A method for feature extraction of multi-mode pulse waveforms in high-pressure water detection test

Abstract

The application belongs to the technical field of data processing, and particularly relates to a feature extraction method for a multi-mode pulse waveform in high-pressure water detection testing, which comprises the following steps: acquiring real-time pressure data and environmental pressure readings output by a high-pressure water detection testing system to construct an original pulse sequence; acquiring a water hammer oscillation suppression factor according to a local pressure variance, an overall fluctuation reference, an absolute value of a pressure difference value, and an overall step reference of the original pulse sequence, and correcting the original pulse sequence to obtain a denoised pulse waveform sequence; extracting a dynamic impact feature sequence according to an extreme value difference and a rise time of a pulse cycle interval in the denoised pulse waveform sequence; acquiring a comprehensive feature index according to a pressure dispersion degree of a pressure maintaining interval and an average value of the dynamic impact feature sequence, and determining a high-pressure water detection testing state according to the comprehensive feature index. The application improves the accuracy of high-pressure water detection testing state determination.

Description

Technical Field

[0001] This invention relates to the field of data processing technology. More specifically, this invention relates to a method for feature extraction of multi-mode pulse waveforms in high-pressure water testing. Background Technology

[0002] High-pressure water testing systems are primarily used to evaluate the pressure-bearing capacity and hydraulic sealing performance of test specimens. For example, an eight-channel high-pressure water testing system typically employs servo control for linear pressurization and supports multiple testing modes, including direct pressure, pulse, holding pressure, and static pressure. In alternating pressure cycles, the hydraulic system applies periodically varying pressure to the interior of the specimen, simulating pressure shocks or pulsations in actual use. During this process, the specimen must withstand instantaneous impact loads while maintaining a stable hydraulic seal at high pressure. To accurately assess the quality of the test specimen, the pressure data output by the high-pressure water testing system during operation needs to be processed to precisely detect the current high-pressure water testing status, thereby identifying potential anomalies such as micro-cracks or seal deformation within the test specimen.

[0003] Existing high-pressure water testing methods typically acquire real-time pressure waveforms in hydraulic lines using pressure sensors and perform basic smoothing and noise reduction on these waveforms. In the feature extraction and condition determination stages, existing methods generally extract the highest pressure value reached by the pressure waveform to assess the impact load on the test piece, and observe the pressure drop during the pressure holding phase to determine if the test piece is leaking.

[0004] However, when the high-pressure water testing system operates in pulse mode, the fluid inertia in the hydraulic lines often generates a water hammer effect during periodic and drastic pressure changes. This results in high-frequency oscillation noise superimposed on the acquired pulse waveform signal. Furthermore, the base pressure build-up rate differs from the inherent background vibration amplitude of the hydraulic system under different testing modes, making it difficult for existing methods to retain the true edge characteristics of the waveform while filtering out high-frequency noise. The destructive force of the pulse waveform on the test piece depends not only on the highest pressure reached by the waveform but also on the abruptness of the pressure application. Existing methods rely solely on the highest pressure extreme value for judgment, making it difficult to accurately extract the instantaneous tearing impact force characteristics of the fluid on the inner wall of the test piece, which can easily lead to misjudgments of the test state. Summary of the Invention

[0005] To address the technical problems of existing high-pressure water testing status detection methods, such as difficulty in accurately extracting features and the risk of misjudgment, this invention provides a feature extraction method for multi-mode pulse waveforms in high-pressure water testing. The method includes: acquiring real-time pressure data output by the high-pressure water testing system and ambient pressure at normal pressure; constructing a pulse sequence based on the difference between the real-time pressure data and the ambient pressure at normal pressure; obtaining a step reference based on the absolute value of the pressure difference between adjacent sampling points in the pulse sequence; obtaining a fluctuation reference and local pressure variance based on the variance of the pressure data within a sliding time window centered on the sampling points in the pulse sequence; and obtaining the feature extraction method based on the fluctuation reference, step reference, local pressure variance, and pressure difference. The absolute value is used to obtain the water hammer oscillation suppression factor. The pressure values ​​of the pulse sequence are corrected using the water hammer oscillation suppression factor to obtain a denoised pulse sequence. The sampling points between adjacent troughs and peaks in the denoised pulse sequence are extracted as pulse cycle intervals. The pressure difference between the peaks and troughs in the pulse cycle interval is taken as the extreme value difference, and the time interval between the trough and the peak is taken as the rise time. The impact characteristic value is obtained based on the extreme value difference and the rise time, and the impact characteristic sequence is constructed. The standard deviation of the pressure in the pressure holding interval data in the denoised pulse sequence is taken as the pressure dispersion. The characteristic index is obtained based on the average value of the impact characteristic value and the pressure dispersion in the impact characteristic sequence. The high-pressure water test status is determined based on the characteristic index.

[0006] This invention constructs an original pulse sequence by acquiring real-time pressure data output from a high-pressure water testing system and ambient pressure readings at normal pressure. It then obtains a water hammer oscillation suppression factor by combining local pressure variance with overall fluctuation and step thresholds. The original pulse sequence is then corrected to obtain a denoised pulse waveform sequence. Furthermore, feature extraction is performed using the extreme value difference and rise time of the denoised pulse waveform sequence to obtain a dynamic impact feature sequence. Finally, pressure dispersion is fused to obtain a comprehensive feature index to determine the test status, achieving accurate detection of the high-pressure water testing status. This invention utilizes the global statistical characteristics of the pulse sequence itself to dynamically evaluate the water hammer oscillation suppression factor. The invention estimates the true signal-to-noise level of the local area, suppressing high-frequency oscillation noise caused by fluid inertia while preserving the true edge features of the pulse waveform. During feature extraction, the invention fully considers the impact of the abruptness of pressure application on the destructive force of the specimen, combining extreme value difference and rise time to extract dynamic impact features that can measure the true destructive load. Furthermore, the invention integrates impact capability features with pressure holding stability features, downwardly correcting for pressure leakage or fluctuations during the pressure holding stage under the same driving force. This allows the comprehensive feature index to characterize excellent waveform fidelity and specimen sealing characteristics, thereby improving the accuracy of high-pressure water testing status determination.

[0007] Preferably, the water hammer oscillation suppression factor is obtained by: recording the ratio of the local pressure variance to the fluctuation benchmark as the first ratio; recording the ratio between the absolute value of the pressure difference and the step benchmark as the second ratio; inputting the difference between the first ratio and the second ratio into an exponential function with the natural constant as the base; and obtaining the water hammer oscillation suppression factor based on the output value of the function.

[0008] This invention obtains the water hammer oscillation suppression factor by using the ratio of local pressure variance to the fluctuation benchmark and the ratio of the absolute value of the pressure difference to the step benchmark. This achieves adaptive smoothing attenuation of abnormal oscillation points and preserves the true edge features of the pulse waveform. The invention uses the relative magnitude of the local pressure variance and the fluctuation benchmark to reflect the relative fluctuation degree of the local area. When the disordered oscillation near the sampling point is higher than the average background noise level under the test mode, the smoothing attenuation effect on the abnormal oscillation point is enhanced; when the local pressure fluctuation is within the normal range, the suppression effect is weakened. Simultaneously, the invention uses the relative magnitude of the absolute value of the pressure difference and the step benchmark to reflect the abrupt intensity of local pressure changes. When there is a mechanical pressurization or depressurization transient process and its intensity is higher than the average change level, the suppression and cancellation effect on the original value is weakened, thereby preserving the true edge features of the pulse waveform. This invention comprehensively evaluates the disordered fluctuation attributes and normal step attributes of local pressure. While effectively suppressing the high-frequency noise of water hammer caused by fluid inertia, it prevents the true pressure step signal from being over-smoothed, improving the waveform fidelity of the denoised pulse sequence, thus benefiting the accuracy of subsequent dynamic impact feature extraction.

[0009] Preferably, the impact characteristic value is obtained by: obtaining the product of the climb time and the preset slewing rate constant, using the ratio of the extreme value difference to the product as the gradient characteristic, inputting the gradient characteristic into an exponential function with the natural constant as the base, and obtaining the impact characteristic value based on the output value of the function and the extreme value difference.

[0010] This invention corrects the extreme value difference by using the ratio of extreme value difference to ramp-up time and a preset pressure slewing rate constant to obtain impact characteristic values, thus achieving a realistic feature extraction effect on the instantaneous tearing impact force of fluid. In feature extraction, this invention fully considers the reflection of the pressure ramp-up gradient on the intensity of transient impact. The smoother the pressurization process and the longer the ramp-up time, the more obvious the downward attenuation correction effect on the extreme value difference, thereby reflecting the weaker impact characteristics under flexible pressurization conditions. The shorter the pressure build-up time, the closer the corrected feature value is to the original extreme value difference, thus realistically extracting extremely strong fluid impact damage characteristics and improving the accuracy of measuring destructive loads.

[0011] Preferably, the characteristic index is obtained as follows: the ratio between the pressure dispersion and the average value of the impact characteristic values ​​in the impact characteristic sequence is recorded as the relative pressure holding volatility; the negative of the product between the preset pressure holding sensitivity coefficient and the relative pressure holding volatility is input into an exponential function with the natural constant as the base to obtain the pressure holding correction term; the product between the average value of the impact characteristic values ​​in the impact characteristic sequence and the pressure holding correction term is recorded as the characteristic index.

[0012] This invention uses the ratio of the average value of impact characteristic values ​​to the pressure dispersion in an impact characteristic sequence, combined with a preset pressure holding sensitivity coefficient, to correct the average value of impact characteristic values ​​and obtain a characteristic index. This achieves a fusion effect of pressure holding stability and impact strength characteristics. The invention utilizes the relative relationship between the average value of characteristic values ​​and the pressure dispersion to characterize the relative pressure holding volatility. The more severe the pressure leakage or fluctuation during the pressure holding stage, the stronger the downward correction effect on the characteristic index. When the hydraulic system can maintain internal pressure stability after impact, the characteristic index is closer to the true average value, thus accurately characterizing excellent waveform fidelity and specimen sealing characteristics, improving the final evaluation quality of feature extraction.

[0013] Preferably, the method for obtaining the ambient pressure under normal pressure includes: controlling the plunger pump and centrifugal pump in the high-pressure water testing system to be in a stopped and depressurized state, and reading the ambient pressure reading of the multi-channel pressure transmitter under normal pressure as the ambient pressure under normal pressure.

[0014] Preferably, obtaining the step reference based on the absolute value of the pressure difference between adjacent sampling points in the pulse sequence includes: extracting the absolute value of the pressure difference between all adjacent sampling points in the pulse sequence, and using the average value of all absolute pressure differences as the step reference.

[0015] Preferably, obtaining the fluctuation benchmark and local pressure variance based on the variance of pressure data within a sliding time window centered on the sampling point in the pulse sequence includes: calculating the average value of the variance of pressure values ​​within all sliding time windows, and using this average value as the fluctuation benchmark.

[0016] Preferably, the step of correcting the pressure values ​​of the pulse sequence using a water hammer oscillation suppression factor to obtain a denoised pulse sequence includes: multiplying the water hammer oscillation suppression factor of the sampling point with the pressure of the sampling point, and recombining all the multiplied sampling points in the original time order to obtain a denoised pulse sequence.

[0017] This invention achieves smoothing and denoising of the original pulse sequence by directly multiplying the pressure values ​​at the sampling points using a water hammer oscillation suppression factor, and then recombining all the multiplied sampling points in their original time order. By utilizing the regulating effect of the water hammer oscillation suppression factor, this invention applies strong attenuation to abnormal oscillation points while weakening the suppression of normal fluctuations or sudden pressure increases and decreases. While filtering out high-frequency oscillation noise caused by the water hammer effect, it preserves the original temporal characteristics and true edge information of the pulse waveform to the greatest extent, providing high-quality data input for feature extraction of multi-mode pulse waveforms.

[0018] Preferably, before extracting the sampling points between adjacent troughs and peaks in the denoised pulse sequence as pulse loop intervals, the method further includes: traversing the denoised pulse sequence, extracting the inflection point sampling point where the pressure value change trend changes from decreasing to increasing as candidate troughs, and extracting the inflection point sampling point where the pressure value change trend changes from increasing to decreasing as candidate peaks; calculating the absolute pressure difference between adjacent candidate troughs and peaks on the time axis, eliminating candidate troughs and peaks with an absolute pressure difference less than or equal to the step reference, and retaining candidate troughs and peaks with an absolute pressure difference greater than the step reference, and marking them as troughs and peaks respectively.

[0019] Preferably, the determination of the high-pressure water test status based on the characteristic index includes: determining the current high-pressure water test status as compliant when the characteristic index is greater than or equal to a preset characteristic threshold; and determining the current high-pressure water test status as abnormal leakage when the characteristic index is less than the preset characteristic threshold.

[0020] The beneficial effects of this invention are as follows: This invention combines the overall fluctuation benchmark and local pressure variance of the pulse sequence, as well as the absolute value of the difference between the overall step benchmark and the local pressure, to obtain a water hammer oscillation suppression factor. This corrects the pressure values ​​of the original pulse sequence, suppressing high-frequency oscillation noise generated by fluid inertia while preserving the true edge features of the pulse waveform. This avoids excessive smoothing of the true pressure step signal, restores the transient physical conditions of the hydraulic system, and provides accurate waveform data for subsequent feature extraction. This invention also combines the extreme value difference of the pulse cycle interval and the climb time to obtain dynamic impact characteristic values, considering the influence of pressure build-up speed on the intensity of transient impact, and extracting the instantaneous tearing impact force characteristics of the fluid on the specimen. Furthermore, this invention combines the pressure holding zone... The invention obtains a comprehensive characteristic index by combining the pressure dispersion and the average value of the dynamic impact characteristic sequence. This combines transient impact characteristics with high-pressure seal maintenance characteristics, restoring the true test state of the specimen in multi-mode test cycles, where it must withstand impact and maintain a high-pressure seal. By fusing impact characteristics and pressure holding characteristics to obtain a comprehensive characteristic index, the invention reduces the interference caused by differences in foundation pressure build-up rates under different test modes and inherent background vibrations of the equipment. This allows the comprehensive characteristic index to reflect the small pressure drop caused by micro-cracks inside the specimen or micro-deformation of the seal. The invention improves the accuracy of abnormal leakage state identification, reduces the risk of misjudgment due to high-frequency oscillation noise or water hammer effect, and facilitates the final accurate determination of the high-pressure water test state. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating a method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to the present invention; Figure 2 This is a schematic diagram illustrating the comparison of denoised pulse sequences in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0024] This invention discloses a method for feature extraction of multi-mode pulse waveforms in high-pressure water testing, referring to... Figure 1 This includes steps S1 to S4: S1. Based on the real-time pressure data output by the high-pressure water testing system within the preset testing cycle, and combined with the zero-point pressure data of the multi-channel pressure transmitter, obtain a pulse sequence containing multi-mode pulse waveforms.

[0025] Specifically, during the system power-on initialization phase, the plunger pump and centrifugal pump in the high-pressure water testing system are controlled to be in a stopped and depressurized state. The ambient pressure reading of the multi-channel pressure transmitter at normal pressure is read and marked as the zero-point pressure data. Subsequently, the cyclic test program is started to acquire the real-time pressure data output by the high-pressure water testing system within the preset test cycle. The difference between the real-time pressure data and the zero-point pressure data is calculated, and the difference is arranged in chronological order to construct a pulse sequence containing multi-mode pulse waveforms.

[0026] S2. Based on the fluctuation reference and step reference of the pulse sequence containing multi-mode pulse waveforms, and combined with the absolute value of the local pressure variance and pressure difference of the sampling points in the pulse sequence containing multi-mode pulse waveforms, obtain the denoised pulse sequence.

[0027] It should be noted that fluid inertia exists in the hydraulic lines of the high-pressure water testing system, which often generates a water hammer effect when the pressure changes drastically and periodically. This results in high-frequency oscillation noise being superimposed on the multi-mode pulse waveform signals acquired by the sensor. The base pressure build-up rate and the inherent background vibration amplitude of the hydraulic system differ to some extent under different water testing modes. To accurately extract features from the multi-mode pulse waveform, this invention utilizes the global statistical characteristics of the waveform itself to dynamically evaluate the local true signal-to-noise level.

[0028] Specifically, the absolute values ​​of pressure differences between all adjacent sampling points in a pulse sequence containing multi-mode pulse waveforms are extracted, and the average of all absolute pressure differences is used as the step reference. A sliding time window is constructed with each sampling point in the pulse sequence containing multi-mode pulse waveforms as the center and a preset time span as the size. The average value of the pressure value variance within all sliding time windows is calculated, and this average value is used as the fluctuation reference. The sampling points in the pulse sequence containing multi-mode pulse waveforms are traversed, and the variance of the pressure data within the corresponding sliding time window of each sampling point is used as the local pressure variance. The absolute value of the pressure difference between the pressure value of a sampling point and the pressure value of its adjacent previous sampling point is obtained. Based on the fluctuation reference, step reference, local pressure variance, and absolute value of pressure difference, the water hammer oscillation suppression factor of the sampling point is obtained.

[0029] For example, the preset time span is 50ms.

[0030] Specifically, the water hammer oscillation suppression factor satisfies the following relationship: ; In the formula, The first pulse in the pulse sequence Water hammer oscillation suppression factor at each sampling point For the first Each sampling point corresponds to the local pressure variance within the sliding time window. As a benchmark for fluctuations, For the first The absolute value of the pressure difference between the pressure value at each sampling point and the pressure value at its adjacent previous sampling point. As a step reference, It is an exponential function with the natural constant as the base.

[0031] in, This represents the relative fluctuation of the local area's fluctuation variance compared to the global average fluctuation variance. A larger value indicates that the disordered oscillations near the current sampling point are significantly higher than the average background noise level under this test mode, making the water hammer oscillation suppression factor closer to 0, thus exerting a stronger smoothing and attenuation effect on the abnormal oscillation point. A smaller value indicates that the local pressure fluctuations are within the global normal range or relatively stable, thus weakening the suppression effect on the original value. This represents the relative intensity of the abrupt change in local pressure variation compared to the global average step change. A larger value indicates that the current situation is more likely to be a normal mechanical pressurization or depressurization transient process, and that its intensity is higher than the global average change level. and Subtraction is used to establish a dynamic adjustment mechanism between filtering out disordered noise and preserving true edges. In actual hydraulic conditions, the true pressure step boundary will inevitably cause a significant increase in local variance during calculation, which is easily misjudged as water hammer oscillation noise and over-smoothed. Through the subtraction operation, the relative abrupt change intensity of the signal edge can actively offset the relative fluctuation caused by the surge in variance. When encountering the true pressure boundary, a larger abrupt change intensity will dominate the difference to go negative, leading to... Rapid decay approaches 0, causing the water hammer oscillation suppression factor to approach 1, thus preserving the true edge characteristics of the multi-mode pulse waveform; if the local pressure change is slower or only a small fluctuation, the intensity of the abrupt change characterizing the true edge decreases, leading to its... If the suppression and cancellation effect weakens, the corresponding noise smoothing action will be performed according to the magnitude of the fluctuation variance.

[0032] Furthermore, the pressure values ​​at the sampling points are multiplied by the water hammer oscillation suppression factor at the sampling points, and all the multiplied sampling points are recombined in their original time order to obtain a denoised pulse sequence.

[0033] For example, Figure 2This is a comparative diagram of the denoised pulse sequences in this invention. As can be seen from the diagram, the data change trajectories of the two pulse sequences during the transient pressure build-up and depressurization phases highly overlap. However, they exhibit significant differences in data patterns within the high-pressure maintenance range. The denoised pulse sequence with stable pressure maintenance shows relatively stable pressure fluctuations around the high-pressure baseline within the maintenance range, with a relatively concentrated distribution of data points. In contrast, the denoised pulse sequence with abnormal pressure drop shows a clear and continuous downward gradient in pressure values ​​within the same maintenance range. This provides a data foundation for subsequently calculating the characteristic index of the multi-mode waveform by combining pressure dispersion, thereby accurately identifying anomalies and triggering audible and visual alarms.

[0034] S3. Based on the extreme value difference of each pulse cycle interval in the denoised pulse sequence, and combined with the rise time of each pulse cycle interval, extract the impact feature sequence of the multi-mode pulse waveform.

[0035] It should be noted that the destructive force of multi-mode pulse waveforms on the specimen depends not only on the highest pressure reached by the waveform but also on the abruptness of pressure application. Under the same extreme value difference, the shorter the pressure build-up time, the stronger the instantaneous tearing impact force of the fluid on the inner wall of the specimen. Therefore, this invention combines the extreme value difference and rise time of multi-mode pulse waveforms to extract dynamic impact characteristics that can measure the actual destructive load.

[0036] Specifically, the denoised pulse sequence is traversed, and inflection points where the pressure value change trend changes from decreasing to increasing are extracted as candidate troughs, and inflection points where the pressure value change trend changes from increasing to decreasing are extracted as candidate peaks. The absolute pressure difference between adjacent candidate troughs and peaks on the time axis is calculated. Candidate troughs and peaks with absolute pressure differences less than or equal to the step reference are removed, while those with absolute pressure differences greater than the step reference are retained and marked as troughs and peaks, respectively. All consecutive sampling points between two adjacent troughs in the denoised pulse sequence are truncated into a pulse cycle interval.

[0037] Furthermore, the peak and trough points of each pulse cycle interval in the denoised pulse sequence are extracted. The difference between the pressure values ​​at the peak and trough points is calculated, and this difference is taken as the extreme value difference of that pulse cycle interval. The time interval from the trough point to the peak point within that pulse cycle interval is calculated, and this time interval is taken as the rise time. Based on the extreme value difference and rise time of each pulse cycle interval, the impact characteristic value of that pulse cycle interval is calculated. The impact characteristic values ​​of all pulse cycle intervals are combined in chronological order to extract the impact characteristic sequence of the multi-mode pulse waveform.

[0038] Specifically, the impact eigenvalues ​​satisfy the following relationship: ; In the formula, For the first Impact characteristic values ​​of each pulse cycle interval, For the first The extreme value difference of each pulse cycle interval For the first The rise time of each pulse cycle interval It is an exponential function with the natural constant as its base. The preset slew rate constant, The empirical value range is [20, 100]. In this embodiment... The value is 50 kPa / ms; the implementer can determine this based on the actual situation. When the rated output flow rate of a centrifugal pump or plunger pump is large, the flow rate can be appropriately increased. To accommodate steeper baseline pressure ramp rates; when the test specimen is a large-volume cavity, the pressure can be appropriately reduced. To match the slower pressure build-up process caused by the volume effect.

[0039] in, This represents the pressure gradient within the pulse cycle; a larger value indicates a more intense transient impact, leading to... The closer it gets to 0, the more the impact eigenvalue... The closer to the original extreme difference This allows for the accurate extraction of extremely strong fluid impact damage characteristics; a smaller value indicates a smoother pressurization process, leading to... The larger the difference between extreme values, the better. The more pronounced the downward attenuation correction effect, the more significant the impact eigenvalue. The difference is less than the extreme value difference, thus reflecting the weaker impact characteristics under flexible pressurization conditions.

[0040] S4. Based on the average value of the impact characteristic value in the impact characteristic sequence, combined with the pressure dispersion of the denoised pulse sequence in the pressure holding range, extract the characteristic index of the multi-mode waveform, and determine the water test status based on the characteristic index of the multi-mode waveform.

[0041] It should be noted that in the multi-mode test cycle, the specimen needs to withstand instantaneous impact loads while maintaining a stable hydraulic seal during the high-pressure phase. If the specimen has micro-cracks or the seal undergoes irreversible deformation, it can lead to microscopic pressure drops or irregular fluctuations during the pressure holding phase. Therefore, this invention extracts and fuses the impact capability and pressure holding stability characteristics to obtain the final multi-mode waveform characteristic evaluation results.

[0042] Specifically, the average value of all impact characteristic values ​​in the impact characteristic sequence of the multi-mode pulse waveform is calculated, and this average value is used as the dynamic impact mean. For each pulse cycle interval, a high-pressure maintenance judgment threshold is obtained. The denoised pulse sequence within the pulse cycle interval is traversed, and the first sampling point with a pressure value greater than or equal to the high-pressure maintenance judgment threshold is extracted as the pressure holding start point, and the last sampling point with a pressure value greater than or equal to the high-pressure maintenance judgment threshold is extracted as the pressure holding end point. The continuous sampling points between the pressure holding start point and the pressure holding end point are extracted as the pressure holding interval data within the preset high-pressure maintenance time period.

[0043] For example, the method for obtaining the high pressure maintenance determination threshold includes: for each pulse cycle interval, calculating the average pressure value of all sampling points in the pulse cycle interval, and using the average value as the first reference pressure; extracting all sampling points in the pulse cycle interval whose pressure value is greater than the first reference pressure, calculating the average pressure value of the extracted sampling points, and using the average value as the high pressure maintenance determination threshold.

[0044] Furthermore, the average value of all impact characteristic values ​​in the impact characteristic sequence of the multi-mode pulse waveform is calculated. All sampling points within the preset high-pressure maintenance time period in the denoised pulse sequence are extracted as pressure holding interval data, and the standard deviation of all pressure values ​​in the pressure holding interval data is used as the pressure dispersion. Based on the average value of the impact characteristic values ​​and the pressure dispersion, the characteristic index of the multi-mode waveform is calculated.

[0045] Specifically, the characteristic index satisfies the following relation: ; In the formula, For the characteristic index of multi-mode waveforms, This represents the average value of the impact characteristic values ​​in the impact characteristic sequence. For pressure dispersion, It is an exponential function with the natural constant as its base. The preset pressure holding sensitivity coefficient, The empirical value range is [1, 5]. In this embodiment... The value is 2, and the implementers can determine the number based on the actual situation. When the leakage tolerance of the test piece is extremely low, the value can be appropriately increased. To amplify the exponential drop caused by minor leaks; when the accumulator in the test system itself exhibits normal fluctuations, the magnitude can be appropriately reduced. To prevent misjudgments caused by system jitter.

[0046] in, This represents the relative pressure holding volatility characteristic based on the average impact intensity. A larger value indicates more severe pressure leakage or fluctuation during the pressure holding phase under the same driving force, causing the exponential function term to approach 0 overall, thus increasing the characteristic exponent of the multi-mode waveform. The smaller this value, the better the hydraulic system can maintain internal pressure stability after high-intensity impacts, causing the exponential function term to approach 1 overall, thus making the characteristic exponent of the multi-mode waveform more pronounced. The closer to the true average This demonstrates excellent waveform fidelity and specimen sealing characteristics.

[0047] Furthermore, if the characteristic index of the multi-mode waveform is greater than or equal to the preset characteristic threshold, the current high-pressure water test status is determined to be compliant; if the characteristic index of the multi-mode waveform is less than the preset characteristic threshold, the current high-pressure water test status is determined to be abnormal leakage, for example, the preset characteristic threshold is 1200 kPa.

[0048] In one embodiment, if the characteristic index of the multi-mode waveform is less than a preset characteristic threshold for three consecutive test cycles, an audible and visual alarm is triggered and the system is shut down and pressure is released; if the characteristic index of the multi-mode waveform is greater than or equal to the preset characteristic threshold, the current test cycle is marked as qualified and the cycle counter is automatically incremented to enter the next test cycle.

Claims

1. A method for feature extraction of multi-mode pulse waveforms in high-pressure water testing, characterized in that, include: Acquire real-time pressure data output by the high-pressure water testing system, as well as ambient pressure at normal pressure; A pulse sequence is constructed based on the difference between real-time pressure data and ambient pressure at normal pressure. The step reference is obtained by taking the absolute value of the pressure difference between adjacent sampling points in the pulse sequence. The fluctuation reference and local pressure variance are obtained by taking the variance of the pressure data within the sliding time window centered on the sampling points in the pulse sequence. The water hammer oscillation suppression factor is obtained by taking the fluctuation reference, step reference, local pressure variance and absolute value of the pressure difference. The pressure values ​​of the pulse sequence are corrected by taking the water hammer oscillation suppression factor to obtain the denoised pulse sequence. The sampling points between adjacent troughs and peaks in the denoised pulse sequence are extracted as the pulse cycle interval. The difference between the pressure at the peak and trough of the pulse cycle interval is taken as the extreme value difference, and the time interval between the trough and the peak is taken as the rise time. The impact characteristic value is obtained based on the extreme value difference and the rise time, and the impact characteristic sequence is constructed. The standard deviation of pressure in the pressure holding interval data of the denoised pulse sequence is used as the pressure dispersion; the characteristic index is obtained based on the average value of the impact characteristic value and the pressure dispersion in the impact characteristic sequence. The condition of the high-pressure water test is determined based on the characteristic index.

2. The method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, The method for obtaining the water hammer oscillation suppression factor is as follows: The ratio of the local pressure variance to the fluctuation benchmark is denoted as the first ratio; the ratio of the absolute value of the pressure difference to the step benchmark is denoted as the second ratio. The difference between the first ratio and the second ratio is input into an exponential function with the natural constant as the base, and the water hammer oscillation suppression factor is obtained based on the output value of the function.

3. The method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, The method for obtaining the impact characteristic value is as follows: The product of the climb time and the preset slewing rate constant is obtained. The ratio of the extreme value difference to the product is used as the gradient feature. The gradient feature is input into an exponential function with the natural constant as the base. The impact feature value is obtained based on the output value of the function and the extreme value difference.

4. The feature extraction method for multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, The characteristic index is obtained as follows: The ratio between the pressure dispersion and the average value of the impact characteristic values ​​in the impact characteristic sequence is denoted as the relative pressure holding volatility. The negative of the product between the preset pressure holding sensitivity coefficient and the relative pressure holding volatility is input into an exponential function with the natural constant as the base to obtain the pressure holding correction term. The product between the average value of the impact characteristic values ​​in the impact characteristic sequence and the pressure holding correction term is denoted as the characteristic index.

5. The method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, The method for obtaining the ambient pressure under normal pressure includes: controlling the plunger pump and centrifugal pump in the high-pressure water testing system to be in a stopped and depressurized state, and reading the ambient pressure reading of the multi-channel pressure transmitter under normal pressure as the ambient pressure under normal pressure.

6. The method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, The step reference obtained based on the absolute value of the pressure difference between adjacent sampling points in the pulse sequence includes: extracting the absolute value of the pressure difference between all adjacent sampling points in the pulse sequence, and using the average value of all the absolute values ​​of the pressure difference as the step reference.

7. The method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, The step of obtaining the fluctuation benchmark and local pressure variance based on the variance of pressure data within a sliding time window centered on the sampling point in the pulse sequence includes: calculating the average value of the variance of pressure values ​​within all sliding time windows, and using this average value as the fluctuation benchmark.

8. The method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, The method of correcting the pressure values ​​of the pulse sequence using a water hammer oscillation suppression factor to obtain a denoised pulse sequence includes: multiplying the water hammer oscillation suppression factor of the sampling point with the pressure of the sampling point, and then recombining all the multiplied sampling points in the original time order to obtain the denoised pulse sequence.

9. The method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, Before extracting the sampling points between adjacent troughs and peaks in the denoised pulse sequence as the pulse cycle interval, the method further includes: traversing the denoised pulse sequence, extracting the inflection point sampling point where the pressure value change trend changes from decreasing to increasing as candidate troughs, and extracting the inflection point sampling point where the pressure value change trend changes from increasing to decreasing as candidate peaks; calculating the absolute pressure difference between adjacent candidate troughs and peaks on the time axis, eliminating candidate troughs and peaks with an absolute pressure difference less than or equal to the step reference, and retaining candidate troughs and peaks with an absolute pressure difference greater than the step reference, and marking them as troughs and peaks respectively.

10. The method for feature extraction of multi-mode pulse waveforms in high-pressure water testing according to claim 1, characterized in that, The determination of the high-pressure water test status based on the characteristic index includes: in response to the characteristic index being greater than or equal to a preset characteristic threshold, the current high-pressure water test status is determined to be a qualified status; in response to the characteristic index being less than the preset characteristic threshold, the current high-pressure water test status is determined to be an abnormal leakage status.

Images