish the damage type mapping table.

[0163] Specifically, the standard threshold data is obtained, and the combination feature values are arranged in time sequence. It is necessary to accumulate the corresponding standard range in multiple device detection or pulse test, and obtain the sequence of one or more reference values by referring to the device manual or existing measurement statistical results. After confirming that the combination feature values of each observation paragraph have been calculated, they are paired and stored according to the time sequence, and it is checked in the comparison process whether the combination feature values of some paragraphs exceed the upper limit range formed on the basis of multiple measurements. For example, if the highest combination feature value recorded in 10 tests is 0.85 and 0.15 is added as a buffer to obtain 1.0, if the actual measured combination feature value exceeds 1.0, it will be listed as an interval that needs to be checked in detail. Then arrange the combination feature values of all paragraphs and the corresponding standard threshold values in the same sequence and mark the deviation of each paragraph. The deviation is the difference between the actual combination feature value of the paragraph and the standard threshold value. If the deviation is greater than the alarm value set in the previous experience, such as 0.2, the paragraph is marked as potential abnormality. After collecting the deviation data corresponding to all paragraphs, the difference is also sequentially represented, and matched with the specific time of each paragraph. Finally, the overall difference between the combination feature value and the standard threshold value is summarized to observe its change in time sequence.

[0164] Based on the difference between the combination value and the standard threshold value, define the damage type discrimination rule, usually combined with industry standards or typical failure modes statistically obtained in multiple tests to list some classification thresholds, such as difference greater than 0.5 is classified as high-risk area, difference within 0.2 to 0.5 interval is classified as moderate-risk area, and difference below 0.2 is classified as normal range. Then judge the difference corresponding to each data point to see which interval it falls into. If the difference of a paragraph reaches the high-risk area standard, the paragraph can be directly marked as severe damage type. If it is only in the moderate-risk range, it is marked as general damage type. After completing the judgment of all paragraphs, map the results of each paragraph to the pre-defined category label, such as label A represents slight damage and label B represents severe damage. Then arrange these mapping relationships into a damage type mapping table. If the user needs to follow up the statistics or early warning of different types of damage information in the actual production line or experimental link, the corresponding records can be quickly found according to the mapping table and the detailed data of the abnormal paragraph is retrieved combined with the time axis, so as to form the judgment result of the overall operation status.

[0165] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in other forms. Any skilled person in the art can modify or change the disclosed technical content into equivalent embodiments with equivalent changes, and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application, without departing from the technical solution content of the present application, still falls within the protection scope of the present application.

Claims

1. A pressure sensitive device damage diagnostic system incorporating a pulse test, characterized by, The method comprises the following steps: The method comprises the following steps: Based on the response signal set, the voltage response signal and the current response signal are respectively wavelet decomposed, the wavelet coefficients of each signal at different scales are extracted, the wavelet decomposition coefficients are classified according to a preset frequency interval, and a high-frequency coefficient group and a low-frequency coefficient group are obtained. The response signal set is obtained by: The pressure sensitive device is connected with the pulse generator, the pulse sequence is started to be emitted, the voltage response signal and the current response signal caused by each pulse sequence are recorded, and preliminary response data is obtained. Based on the preliminary response data, the voltage response signal and the current response signal are classified and arranged to obtain comprehensive response data. Based on the comprehensive response data, all voltage response signals and current response signals are integrated to form a response signal set. The frequency phase vector is obtained by: Based on the time-frequency feature matrix, the components of the high-frequency coefficient group and the low-frequency coefficient group are extracted, the instantaneous amplitude change rate of each component is calculated, the instantaneous change feature is obtained by combining the time information, and instantaneous change feature data is obtained. ； in, Representing the The energy distribution value of the group Representing the The first in the group The absolute value of each wavelet decomposition coefficient Representing the The first in the group The deviation term of each coefficient, The number of wavelet coefficients; Based on the instantaneous change feature data, the instantaneous frequency and phase values are calculated, the time sequence information of each component is arranged, the instantaneous frequency and phase change of each component are associated, and instantaneous frequency data and phase data are obtained.

2. The pressure sensitive device damage diagnostic system in combination with a pulse test according to claim 1, characterized in that, Based on the instantaneous frequency data and phase data, the data is matched and classified according to the corresponding frequency interval, and the corresponding instantaneous frequency and phase values are combined into a frequency phase vector. The fault feature spectrum is obtained by: ​ ​ 3. The pressure sensitive device damage diagnostic system in combination with a pulse test of claim 1, wherein, ​ ​ ​ ​ 4. The pressure sensitive device damage diagnostic system in combination with a pulse test of claim 1, wherein, ​ Based on the frequency phase vector, the instantaneous frequency and phase values corresponding to each time point are extracted, sorted in time sequence order, interpolated and completed, and the frequency points with abnormal changes are smoothed to obtain the reconstructed frequency sequence; The Hilbert transform is performed on the reconstructed frequency sequence, and the transformed instantaneous amplitude spectrum value is calculated, and the calculation formula is: ； wherein, is the instantaneous amplitude spectrum value for the th frequency point, is the original component value in the frequency sequence, is the component after applying the Hilbert transform, is the component after applying the Hilbert transform, Based on the instantaneous amplitude spectrum value, a feature matrix including amplitude information is constructed, the amplitude values corresponding to each time point are classified according to frequency distribution, and the classified data is arranged in time sequence to form a fault feature spectrum.

5. The pressure sensitive device damage diagnostic system in combination with a pulse test of claim 1, wherein, The acquisition step of the time domain fitting parameter set is: Based on the fault feature spectrum, the amplitude data and phase data in each time period are extracted according to the time axis, and the segmented amplitude spectrum and phase spectrum are formed; According to the segmented amplitude spectrum and phase spectrum, the linear fitting parameters of each time period are calculated, and the calculation formula is: ； wherein, represents the linear fit parameter of the segment, represents the linear fit parameter of the segment, amplitude data point of the segment, phase data point of the segment; Based on the linear fitting parameters, the fitting parameters of all time periods are arranged in time sequence, and a complete time domain fitting parameter set is constructed, and the fitting parameters of each time period are associated according to the distribution relationship of the amplitude spectrum and the phase spectrum to generate the time domain fitting parameter set.

6. The pressure sensitive device damage diagnostic system in combination with a pulse test of claim 1, wherein, The acquisition step of the feature drift curve is: Based on the time domain fitting parameter set, the linear fitting parameters of each time period are extracted, and the voltage current ratio of the corresponding time period is obtained, and the sliding window method is used for smoothing to form time period matching data; According to the time period matching data, the threshold drift amount of each time period is calculated, and the calculation formula is: ； in, Representing the Threshold drift over a time period Representing the Within the time period One linear fitting parameter value, Representing the Within the time period The voltage-current ratio Representing the The total number of data points within the time period; Based on the threshold drift amount, the drift data of all time periods are arranged in time sequence, and a feature drift curve including the drift features of all time periods is constructed.

7. The pressure sensitive device damage diagnostic system in combination with a pulse test of claim 1, wherein, The acquisition step of the combined feature value is: According to the feature drift curve, the combined feature value is calculated, and the calculation formula is: ； in, Representing the Combined feature values ​​of time periods Representing the Voltage response signal function over a time period Representing the Current response signal function over a time period Representing the Time period Each voltage response signal value, Representing the Time period One current response signal value Representing the Time period A threshold drift amount, Representing the The total number of data points within the time period, where Q is the total range of signal variation within the time period.

8. The pressure sensitive device damage diagnostic system in combination with a pulse test of claim 1, wherein, The acquisition step of the damage type mapping table is: Obtain the standard threshold data, arrange the combined feature values in time sequence to form the combined feature values and the standard threshold data; Based on the combined feature values and the standard threshold data, the difference amount of each time period is calculated, the calculated difference amount is matched according to the time axis, and all the difference amounts are arranged in time sequence of data points to obtain the difference amount between the combined value and the standard threshold value; Based on the difference amount between the combined value and the standard threshold value, the damage type discrimination rule is defined, the damage types corresponding to all data points are classified according to the preset classification threshold, and the data of different damage types are mapped to the corresponding categories to establish the damage type mapping table.

Images