Hydraulic pipe fitting test error compensation method based on intelligent sensor

Abstract

The application relates to the technical field of hydraulic pipe testing, and discloses a hydraulic pipe testing error compensation method based on an intelligent sensor, which comprises the following steps: synchronously collecting and verifying the effectiveness of channels through a multi-source sensor array, combining spectrum aliasing risk prediction and path shunting, guaranteeing the data quality and working condition adaptability of the core decoupling link, realizing the accurate separation of fluid pulsation and mechanical vibration under the condition of spectrum overlap through progressive fusion of pipeline axial coherence discrimination and acoustic-mechanical dual-mode projection, realizing the accurate correction of phase delay under the condition of time-varying pressure wave velocity through dynamic interpolation of the three-dimensional calibration relationship of temperature, gas content and wave velocity and adaptive window coherence tracking, improving the testing stability and reliability in complex environment fluctuations, ensuring that the compensation result continuously meets the accuracy threshold requirement through residual error uncertainty closed-loop verification, reducing the production line interruption caused by frequent calibration of the reference pipeline, and improving the batch testing efficiency and qualification rate of the hydraulic pipe.

Description

Technical Field

[0001] This invention relates to the field of hydraulic pipe fitting testing technology, specifically to a hydraulic pipe fitting testing error compensation method based on intelligent sensors. Background Technology

[0002] During the testing of the pressure-bearing and sealing performance of hydraulic pipe fittings, it is usually necessary to accurately measure and analyze the internal pressure signal of the pipeline. However, in the actual testing environment, fluid pulsation and pipeline structure vibration coexist in the hydraulic system, and the two often overlap in the frequency domain, resulting in significant mechanical vibration noise mixed into the acquired pressure signal, thus affecting the accuracy of the test results.

[0003] Existing methods for compensating hydraulic pipe fitting testing errors based on smart sensors cannot separate hydraulic pulsations from mechanical vibrations when the spectra overlap. They fail to filter out noise while attenuating the true pressure characteristics, calculate phase delays based on the assumption of a fixed wave velocity, and do not consider the time-varying wave velocity caused by temperature and gas content changes, resulting in coherence mismatch. They also lack axial multi-point coherence discrimination and dual-mode verification mechanisms, making it difficult to distinguish between axial fluid pulsations and local structural vibrations. Furthermore, the aging drift of the reference pipeline after long-term calibration is not identified, leading to implicit fractures in the compensation reference. Therefore, their practicality has certain limitations. Summary of the Invention

[0004] This invention provides a hydraulic pipe fitting testing error compensation method based on intelligent sensors to solve the problems in the prior art where, during the testing of hydraulic pipe fittings, the fluid pulsation signal and the mechanical vibration signal overlap in the spectrum, making it difficult for traditional filtering methods to effectively separate them, thus causing test result distortion. Furthermore, the pressure wave propagation speed changes with temperature and gas content fluctuations, further introducing phase errors and reducing test accuracy.

[0005] This invention provides the following technical solution: a method for compensating for testing errors in hydraulic pipe fittings based on intelligent sensors, comprising:

[0006] Multiple signals of pressure, acceleration, temperature and gas content are collected, and clock synchronization and numerical validity are jointly verified. The set of valid measurement points is selected based on comprehensive validity criteria.

[0007] Frequency domain analysis is performed on the effective pressure signal and acceleration signal to extract their respective main frequency bands and calculate the overlap bandwidth ratio. The results are compared with a preset threshold to determine the risk of spectral aliasing and to determine the processing path.

[0008] When spectral aliasing is determined, the coherence function and cross-spectral phase between measurement points are calculated. Within the overlapping frequency band, the candidate fluid pulsation component and candidate vibration noise component are determined by the joint constraint of coherence peak value and phase deviation.

[0009] The candidate fluid pulsation components are projected onto a preset acoustic modal basis, the modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline. The fluid pulsation is determined based on the joint criterion of energy concentration and physical residual.

[0010] Candidate vibration noise components and components with unclear attribution are projected onto a preset mechanical mode basis. The energy concentration of the mechanical mode projection is calculated, and the physical residual is calculated in combination with the structural dynamics equation. For components that cannot be determined by a single mode, arbitration by minimizing the acoustic and mechanical mode residuals is performed to determine the final attribution.

[0011] For components identified as fluid pulsations, the stability of pressure wave propagation speed is determined based on the rate of temperature change and the fluctuation of gas content.

[0012] When the propagation speed of pressure waves is unstable, the wave speed is dynamically interpolated and corrected based on temperature and gas content, the expected value of phase delay is updated, and the coherence function and cross-spectral phase are recalculated within an adaptive window to achieve coherent tracking correction under time-varying wave speed conditions.

[0013] The separated fluid pulsation signal is reconstructed by inverse transformation, the residual error uncertainty is calculated and compared with the accuracy threshold. If the requirements are met, the compensation result is output; otherwise, the working condition review or recalibration process is triggered.

[0014] As an optional solution to the hydraulic pipe fitting testing error compensation method based on intelligent sensors described in this invention, wherein: when spectral aliasing is determined to exist, the coherence function and cross-spectral phase between measurement points are calculated, specifically:

[0015] After clock synchronization and numerical validity verification, the installation position coordinates of all valid pressure measuring points along the pipe fitting axis are obtained based on the comprehensive validity flag and the set of valid measuring points is selected. The geometric distance between each pair of measuring points is calculated, the axial geometric coordinates and distance of the measuring points are obtained, and the physical scale benchmark required for spatial propagation analysis is established to support the subsequent accurate calculation of phase delay.

[0016] Based on the fluid medium physical properties and pipeline structural parameters, the theoretical pressure wave velocity under the current working condition is calculated, and the expected propagation delay and phase difference between each measuring point are derived accordingly. Based on the fluid-pipeline coupled wave velocity model, the theoretical propagation delay and expected phase difference are derived to provide a physical reference benchmark for coherent matching.

[0017] The effective pressure measurement point signals are segmented, windowed, and averaged in the frequency domain to estimate the cross-spectral density between each measurement point pair, suppress random noise, and extract stable frequency domain coupling features.

[0018] The coherence function is calculated from the cross-spectral density and the self-spectral density, and the cross-spectral phase is calculated from the cross-spectral density argument, thus quantifying the degree of linear correlation and phase propagation relationship between measurement points.

[0019] Frequency domain analysis is performed on the effective pressure signal and acceleration signal to extract their respective main frequency bands and search for the peak value of the coherent function within the overlapping frequency range determined by the overlap bandwidth ratio. The degree of agreement between the measured phase difference and the theoretical expected phase difference is compared to determine whether the signal belongs to axial coherent fluid pulsation or local incoherent vibration.

[0020] If the coherence consistency flag of the measuring point pair is 1, it is determined that the signal of the measuring point pair has axial propagation coherence, which is extracted as a candidate fluid pulsation signal and the candidate fluid pulsation component is projected onto the preset acoustic mode base. The modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline.

[0021] If the coherence consistency flag of the measuring point pair is 0, the signal of the measuring point pair is determined to be local incoherent vibration. It is extracted as a candidate vibration noise signal and the candidate vibration noise component and the component with unclear affiliation are projected onto the preset mechanical mode basis. The energy concentration of the mechanical mode projection is calculated and the physical residual is calculated in combination with the structural dynamics equation.

[0022] As an optional scheme of the hydraulic pipe fitting testing error compensation method based on intelligent sensors described in this invention, the following steps are taken: projecting the candidate fluid pulsation component onto a preset acoustic modal basis, calculating the modal projection energy concentration, and calculating the physical residual in conjunction with the one-dimensional wave equation of the pipeline, specifically:

[0023] When spectral aliasing is determined, all valid measurement point signals with a coherence consistency flag of 1 in the calculation of the coherence function and cross-spectral phase between measurement points are obtained. Frequency domain averaging is performed in the coherence band to construct the frequency domain representation of the candidate fluid pulsation signal. The candidate signals extracted from the calculation of the coherence function and cross-spectral phase between measurement points when spectral aliasing is determined are reconstructed and represented in the frequency domain to establish the frequency domain input required for subsequent modal projection.

[0024] Obtain the pre-calibrated acoustic modal basis and verify its normalized state to ensure that the mathematical basis of modal projection is strictly valid;

[0025] Calculate the projection coefficients of candidate fluid pulsation signals onto the acoustic mode basis, and quantize the coupling strength between the signal and the fluid pulsation mode;

[0026] Calculate the proportion of the energy projected onto the acoustic mode basis to the total energy of the candidate signal, evaluate the proportion of the projected energy to the total energy of the candidate signal, and determine whether the signal energy is concentrated in the acoustic mode subspace;

[0027] Substitute the signal reconstructed by modal projection into the one-dimensional wave equation of the pipeline, calculate the physical residual, and verify whether the signal conforms to the fluid propagation law.

[0028] Based on the joint criterion of projected energy concentration and physical residual, the modal assignment of candidate components is determined. By combining the dual criteria of energy concentration and physical residual, the candidate components are accurately assigned to real fluid pulsation or suspected vibration noise.

[0029] If the acoustic mode attribution flag is 1, the candidate component is determined to be a real fluid pulsation component and is retained. For components determined to be fluid pulsations, the stability of pressure wave propagation speed is determined based on the temperature change rate and gas content fluctuation.

[0030] If the acoustic mode attribution determination flag is 0, the attribution of the candidate component is questionable. In this case, the component is preferentially transferred to the preset mechanical mode basis for projecting the candidate vibration noise component and the component with unclear attribution. The energy concentration of the mechanical mode projection is calculated and the physical residual is calculated in combination with the structural dynamics equation as a vibration noise candidate to participate in the mechanical mode projection verification and dual-mode residual arbitration. If the attribution still cannot be determined after arbitration, the process is back to when spectral aliasing is determined, and the coherence function between measurement points and the cross-spectral phase adjustment coherence boundary threshold are calculated for re-screening.

[0031] As an optional scheme of the hydraulic pipe fitting test error compensation method based on intelligent sensors described in this invention, the following steps are taken: projecting candidate vibration noise components and components with unclear attribution onto a preset mechanical modal basis, calculating the energy concentration of the mechanical modal projection, and calculating the physical residuals in conjunction with the structural dynamics equations, specifically:

[0032] When spectral aliasing is determined, the coherence function between measurement points and all valid measurement point signals with a coherence consistency flag of 0 in the cross-spectral phase are calculated. The candidate fluid pulsation components are projected onto a preset acoustic mode base, the modal projection energy concentration is calculated, and the fuzzy components with an acoustic mode attribution flag of 0 in the physical residual are calculated in combination with the one-dimensional wave equation of the pipeline. Frequency domain averaging and set aggregation are performed respectively to construct the frequency domain characterization of the candidate vibration noise signal and the fuzzy components to be arbitrated. The incoherent measurement point signals and the acoustic mode attribution doubtful components are aggregated to establish a unified input set for mechanical mode projection and critical arbitration.

[0033] Call the pre-calibrated mechanical modal basis and verify its normalized state to ensure that the mathematical basis of the structural vibration modal projection is strictly valid;

[0034] Calculate the projection coefficients of candidate vibration noise signals onto the mechanical mode basis, and quantify the coupling strength between the signals and the structural vibration modes;

[0035] The energy concentration projected onto the mechanical modal basis is calculated, and the modal projection reconstruction signal is substituted into the structural dynamics equation to calculate the physical residual. The energy proportion of the mechanical modal projection and the fit of the structural dynamics equation are evaluated, providing a dual quantitative criterion for the clear attribution of vibration and noise.

[0036] Based on the joint criterion of projected energy concentration and structural dynamic residual, the modal attribution of candidate vibration noise components is determined. Mechanical mode dual verification is performed on incoherent measurement point signals. Those with clear attribution are directly eliminated, and those with ambiguous attribution are transferred to the arbitration process.

[0037] If the mechanical mode attribution flag is 1, then the candidate component is determined to be a mechanical vibration component and is eliminated.

[0038] If the mechanical mode attribution flag is 0, the candidate component is determined to be of unclear attribution and is projected onto the acoustic mode basis and mechanical mode basis respectively as a critical fuzzy component for reconstruction. The dual-modal physical residual is calculated and minimized through arbitration.

[0039] The components with unclear attribution and the candidate fluid pulsation components are projected onto the preset acoustic modal basis, the modal projection energy concentration is calculated, and the fuzzy components of the physical residual back are calculated in combination with the one-dimensional wave equation of the pipeline. They are then projected and reconstructed onto the acoustic modal basis and the mechanical modal basis respectively. The dual-modal physical residual is calculated and minimized. For the critical components where the single-modal criterion fails, dual-modal residual minimization arbitration is implemented. The physical equation fit is used as the final attribution basis.

[0040] If the final attribution flag of the bimodal arbitration is 1, then the determination residual points to the acoustic mode, and this component is retained as fluid pulsation;

[0041] If the final attribution flag of the dual-modal arbitration is 0, the residual is determined to point to the mechanical mode, and this component is discarded as vibration noise.

[0042] The final assignment status of all frequency band components is summarized, and a frequency domain characterization of the pure fluid pulsation signal after processing by the axial coherent dual-mode projection decoupling method is constructed. All frequency band components identified as fluid pulsation are aggregated, and the pure pressure pulsation frequency domain signal after three-stage joint decoupling by the axial coherent dual-mode projection decoupling method is reconstructed. For the components identified as fluid pulsation, the stability of the pressure wave propagation speed is determined based on the temperature change rate and the gas content fluctuation.

[0043] As an optional solution to the hydraulic pipe fitting testing error compensation method based on intelligent sensors described in this invention, the stability of the pressure wave propagation speed is determined based on the temperature change rate and gas content fluctuation for the component identified as fluid pulsation, specifically:

[0044] Acquire current and historical values ​​of fluid medium temperature, gas content, and pipe material elastic modulus, constructing a length of [missing information]. The sliding analysis window provides a continuous operating condition data basis for dynamic stability analysis;

[0045] The average rate of temperature change is calculated within a sliding window to quantify the severity of temperature drift and identify time-varying operating conditions that exceed the steady-state tolerance.

[0046] The standard deviation of gas content is calculated within a sliding window as a volatility index to quantify the degree of medium inhomogeneity caused by gas precipitation or mixing.

[0047] The volumetric elastic modulus of hydraulic oil is dynamically corrected based on real-time temperature to establish a temperature-property coupling relationship, ensuring that the wave velocity calculation reflects the true state of the medium.

[0048] Based on the corrected elastic modulus, oil modulus and density, the theoretical pressure wave velocity under the current working condition is updated to provide a dynamic reference for the expected value of phase delay.

[0049] By comparing the steady-state tolerance thresholds corresponding to the rate of temperature change and the fluctuation of gas content, it is independently determined whether the temperature and gas content are in the steady-state range.

[0050] If the temperature steady-state flag is output as 1, then the temperature is determined to be in a steady state.

[0051] If the steady-state temperature flag output is 0, it is determined that the temperature is changing drastically.

[0052] If the steady-state flag for gas content is output as 1, then the gas content is determined to be in a steady state.

[0053] If the steady-state gas content flag output is 0, then the gas content fluctuation is determined to be excessive.

[0054] The steady-state temperature flag and the steady-state gas content flag are logically ANDed to determine the wave velocity stability level and subsequent processing path. The wave velocity stability level is determined by combining the steady-state temperature and gas content flags. If it is stable, the separated fluid pulsation signal is directly reconstructed by inverse transformation. The residual error uncertainty is calculated and compared with the accuracy threshold. If it is time-varying, the expected value of phase delay is updated and the coherence function and cross-spectral phase are recalculated within the adaptive window.

[0055] If the comprehensive flag output of wave velocity stability is 1, it is determined that the temperature and gas content are both in the steady state range and the pressure wave propagation speed is constant. When spectral aliasing is determined, the fixed phase delay assumption of the coherence function and cross-spectral phase between measurement points is valid. When spectral aliasing is determined, the coherence function and cross-spectral phase between measurement points are calculated to project the candidate vibration noise components and the components with unclear attribution onto the preset mechanical mode basis. The separation result of calculating the energy concentration of the mechanical mode projection and combining it with the structural dynamics equation to calculate the physical residual is used as the intermediate result of error compensation. The fluid pulsation signal obtained after separation is directly reconstructed by inverse transformation. The residual error uncertainty is calculated and compared with the accuracy threshold for final accuracy verification.

[0056] If the output of the wave velocity stability comprehensive flag is 0, it is determined that the temperature change rate or gas content fluctuation exceeds the steady-state tolerance threshold, and the pressure wave propagation speed has undergone significant time variation. When spectral aliasing is determined to exist, the fixed phase delay assumption of the coherence function and cross-spectral phase between measurement points is assumed to have a mismatch risk. The expected value of the phase delay is updated, and the coherence function and cross-spectral phase are recalculated within the adaptive window to start the time-varying propagation speed adaptive coherent tracking correction.

[0057] As an optional scheme of the hydraulic pipe fitting testing error compensation method based on intelligent sensors described in this invention, the following steps are taken: updating the expected value of the phase delay and recalculating the coherence function and cross-spectral phase within an adaptive window:

[0058] Based on the current real-time temperature and gas content, interpolation is performed on the pre-established three-dimensional calibration surface of wave velocity-temperature-gas content to obtain the actual pressure wave velocity under the current working conditions and establish a dynamic mapping relationship between temperature, gas content and wave velocity.

[0059] Based on the actual wave speed obtained by interpolation, the theoretical propagation delay and expected phase difference between each measuring point are recalculated. The theoretical propagation delay and expected phase difference are then recalculated using the actual wave speed, replacing the outdated expected value under the fixed wave speed assumption.

[0060] The adaptive length of the coherence analysis window is determined based on the relative rate of change of wave velocity. When the wave velocity changes drastically, the window is shortened to track quickly, and when it is stable, the window is lengthened to improve frequency resolution.

[0061] Within a sliding adaptive window, the cross-spectral density is re-estimated, and the updated coherence function and cross-spectral phase are calculated to obtain the measured frequency domain coupling characteristics that match the current wave velocity.

[0062] Search for the peak value of the adaptive coherence function within the overlapping frequency range, and compare the degree of agreement between the adaptive cross-spectral phase and the updated expected phase difference to determine whether the signal attribution has recovered after the wave velocity correction.

[0063] If the adaptive coherence consistency flag is 1, it is determined that the corrected coherence peak value and the updated expected phase delay are rematched. The corrected fluid pulsation component is extracted and the candidate fluid pulsation component is projected onto the preset acoustic mode base. The modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline to verify the acoustic mode attribution.

[0064] If the adaptive coherence consistency flag is output, it is determined that the adaptive tracking still cannot match. The adaptive window length is expanded by a preset step size, and the updated coherence function and cross-spectral phase are recalculated and the degree of match between the adaptive cross-spectral phase and the updated expected phase difference is compared. At the same time, an expansion upper limit is set, and the matching is restored through the window iteration expansion strategy. If the upper limit is reached and the matching still fails, the array verification is triggered to avoid misjudging it as fluid pulsation.

[0065] If the sensor array status verification trigger flag is 1, then adaptive tracking is determined to have failed, and the sensor array status verification process is triggered.

[0066] If the sensor array status verification trigger flag is 0, the match is successful after the window is expanded, and the corrected fluid pulsation component is extracted and returned to project the candidate fluid pulsation component onto the preset acoustic mode base. The modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline.

[0067] The present invention has the following beneficial effects:

[0068] 1. This hydraulic pipe fitting test error compensation method based on intelligent sensors effectively avoids the separation failure problem caused by traditional filtering methods under spectral overlap conditions by constructing a spectrum aliasing risk judgment and path diversion mechanism, thereby improving the signal processing adaptability under complex working conditions.

[0069] 2. This hydraulic pipe fitting testing error compensation method based on intelligent sensors introduces the joint constraint of coherence function and cross-spectral phase to achieve a preliminary distinction between fluid pulsation and mechanical vibration. On this basis, it combines acoustic modal and mechanical modal projection analysis to establish a dual criterion of "energy concentration + physical residual". Compared with the single signal feature determination method, it significantly improves the accuracy of signal attribution judgment.

[0070] 3. The hydraulic fitting test error compensation method based on intelligent sensors sets up a dual-modal physical residual minimization arbitration mechanism to make physical consistency judgment on critical aliasing components that are difficult to determine in traditional methods. This avoids the uncertainty caused by the selection of empirical thresholds and improves the stability and reliability of decoupling results.

[0071] 4. This intelligent sensor-based hydraulic pipe fitting testing error compensation method effectively reduces the impact of temperature fluctuations and medium state changes on phase analysis by constructing a coupling relationship between temperature, gas content, and wave velocity, and performing dynamic correction and adaptive coherent tracking when the wave velocity changes over time, thereby improving the testing accuracy in complex environments.

[0072] 5. This intelligent sensor-based hydraulic pipe fitting testing error compensation method introduces a residual error uncertainty assessment and accuracy threshold closed-loop verification mechanism to achieve self-verification and dynamic correction of the error compensation process, ensuring that the compensation results continuously meet the test specification requirements, reducing the need for frequent calibration of the reference pipeline, and improving test efficiency. Attached Figure Description

[0073] Figure 1 This is a flowchart of the hydraulic fitting testing error compensation method based on intelligent sensors according to the present invention;

[0074] Figure 2 This is a flowchart of the axial coherent dual-mode projection decoupling method according to claims 4-6 of the present invention;

[0075] Figure 3 The flowchart of the time-varying wave velocity adaptive coherent tracking method according to claims 7-8 of this invention is shown below.

[0076] Figure 4 This is a schematic diagram of the overall device structure and signal flow of the present invention. Detailed Implementation

[0077] Example 1: A method for compensating for testing errors in hydraulic pipe fittings based on intelligent sensors, see reference [link to documentation]. Figure 1 ,include:

[0078] Multiple signals of pressure, acceleration, temperature and gas content are collected, and clock synchronization and numerical validity are jointly verified. The set of valid measurement points is selected based on comprehensive validity criteria.

[0079] Frequency domain analysis is performed on the effective pressure signal and acceleration signal to extract their respective main frequency bands and calculate the overlap bandwidth ratio. The results are compared with a preset threshold to determine the risk of spectral aliasing and to determine the processing path.

[0080] When spectral aliasing is determined, the coherence function and cross-spectral phase between measurement points are calculated. Within the overlapping frequency band, the candidate fluid pulsation component and candidate vibration noise component are determined by the joint constraint of coherence peak value and phase deviation.

[0081] The candidate fluid pulsation components are projected onto a preset acoustic modal basis, the modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline. The fluid pulsation is determined based on the joint criterion of energy concentration and physical residual.

[0082] Candidate vibration noise components and components with unclear attribution are projected onto a preset mechanical mode basis. The energy concentration of the mechanical mode projection is calculated, and the physical residual is calculated in combination with the structural dynamics equation. For components that cannot be determined by a single mode, arbitration by minimizing the acoustic and mechanical mode residuals is performed to determine the final attribution.

[0083] For components identified as fluid pulsations, the stability of pressure wave propagation speed is determined based on the rate of temperature change and the fluctuation of gas content.

[0084] When the propagation speed of pressure waves is unstable, the wave speed is dynamically interpolated and corrected based on temperature and gas content, the expected value of phase delay is updated, and the coherence function and cross-spectral phase are recalculated within an adaptive window to achieve coherent tracking correction under time-varying wave speed conditions.

[0085] The separated fluid pulsation signal is reconstructed by inverse transformation, the residual error uncertainty is calculated and compared with the accuracy threshold. If the requirements are met, the compensation result is output; otherwise, the working condition review or recalibration process is triggered.

[0086] Example 2 is an improvement upon Example 1. This method for compensating for testing errors in hydraulic pipe fittings based on intelligent sensors performs joint verification of clock synchronization and numerical validity, and selects a set of valid measurement points based on comprehensive valid indicators. Specifically:

[0087] A raw data vector containing an axial pressure sensor array, an acceleration sensor, a temperature sensor, and a gas content monitoring channel is constructed, and all channels are marked as pending verification. The raw sampling data from the multi-source heterogeneous sensors are aggregated to establish a unified data frame structure, providing a standardized input interface for subsequent hierarchical verification.

[0088] ;

[0089] ;

[0090] In the formula, For discrete sampling time index, , For the first The original data vector of multi-source signals at any given time. For the first The first axial pressure measuring point The original sampled value at time 10:00. , This represents the total number of measuring points for the axial pressure sensor. For the accelerometer sensor Original sampled value at time, For temperature sensor number Original sampled value at time, For the gas content sensor Original sampled value at time, The initial state label vector for the channel. For the first The pressure channel initial state flag, with a value of 1, indicates that it has entered the verification state. , , These are the initial state indicators for the acceleration, temperature, and gas content channels, respectively.

[0091] Calculate the deviation of each pressure channel sampling timestamp relative to the master clock reference timestamp to determine if clock synchronization failure exists, identify timing misalignments caused by communication delays or sampling clock drift, and prevent asynchronous data from entering coherence analysis and causing pseudo-coherence or false phase.

[0092] ;

[0093] ;

[0094] In the formula, For the first Pressure Channel timestamp deviation at sampling time For the first The actual sampling timestamp of the channel. The master clock reference timestamp is output by the system synchronization clock module. This is the sampling clock synchronization tolerance threshold. This is a clock synchronization check flag;

[0095] If the clock synchronization check flag If the output is 1, the clock synchronization of this channel is deemed to be qualified.

[0096] If the clock synchronization check flag If the output is 0, it is determined that the channel has a clock synchronization error.

[0097] Verify whether the sampled values ​​of each channel exceed the measurement range, and whether the changes in sampled values ​​at adjacent time points exceed the physically reasonable threshold. Eliminate outliers caused by sensor range overflow, electromagnetic interference spikes, and communication packet loss to ensure that the sampled values ​​used in subsequent calculations are physically reliable.

[0098] ;

[0099] ;

[0100] In the formula, For the first Pressure Channel The sampling jump variable at a given time relative to the previous time. , These are the lower and upper limits of the sensor's measurement range, respectively. This is the maximum allowable jump threshold between adjacent sampling times. This serves as a marker of numerical validity.

[0101] If the numerical validity flag If the output is 1, the value of that channel is considered normal.

[0102] If the numerical validity flag If the output is 0, it is determined that the channel has saturation or abrupt change anomalies;

[0103] A logical AND operation is performed between the clock synchronization flag and the numerical validity flag to obtain a comprehensive validity flag. The total number of valid measurement points is counted, and the data quality level is determined. By combining the results of the dual verification of clock and numerical values, the redundancy of valid measurement points is quantified, providing a rigid admission criterion for whether axial coherence analysis conditions are met.

[0104] ;

[0105] ;

[0106] ;

[0107] In the formula, For the first The pressure channel validity flag is obtained by a logical AND operation between the clock synchronization flag and the numerical validity flag. If... and If so, the channel is deemed to be valid. Take 1, if or If so, the channel is deemed to be in a state of overall failure. Take 0, For the first The total number of valid pressure measurement points at the sampling time This represents the minimum number of effective measurement points required for axial array coherence analysis. This serves as a marker of data quality compliance.

[0108] If the data quality is qualified If the output is 1, the data quality is deemed acceptable. Frequency domain analysis is performed on the effective pressure signal and acceleration signal to extract their respective main frequency bands and calculate the overlap bandwidth ratio.

[0109] If the data quality is qualified If the output is 0, it is determined that the data quality does not meet the joint decoupling condition, and the sensor array self-test and channel switching process is triggered. That is, communication handshake and power supply diagnosis are performed channel by channel to identify channels with communication interruption or sampling abnormality. Power-off restart is performed on abnormal channels. If the comprehensive validity flag of the channel changes from 0 to 1 after restart, it is determined that the channel has recovered and is re-included in the array. If the comprehensive validity flag of the channel is still 0 after restart, it is determined that the channel has a hardware failure, and it is automatically switched to the redundant backup channel and clock synchronization and numerical validity joint verification are performed again. The set of valid measurement points is selected based on the comprehensive validity flag.

[0110] This embodiment also provides frequency domain analysis of the effective pressure signal and acceleration signal, extracting their respective main frequency bands and calculating the overlap bandwidth ratio, specifically:

[0111] The process involves joint clock synchronization and numerical validity verification, followed by selection of a valid measurement point set based on comprehensive validity indicators to obtain the pressure channel signal after comprehensive validity verification. Invalid channel data is then eliminated to construct a valid signal set for spectrum analysis. Based on the joint clock synchronization and numerical validity verification, and the selection of a valid measurement point set based on comprehensive validity indicators, invalid channels are filtered out to ensure that only valid measurement point signals enter the frequency domain analysis, thus avoiding aliasing detection due to abnormal channels.

[0112] ;

[0113] In the formula, For the first Pressure Channel The effective sampled values ​​that participate in the spectrum analysis at all times. These are the original sampled values. To perform joint verification of clock synchronization and numerical validity, a comprehensive valid flag is output based on the selection of a set of valid measurement points using a comprehensive valid flag.

[0114] If the comprehensive valid indicators If the output is 1, the channel is considered valid, and the valid sample value is... Original sampled values ;

[0115] If the comprehensive valid indicators If the output is 0, the channel is considered invalid, and the valid sample value is... =0;

[0116] Discrete Fourier transforms are performed on the effective pressure and acceleration signals, and the power spectral density is estimated. The time-domain signals are then mapped to the frequency domain, and the energy distribution of each frequency component is quantified, providing a frequency domain representation for subsequent main frequency band extraction and overlap calculation.

[0117] ;

[0118] ;

[0119] ;

[0120] ;

[0121] In the formula, For the first Pressure channel at discrete frequency spectral components at that location, For acceleration channels at discrete frequencies spectral components at that location, For the first A discrete frequency point, , For frequency resolution, This refers to the number of sampling points per frame. The imaginary unit satisfies , It is a natural constant. The power spectral density of the pressure channel. Power spectral density of the acceleration channel;

[0122] Based on the power spectral density significance threshold, the dominant frequency distribution intervals of pressure and acceleration signals are extracted separately. From the entire frequency band, the dominant frequency intervals with significant energy are screened out, low-energy noise bases are excluded, and the focus is on the frequency bands that are actually dominant in pressure pulsation and mechanical vibration.

[0123] ;

[0124] ;

[0125] In the formula, For the first The set of main frequency bands of the pressure channel is composed of the union of continuous frequency intervals whose power spectral density exceeds the significance threshold. For the main frequency band of acceleration signals, The significance coefficient is extracted from the dominant frequency, and its value range is [value range missing]. , Represents all discrete frequency points Take the maximum value. The union operator represents merging all consecutive frequency ranges that meet the conditions into a total main frequency band.

[0126] The proportion of the overlap between the pressure and acceleration frequency bands within the band of interest is calculated to the total bandwidth of the pressure frequency band. This quantifies the degree of overlap between the pressure pulsation frequency band and the mechanical vibration frequency band, providing a numerical criterion for evaluating whether conventional filtering methods can achieve lossless separation.

[0127] ;

[0128] ;

[0129] ;

[0130] ;

[0131] In the formula, The overlapping frequency range of the pressure main frequency band and the acceleration main frequency band is determined by intersection operation. Sure, The total bandwidth of the overlapping frequency range. For frequency differential elements, For the first Total bandwidth of the main frequency band of the pressure channel. This represents the percentage of overlapping bandwidth.

[0132] Based on the comparison between the overlapping bandwidth ratio and a preset threshold, the risk level of spectral aliasing is determined and the subsequent processing path is identified. A rigid quantitative threshold for the degree of spectral overlap is established, and the test conditions are diverted to a joint decoupling process or a conventional filtering channel to avoid method mismatch.

[0133] ;

[0134] In the formula, For the first Risk indicator of spectral aliasing in pressure channels The preset separation difficulty threshold is set based on the transition band characteristics of conventional digital filters and the test accuracy requirements;

[0135] If the overlap bandwidth ratio Greater than or equal to the preset separation difficulty threshold If the hydraulic pulsation and mechanical vibration exhibit spectral aliasing, and conventional filtering methods cannot separate them without loss, then the risk of spectral aliasing in the pressure channel is identified. Take 1 and when spectral aliasing is determined to exist, calculate the coherence function between measurement points and the cross-spectral phase to start the axial coherence-dual-mode projection joint decoupling process of the pipeline;

[0136] If the overlap bandwidth ratio Less than the preset separation difficulty threshold If the spectrum is determined to be separable, then the risk of spectral aliasing in the pressure channel is identified. After setting the value to 0 and performing error compensation using conventional digital filtering, the separated fluid pulsation signal is directly reconstructed by inverse transformation. The residual error uncertainty is calculated and compared with the accuracy threshold for accuracy verification and output.

[0137] This embodiment also provides that the separated fluid pulsation signal is reconstructed by inverse transformation, the residual error uncertainty is calculated and compared with the accuracy threshold, specifically:

[0138] Candidate vibration noise components and components with unclear attribution are projected onto a preset mechanical mode basis. The energy concentration of the mechanical mode projection is calculated, and the physical residual or the expected value of the updated phase delay is calculated in combination with the structural dynamics equations. The pure fluid pulsation frequency domain characterization of the coherence function and cross-spectral phase output is recalculated within an adaptive window and subjected to inverse discrete Fourier transform to reconstruct the time domain waveform. The pure fluid pulsation components separated in the frequency domain are mapped back to the time domain through inverse transform to reconstruct the physical pressure waveform that can be used for pipe performance evaluation.

[0139] ;

[0140] In the formula, For the first Time-domain sampled value of pure fluid pulsation signal at any given moment. To project candidate vibration noise components and components with unclear attribution onto a preset mechanical modal basis, the energy concentration of the mechanical modal projection is calculated, and the physical residual or the expected value of the updated phase delay is calculated in conjunction with the structural dynamics equations. Within an adaptive window, the clean signal spectral components of the coherence function and cross-spectral phase output are recalculated. This refers to the number of sampling points per frame. The imaginary unit satisfies , It is a natural constant;

[0141] The arithmetic mean of the original signals from all valid measurement points is used to construct a mixed signal benchmark. This benchmark is then differentially analyzed point-by-point with the pure signal to obtain the residual error sequence. This process is repeated to construct the original mixed signal arithmetic mean benchmark, and the compensation residual sequence is extracted by differentially analyzing point-by-point with the pure signal, providing error samples for accuracy assessment.

[0142] ;

[0143] ;

[0144] In the formula, For the first The original mixed signal reference value at time point, The total number of valid measurement points. For the set of valid measurement point indices, For the first Residual error at time;

[0145] Calculate the mean, standard deviation, and maximum absolute deviation of the residual error sequence; synthesize the standard uncertainty of the compensated signal; statistically analyze the residual error distribution characteristics and synthesize the standard uncertainty; quantify the degree of deviation between the compensated signal and the original mixed signal.

[0146] ;

[0147] ;

[0148] ;

[0149] ;

[0150] In the formula, The mean of the residual error. The standard deviation of the residual error. The maximum absolute deviation of the residual error. For the combined standard uncertainty, Indicates all sampling times Take the maximum value;

[0151] The synthesized standard uncertainty is compared with the accuracy threshold specified in the test procedure to establish a rigid comparison between the residual uncertainty and the procedure threshold, and to determine whether the compensation result meets the accuracy level required for the performance evaluation of the pipe fittings.

[0152] ;

[0153] In the formula, This serves as a mark for determining accuracy and compliance. The accuracy threshold specified in the test procedure;

[0154] If the accuracy compliance judgment mark If the output is 1, the error compensation is deemed valid.

[0155] If the accuracy compliance judgment mark If the output is 0, it indicates that the error compensation has not converged.

[0156] The final processing path is determined based on the status of the accuracy compliance judgment mark. If it is qualified, the final pure pressure signal is output; if it is not qualified, the working condition re-evaluation or benchmark recalibration process is initiated.

[0157] If the error compensation is deemed effective, then the pure fluid pressure pulsation signal will be... As the final error compensation result output, it is used to evaluate the sealing performance and pressure-bearing deformation performance of pipe fittings;

[0158] If the error compensation fails to converge, the process is backtracked to the component identified as fluid pulsation. Based on the temperature change rate and gas content fluctuation, the stability of the pressure wave propagation speed is reassessed to re-evaluate the stability of the operating conditions. Alternatively, the recalibration process of the reference pipeline is triggered. This involves installing the metrologically verified standard reference pipeline into the test station, collecting the output values ​​of each sensor under no-load and rated pressure conditions, reconstructing the original error reference spectrum and acoustic-mechanical modal basis, synchronously updating and performing joint verification of clock synchronization and numerical validity, selecting the set of valid measurement points based on comprehensive valid indicators, performing inverse transformation reconstruction on the separated fluid pulsation signal, calculating the residual error uncertainty and comparing it with all relevant discrimination thresholds, and re-executing the complete error compensation process.

[0159] Example 3 is an improvement upon Example 2. (See attached document for details.) Figure 2 In this embodiment, when spectral aliasing is determined to exist, the coherence function and cross-spectral phase between measurement points are calculated, specifically as follows:

[0160] After clock synchronization and numerical validity verification, the effective measurement point set is selected based on the comprehensive validity flag. The installation position coordinates of all effective pressure measurement points along the pipe fitting axis are obtained. The geometric spacing between each pair of measurement points is calculated. The axial geometric coordinates and spacing of the measurement points are obtained. The physical scale benchmark required for spatial propagation analysis is established to support the subsequent accurate calculation of phase delay.

[0161] ;

[0162] In the formula, For the first measuring point and the first The geometric spacing of the measuring points along the axial direction of the pipe fitting. For the first The coordinates of the measuring point's installation position along the pipe fitting's axial direction. For the first The coordinates of the measuring point's installation position along the pipe fitting's axial direction;

[0163] Based on the fluid medium properties and pipeline structural parameters, the theoretical pressure wave velocity under the current operating conditions is calculated, and the expected propagation delay and phase difference between each measuring point are derived accordingly. Based on the fluid-pipeline coupled wave velocity model, the theoretical propagation delay and expected phase difference are derived, providing a physical reference for coherent matching.

[0164] ;

[0165] ;

[0166] ;

[0167] In the formula, For the theoretical pressure wave velocity, The elastic modulus of the pipe fitting material. The bulk modulus of hydraulic oil. The density of the hydraulic oil, The inner diameter of the pipe fitting. For the pipe wall thickness, For the first measuring point and the first Theoretical propagation delay between measurement points For discrete frequencies Theoretical expected phase difference at;

[0168] Segmented windowing and frequency domain averaging are applied to the effective pressure measurement point signals to estimate the cross-spectral density between each measurement point pair, suppress random noise, and extract stable frequency domain coupling features.

[0169] ;

[0170] In the formula, For the first measuring point and the first The signal at the measuring point is at the frequency Cross-spectral density estimation at [location] The average number of segments, For the first Measurement point number After windowing processing, the segment signal is at frequency spectral components at that location, This is a complex conjugate operation;

[0171] The coherence function is calculated from the cross-spectral density and the self-spectral density, and the cross-spectral phase is calculated from the cross-spectral density argument, quantifying the degree of linear correlation and phase propagation relationship between measurement points:

[0172] ;

[0173] ;

[0174] In the formula, For the first measuring point and the first The signal at the measuring point is at the frequency coherence function value at , The first , No. The autospectral density of the signal at the measurement point, For cross-spectral phase angle, For complex argument operations, To perform operations on the imaginary part, To perform the operation of taking the real part;

[0175] Frequency domain analysis is performed on the effective pressure signal and acceleration signal. The dominant frequency bands of each signal are extracted, and the peak value of the coherent function is searched within the overlapping frequency range determined by the overlap bandwidth ratio. The degree of agreement between the measured phase difference and the theoretically expected phase difference is compared to determine whether the signal belongs to axial coherent fluid pulsation or local incoherent vibration.

[0176] ;

[0177] ;

[0178] ;

[0179] In the formula, Overlapping frequency range Peak value of internal coherence function The frequency point corresponding to the coherent peak value. This represents the absolute deviation between the measured phase difference and the expected phase difference. The coherence threshold, This is the phase tolerance threshold. For measuring point pair Coherence consistency indicator;

[0180] If the measuring point is correct If the coherence consistency flag output is 1, it is determined that the measurement point has axial propagation coherence to the signal, which is extracted as a candidate fluid pulsation signal and the candidate fluid pulsation component is projected onto the preset acoustic mode base. The modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline.

[0181] If the measuring point is correct If the coherence consistency flag output is 0, the signal of the measurement point is determined to be a local incoherent vibration. It is extracted as a candidate vibration noise signal and the candidate vibration noise components and the components with unclear attribution are projected onto the preset mechanical mode basis. The energy concentration of the mechanical mode projection is calculated and the physical residual is calculated in combination with the structural dynamics equation.

[0182] This embodiment also provides that the candidate fluid pulsation component is projected onto a preset acoustic mode base, the modal projection energy concentration is calculated, and the physical residual is calculated in conjunction with the one-dimensional wave equation of the pipeline, specifically:

[0183] When spectral aliasing is determined, all valid measurement point signals with a coherence consistency flag of 1 in the inter-measurement coherence function and cross-spectral phase are obtained. Frequency domain averaging is performed within the coherence band to construct a frequency domain representation of the candidate fluid pulsation signal. The candidate signals extracted from the inter-measurement coherence function and cross-spectral phase when spectral aliasing is determined are then reconstructed and represented in the frequency domain to establish the frequency domain input required for subsequent modal projection.

[0184] ;

[0185] In the formula, For candidate fluid pulsation signals at frequency The average spectral components at that location, To determine the coherence consistency flag in the cross-spectral phase and the coherence function between measurement points when spectral aliasing is detected, the calculation is performed. The set of measurement point indices For set The number of measuring points in the middle, For the first The signal at the measuring point is at the frequency spectral components at the location;

[0186] Obtain the pre-calibrated acoustic modal basis and verify its normalized state to ensure that the mathematical basis of modal projection is strictly valid:

[0187] ;

[0188] ;

[0189] In the formula, The acoustic modal basis matrix is ​​formed by... The first-order acoustic mode shape function is composed of... For the first First-order acoustic mode shape function, , This is the axial length of the pipeline. For the first Modal normalization coefficients, It is an axial spatial differential element. From 0 to along the axial direction of the pipeline Definite integral operations;

[0190] Calculate the projection coefficients of the candidate fluid pulsation signal onto the acoustic mode basis, that is, calculate the projection coefficients of the candidate signal onto each order of acoustic mode, and quantify the coupling strength between the signal and the fluid pulsation mode:

[0191] ;

[0192] In the formula, For candidate signals in the th Projection coefficients on the first acoustic mode, For the first The coordinates of the installation position of the measuring point along the axial direction of the pipeline;

[0193] Calculate the proportion of the energy projected onto the acoustic mode basis to the total energy of the candidate signal, evaluate the proportion of projected energy to the total candidate energy, and determine whether the signal energy is concentrated in the acoustic mode subspace:

[0194] ;

[0195] ;

[0196] ;

[0197] In the formula, For frequency The sum of energies projected onto the acoustic modal basis. For frequency The total energy of the candidate signal, For projected energy concentration, For modal order From 1 to Summation;

[0198] Substituting the modally reconstructed signal into the one-dimensional wave equation of the pipeline, calculating the physical residual, and verifying whether the signal conforms to the fluid propagation law:

[0199] ;

[0200] ;

[0201] In the formula, For the candidate signal reconstructed by modal projection in the th Measurement point frequency spectral components at that location, It is a second-order difference operator in axial space. Angular frequency, For frequency Physical residuals of the pipeline fluid dynamics equations at the location;

[0202] Based on the joint criterion of projected energy concentration and physical residual, the modal assignment of candidate components is determined. By combining the dual criteria of energy concentration and physical residual, the candidate components are accurately assigned to real fluid pulsation or suspected vibration noise.

[0203] ;

[0204] In the formula, As a criterion for determining acoustic mode attribution, The threshold for projected energy concentration. The threshold value for the residual constraint of the fluid dynamics equations;

[0205] If the acoustic mode attribution criterion If the value is 1, the candidate component is determined to be a real fluid pulsation component and is retained. For components determined to be fluid pulsations, the stability of the pressure wave propagation speed is determined based on the temperature change rate and the gas content fluctuation.

[0206] If the acoustic mode attribution criterion If the value is 0, the candidate component is deemed to have questionable attribution. In this case, the component is preferentially transferred to the preset mechanical modal basis for projecting the candidate vibration noise component and the component with unclear attribution. The energy concentration of the mechanical modal projection is calculated, and the physical residual is calculated in combination with the structural dynamics equation as a vibration noise candidate to participate in the mechanical modal projection verification and dual-modal residual arbitration. If the attribution still cannot be clearly determined after arbitration, the process is backtracked to when spectral aliasing is determined, and the coherence function between measurement points and the cross-spectral phase adjustment coherence boundary threshold are calculated for re-screening.

[0207] This embodiment also provides that candidate vibration noise components and components with unclear attribution are projected onto a preset mechanical modal basis, the energy concentration of the mechanical modal projection is calculated, and the physical residual is calculated in conjunction with the structural dynamics equations. Specifically:

[0208] When spectral aliasing is detected, the system calculates the coherence function between measurement points and all valid measurement point signals with a coherence consistency flag of 0 in the cross-spectral phase. It also projects candidate fluid pulsation components onto a preset acoustic mode base, calculates the modal projection energy concentration, and calculates the fuzzy components with an acoustic mode attribution flag of 0 in the physical residual using the one-dimensional wave equation of the pipeline. Frequency domain averaging and ensemble aggregation are then performed to construct frequency domain representations of candidate vibration noise signals and fuzzy components to be arbitrated. Finally, it aggregates incoherent measurement point signals and acoustic mode attribution doubtful components to establish a unified input set for mechanical mode projection and critical arbitration.

[0209] ;

[0210] ;

[0211] In the formula, For candidate vibration noise signals at frequency The average spectral components at that location, To determine the coherence consistency flag in the cross-spectral phase and the coherence function between measurement points when spectral aliasing is detected, the calculation is performed. The set of measurement point indices For set The number of measuring points in the middle, To project the candidate fluid pulsation components onto a preset acoustic modal basis, the modal projection energy concentration is calculated, and the fuzzy component frequency domain representation set of the physical residual back-out is calculated in conjunction with the one-dimensional wave equation of the pipeline. To project the candidate fluid pulsation components onto a preset acoustic modal basis, the modal projection energy concentration is calculated, and the frequency domain characterization of the candidate fluid pulsation signal is constructed by combining the one-dimensional wave equation of the pipeline with the physical residual.

[0212] The pre-calibrated mechanical modal basis is invoked, and its normalized state is verified to ensure that the mathematical basis of the structural vibration modal projection is strictly valid.

[0213] ;

[0214] ;

[0215] In the formula, The mechanical modal basis matrix is ​​formed by... The mechanical modal vibration mode functions are composed of the first order. Let p be the p-th order mechanical mode shape function. , This represents the total number of mechanical modal orders. For the first Normalized coefficients of mechanical modes, This is the axial length of the pipeline. It is an axial spatial differential element. From 0 to along the axial direction of the pipeline Definite integral operations;

[0216] Calculate the projection coefficients of candidate vibration noise signals onto the mechanical mode basis, and quantize the coupling strength between the signal and the structural vibration modes:

[0217] ;

[0218] In the formula, For the candidate vibration noise signal in the th Projection coefficients on the first mechanical mode, For the first The coordinates of the installation position of the measuring point along the axial direction of the pipeline;

[0219] The energy concentration projected onto the mechanical modal basis is calculated, and the modal projection reconstruction signal is substituted into the structural dynamics equations to calculate the physical residuals. The energy proportion of the mechanical modal projection and the fit of the structural dynamics equations are evaluated, providing dual quantitative criteria for the clear attribution of vibration and noise.

[0220] ;

[0221] ;

[0222] ;

[0223] ;

[0224] ;

[0225] In the formula, For frequency The sum of energies projected onto the mechanical modal basis at a given point. For frequency The total energy of the candidate vibration noise signal, For mechanical modal projection energy concentration, The vibration noise signal reconstructed from mechanical modal projection is in the first... Measurement point frequency spectral components at that location, For the propagation speed of structural waves, For frequency Physical residuals of the structural dynamics equations at the location;

[0226] Based on the joint criterion of projected energy concentration and structural dynamic residuals, the modal attribution of candidate vibration noise components is determined. Mechanical modal dual verification is performed on incoherent measurement point signals. Components with clearly defined attributions are directly eliminated, while those with ambiguous attributions are transferred to the arbitration process.

[0227] ;

[0228] In the formula, This serves as a criterion for determining the mechanical mode attribution. The threshold for the projected energy concentration of the mechanical modality. The residual threshold is set as a constraint on the structural dynamics equations;

[0229] If the mechanical mode attribution determination flag If the value is 1, the candidate component is determined to be a mechanical vibration component and is discarded.

[0230] If the mechanical mode attribution determination flag If the value is 0, the candidate component is determined to be of unclear ownership and is used as a critical fuzzy component to be projected and reconstructed onto the acoustic modal basis and mechanical modal basis respectively. The dual-modal physical residual is calculated and minimized in arbitration.

[0231] The components with unclear attribution and the candidate fluid pulsation components are projected onto a preset acoustic modal basis. The modal projection energy concentration is calculated, and the fuzzy components returned by the physical residual are calculated in conjunction with the one-dimensional wave equation of the pipeline. These components are then projected and reconstructed onto the acoustic and mechanical modal bases respectively. The dual-modal physical residual is calculated and minimized. For critical components where the single-modal criterion fails, dual-modal residual minimization arbitration is implemented. The fit of the physical equation is used as the final attribution criterion.

[0232] ;

[0233] ;

[0234] ;

[0235] ;

[0236] ;

[0237] In the formula, After the components to be arbitrated are projected onto the acoustic modal basis and reconstructed, the first... Measurement point frequency Spectral components at that location For the component to be arbitrated in the Projection coefficients on the first acoustic mode, After the components to be arbitrated are projected onto the mechanical modal basis and reconstructed, the first... Measurement point frequency Spectral components at that location For the component to be arbitrated in the Projection coefficients on the first mechanical mode, For acoustic modal projection physical residuals, For the physical residuals of the mechanical modal projection, As the final attribution marker in bimodal arbitration;

[0238] If the final outcome of the dual-modal arbitration is determined by the following criteria: If the value is 1, the residual is determined to point to the acoustic mode, and this component is retained as fluid pulsation;

[0239] If the final outcome of the dual-modal arbitration is determined by the following criteria: If the value is 0, the residual is determined to point to the mechanical mode, and this component is discarded as vibration noise;

[0240] The final assignment status of all frequency band components is summarized, and a frequency domain representation of the pure fluid pulsation signal after processing by the axial coherent dual-mode projection decoupling method is constructed. All frequency band components identified as fluid pulsations are aggregated, and the pure pressure pulsation frequency domain signal after three-stage joint decoupling by the axial coherent dual-mode projection decoupling method is reconstructed. For the components identified as fluid pulsations, the stability of the pressure wave propagation speed is determined based on the temperature change rate and gas content fluctuation.

[0241] ;

[0242] In the formula, To determine the presence of spectral aliasing, the coherence function and cross-spectral phase between measurement points are calculated. Candidate vibration noise components and components with unclear attribution are projected onto a preset mechanical mode base. The energy concentration of the projected mechanical modes is calculated, and the physical residuals are determined by combining the structural dynamics equations. Finally, at the frequency... The pure fluid pulsation signal spectral components retained at that location. This involves projecting the candidate fluid pulsation components onto a preset acoustic modal basis, calculating the modal projection energy concentration, and summing all frequency band components with an acoustic modal attribution flag of 1 in the physical residual, based on the one-dimensional wave equation of the pipeline. This means that for components whose classification is unclear and candidate fluid pulsation components are projected onto a preset acoustic modal basis, the modal projection energy concentration is calculated, and the fuzzy components of the physical residual back are calculated in combination with the one-dimensional wave equation of the pipeline. The components are then projected and reconstructed onto the acoustic modal basis and the mechanical modal basis respectively. The dual-modal physical residual is calculated and the summation of all frequency band components with the dual-modal arbitration classification flag of 1 is performed to minimize the arbitration.

[0243] Example 4 is an improvement upon Example 3. (See attached document for details.) Figure 3 In this embodiment, for the component determined to be fluid pulsation, the stability of the pressure wave propagation speed is determined based on the rate of temperature change and the fluctuation of gas content, specifically as follows:

[0244] Acquire current and historical values ​​of fluid medium temperature, gas content, and pipe material elastic modulus, constructing a length of [missing information]. The sliding analysis window provides a continuous operating condition data foundation for dynamic stability analysis:

[0245] ;

[0246] ;

[0247] In the formula, For the first The sliding time-series window vector of the temperature monitoring values ​​at any given time. For the first The sliding time-series window vector of the gas content monitoring value at each moment. For the first Real-time fluid medium temperature monitoring value, For the first real-time gas content monitoring value The length of the sliding window. Index for discrete sampling time;

[0248] Calculate the average rate of temperature change within a sliding window, quantify the severity of temperature drift, and identify time-varying operating conditions that exceed the steady-state tolerance:

[0249] ;

[0250] In the formula, For the first Average rate of temperature change over time. The time interval between adjacent sampling times. To calculate the absolute difference between adjacent sampling points within the window from to Summation;

[0251] The standard deviation of gas content is calculated within a sliding window as a volatility index to quantify the degree of inhomogeneity of the medium caused by gas precipitation or mixing.

[0252] ;

[0253] ;

[0254] In the formula, This represents the average gas content within the sliding window. For the first Temporal gas content fluctuation;

[0255] The bulk modulus of hydraulic oil is dynamically corrected based on real-time temperature to establish a temperature-property coupling relationship, ensuring that the wave velocity calculation reflects the actual state of the medium.

[0256] ;

[0257] In the formula, For the first The volumetric elastic modulus of hydraulic oil after temperature correction. Reference temperature The bulk modulus of elasticity of the hydraulic oil under the following conditions. To calibrate the reference temperature, The temperature sensitivity coefficient of the bulk modulus of hydraulic oil. It is a natural exponential function;

[0258] Based on the corrected elastic modulus, oil modulus, and density, the theoretical pressure wave velocity under the current operating conditions is updated to provide a dynamic reference for the expected phase delay value:

[0259] ;

[0260] In the formula, For the first Theoretical pressure wave velocity at any given moment For the first The constant value of the elastic modulus of the pipe fitting material. For the first Hydraulic oil density at all times The inner diameter of the pipe fitting. For pipe fitting wall thickness;

[0261] By comparing the steady-state tolerance thresholds corresponding to the rate of temperature change and the fluctuation of gas content, it is independently determined whether the temperature and gas content are within the steady-state range:

[0262] ;

[0263] ;

[0264] In the formula, As a marker of steady-state temperature, The steady-state tolerance threshold for the rate of temperature change. As a steady-state indicator of gas content, This is the steady-state tolerance threshold for gas content fluctuation.

[0265] If the temperature steady state indicator If the output is 1, the temperature is determined to be in a steady state.

[0266] If the temperature steady state indicator If the output is 0, it indicates that the temperature has changed drastically.

[0267] If the steady-state indicator of gas content If the output is 1, it is determined that the gas content is in a steady state;

[0268] If the steady-state indicator of gas content If the output is 0, it is determined that the gas content fluctuation exceeds the limit;

[0269] A logical AND operation is performed between the steady-state temperature flag and the steady-state gas content flag to determine the wave velocity stability level and subsequent processing path. Combining the steady-state temperature and gas content flags, the wave velocity stability level is determined. If stable, the separated fluid pulsation signal is directly reconstructed using an inverse transform, and the residual error uncertainty is calculated and compared with the accuracy threshold. If time-varying, the expected phase delay value is updated, and the coherence function and cross-spectral phase are recalculated within an adaptive window.

[0270] ;

[0271] In the formula, As a comprehensive indicator of wave velocity stability;

[0272] If the comprehensive indicator of wave speed stability If the output is 1, it is determined that the temperature and gas content are both in the steady state range and the pressure wave propagation speed is constant. When spectral aliasing is determined, the fixed phase delay assumption of the coherence function and cross-spectral phase between measurement points is valid. When spectral aliasing is determined, the coherence function and cross-spectral phase between measurement points are calculated to project the candidate vibration noise components and the components with unclear attribution onto the preset mechanical mode basis. The mechanical mode projection energy concentration is calculated and the separation result of the physical residual is calculated in combination with the structural dynamics equation as the intermediate result of error compensation. The fluid pulsation signal obtained after separation is directly reconstructed by inverse transformation. The residual error uncertainty is calculated and compared with the accuracy threshold for final accuracy verification.

[0273] If the comprehensive indicator of wave speed stability If the output is 0, it is determined that the rate of temperature change or gas content fluctuation exceeds the steady-state tolerance threshold, and the pressure wave propagation speed has undergone significant time variation. When spectral aliasing is determined to exist, the fixed phase delay assumption of the coherence function and cross-spectral phase between measurement points is assumed to have a mismatch risk. The expected value of the phase delay is updated, and the coherence function and cross-spectral phase are recalculated within the adaptive window to initiate time-varying propagation speed adaptive coherent tracking correction.

[0274] This embodiment also provides updating the expected phase delay value and recalculating the coherence function and cross-spectral phase within an adaptive window, specifically:

[0275] Based on the current real-time temperature and gas content, interpolation is performed on a pre-established three-dimensional calibration surface of wave velocity-temperature-gas content to obtain the actual pressure wave velocity under the current operating conditions, and to establish a dynamic mapping relationship between temperature, gas content, and wave velocity.

[0276] ;

[0277] In the formula, For the first The actual pressure wave velocity obtained by interpolation at any given time The wave velocity-temperature-gas content three-dimensional calibration mapping operator is established by measuring discrete operating points and fitting surfaces during the calibration stage. For the first Real-time temperature of the fluid medium For the first Real-time monitoring value of gas content at all times;

[0278] Based on the actual wave velocity obtained by interpolation, the theoretical propagation delay and expected phase difference between each measuring point are recalculated. The theoretical propagation delay and expected phase difference are then recalculated using the actual wave velocity, replacing the outdated expected values ​​under the fixed wave velocity assumption.

[0279] ;

[0280] ;

[0281] In the formula, For the first The latest update measuring point and the first Theoretical propagation delay between measurement points For the first After the time update at discrete frequency The expected phase difference at the location;

[0282] The adaptive length of the coherence analysis window is determined based on the relative rate of change of wave velocity. When the wave velocity changes drastically, the window is shortened for fast tracking, and when it is stable, the window is lengthened to improve frequency resolution.

[0283] ;

[0284] In the formula, For the first The window length adapts to the sliding movement at all times, and is a positive integer. The reference window length determined during the calibration phase. The minimum allowed window length, The maximum allowed window length, Let be the window shrinkage sensitivity coefficient, and take a positive real number. For the first The actual pressure wave velocity at any given moment, To perform the maximum value operation, This is for calculating the minimum value;

[0285] Within a sliding adaptive window, the cross-spectral density is re-estimated, and the updated coherence function and cross-spectral phase are calculated to obtain the measured frequency domain coupling characteristics matching the current wave velocity.

[0286] ;

[0287] ;

[0288] ;

[0289] In the formula, For the first Time-adaptive window within the first measuring point and the first The signal at the measuring point is at the frequency Cross-spectral density estimation at [location] For adaptive window within the first Duan Di After windowing processing, the measurement point signal is at the frequency Spectral components at that location For adaptive coherence function values, For adaptive cross-spectral phase angle, For complex argument operations;

[0290] Search for the peak value of the adaptive coherence function within the overlapping frequency range, and compare the degree of agreement between the adaptive cross-spectral phase and the updated expected phase difference to determine whether the signal attribution has recovered to a matching state after wave velocity correction.

[0291] ;

[0292] ;

[0293] ;

[0294] In the formula, For the overlapping frequency range within the adaptive window The peak value of the coherence function in the middle. The frequency point corresponding to the adaptive coherence peak. To adapt to the absolute deviation between the measured phase difference and the updated expected phase difference, This serves as an adaptive coherence consistency flag.

[0295] If adaptive coherence consistency flag If the output is 1, it is determined that the corrected coherent peak value and the updated expected phase delay are rematched, the corrected fluid pulsation component is extracted and the candidate fluid pulsation component is projected onto the preset acoustic mode base, the modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline to verify the acoustic mode attribution.

[0296] If adaptive coherence consistency flag If the output is negative, it indicates that adaptive tracking still fails to match. The adaptive window length is then expanded by a preset step size, and the updated coherence function and cross-spectral phase are recalculated. The degree of agreement between the adaptive cross-spectral phase and the updated expected phase difference is compared. An expansion upper limit is set, and matching is restored through an iterative window expansion strategy. If the upper limit is reached and the problem persists, array verification is triggered to avoid misjudging it as fluid pulsation.

[0297] ;

[0298] ;

[0299] In the formula, For the expanded adaptive window length, Expand the window step size, taking a positive integer. This serves as a trigger flag for sensor array status verification.

[0300] If the sensor array status verification trigger flag is active. If the value is 1, the adaptive tracking is deemed to have failed, triggering the sensor array status verification process. This involves checking the physical installation tightness, zero-point drift, and communication transmission delay of all sensors. Sensors deemed abnormal are re-tightened and undergo local zero-point calibration. If the sensor's overall valid flag changes from 0 to 1 after calibration, the performance is deemed to have returned to normal. If the sensor's overall valid flag remains 0 after calibration, the sensor is deemed to have failed. After replacing the spare parts, clock synchronization and numerical validity are re-verified, and a set of valid measurement points is selected based on the overall valid flag.

[0301] If the sensor array status verification trigger flag is active. If the value is 0, it indicates that the match was successful after the window was expanded, and the corrected fluid pulsation component was extracted and returned. The candidate fluid pulsation component was then projected onto the preset acoustic modal basis, the modal projection energy concentration was calculated, and the physical residual was calculated in combination with the one-dimensional wave equation of the pipeline.

[0302] In this embodiment, pressure, acceleration, temperature, and gas content signals are simultaneously acquired and validated using a multi-source sensor array. In the frequency domain, the main frequency band of the pressure and acceleration signals is extracted and overlap analysis is performed to establish a spectral aliasing risk assessment mechanism. When spectral aliasing is detected, the coherence function and cross-spectral phase between measurement points are introduced for joint constraint analysis to extract candidate fluid pulsation components and vibration noise components. Furthermore, the candidate components are projected onto preset acoustic and mechanical modal bases, respectively. A joint criterion is constructed by combining the modal projection energy concentration and the corresponding physical control equation residuals. Components with unclear single-mode determination are then addressed. A dual-mode physical residual minimization arbitration of acoustic and mechanical modes is adopted to achieve high-precision decoupling and separation of fluid pulsation signals and mechanical vibration signals. Based on this, the stability of pressure wave propagation speed is determined by monitoring the rate of temperature change and gas content fluctuation. When the wave speed changes over time, the wave speed is dynamically corrected based on the calibration relationship between temperature and gas content. The coherence function and phase information are recalculated within an adaptive window to achieve coherent tracking compensation under time-varying conditions. Finally, the separated fluid pulsation signal is reconstructed by inverse transformation, and an error compensation closed-loop verification mechanism is constructed by comparing the residual error uncertainty with the test accuracy threshold.

Claims

1. A method for compensating for testing errors in hydraulic pipe fittings based on intelligent sensors, characterized in that: include: Multiple signals of pressure, acceleration, temperature and gas content are collected, and clock synchronization and numerical validity are jointly verified. The set of valid measurement points is selected based on comprehensive validity criteria. Frequency domain analysis is performed on the effective pressure signal and acceleration signal to extract their respective main frequency bands and calculate the overlap bandwidth ratio. The results are compared with a preset threshold to determine the risk of spectral aliasing and to determine the processing path. When spectral aliasing is determined, the coherence function and cross-spectral phase between measurement points are calculated. Within the overlapping frequency band, the candidate fluid pulsation component and candidate vibration noise component are determined by the joint constraint of coherence peak value and phase deviation. The candidate fluid pulsation components are projected onto a preset acoustic modal basis, the modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline. The fluid pulsation is determined based on the joint criterion of energy concentration and physical residual. Candidate vibration noise components and components with unclear attribution are projected onto a preset mechanical mode basis. The energy concentration of the mechanical mode projection is calculated, and the physical residual is calculated in combination with the structural dynamics equation. For components that cannot be determined by a single mode, arbitration by minimizing the acoustic and mechanical mode residuals is performed to determine the final attribution. For components identified as fluid pulsations, the stability of pressure wave propagation speed is determined based on the rate of temperature change and the fluctuation of gas content. When the propagation speed of pressure waves is unstable, the wave speed is dynamically interpolated and corrected based on temperature and gas content, the expected value of phase delay is updated, and the coherence function and cross-spectral phase are recalculated within an adaptive window to achieve coherent tracking correction under time-varying wave speed conditions. The separated fluid pulsation signal is reconstructed by inverse transformation, the residual error uncertainty is calculated and compared with the accuracy threshold. If the requirements are met, the compensation result is output; otherwise, the working condition review or recalibration process is triggered.

2. The hydraulic fitting testing error compensation method based on intelligent sensors according to claim 1, characterized in that: A joint verification of clock synchronization and numerical validity is performed, and a set of valid measurement points is selected based on comprehensive validity criteria. Specifically: A raw data vector containing an axial pressure sensor array, an acceleration sensor, a temperature sensor, and a gas content monitoring channel is constructed, and all channels are marked as pending verification. The raw sampling data of the multi-source heterogeneous sensors are aggregated to establish a unified data frame structure, providing a standardized input interface for subsequent hierarchical verification. Calculate the deviation of the sampling timestamp of each pressure channel relative to the master clock reference timestamp, determine whether there is clock synchronization failure, identify timing misalignment caused by communication delay or sampling clock drift, and prevent asynchronous data from entering coherent analysis and causing pseudo-coherence or false phase. If the clock synchronization check flag is output as 1, then the clock synchronization of this channel is deemed to be qualified. If the clock synchronization check flag outputs 0, it is determined that the channel has a clock synchronization failure. Check whether the sampled values ​​of each channel exceed the range and whether the changes in sampled values ​​at adjacent times exceed the physically reasonable threshold. Eliminate outliers caused by sensor range overflow, electromagnetic interference spikes and communication packet loss to ensure that the sampled values ​​participating in subsequent calculations are physically reliable. If the value validity flag is output as 1, the value of that channel is considered normal. If the numerical validity flag is output as 0, it is determined that the channel has saturation or jump anomalies. The clock synchronization flag and the numerical validity flag are logically ANDed to obtain a comprehensive validity flag. The total number of valid measurement points is counted and the data quality level is determined. The results of the combined clock and numerical verification are used to quantify the redundancy of valid measurement points and provide a rigid admission judgment for whether the conditions for axial coherence analysis are met. If the data quality pass flag output is 1, the data quality is determined to be passable. Frequency domain analysis is performed on the effective pressure signal and acceleration signal to extract their respective main frequency bands and calculate the overlap bandwidth ratio. If the data quality pass flag output is 0, it is determined that the data quality does not meet the joint decoupling conditions, and the sensor array self-test and channel switching process is triggered.

3. The method for compensating for testing errors of hydraulic pipe fittings based on intelligent sensors according to claim 1, characterized in that: Frequency domain analysis was performed on the effective pressure signal and acceleration signal to extract their respective main frequency bands and calculate the overlap bandwidth ratio, specifically: The process involves joint verification of clock synchronization and numerical validity, selection of valid measurement point set based on comprehensive validity indicators to obtain pressure channel signals after comprehensive validity verification, elimination of invalid channel data, construction of valid signal set for spectrum analysis, and filtering of invalid channels based on the verification results of joint verification of clock synchronization and numerical validity and selection of valid measurement point set based on comprehensive validity indicators. This ensures that only valid measurement point signals enter the frequency domain analysis and avoids abnormal channel contamination and aliasing judgment. If the overall valid flag output is 1, then the channel is considered valid, and the valid sample value is the original sample value; If the overall valid flag output is 0, then the channel is determined to be invalid, and the valid sample value is 0. Discrete Fourier transform is performed on the effective pressure signal and acceleration signal, and the power spectral density is estimated. The time domain signal is mapped to the frequency domain, and the energy distribution of each frequency component is quantified, providing a frequency domain characterization for subsequent main frequency band extraction and overlap calculation. Based on the power spectral density significance threshold, the main frequency distribution ranges of pressure signals and acceleration signals are extracted respectively. The dominant frequency ranges with significant energy are screened from the entire frequency band, low-energy noise base is excluded, and the actual dominant frequency bands of pressure pulsation and mechanical vibration are focused. The proportion of the overlap between the pressure main frequency band and the acceleration main frequency band in the frequency band of interest is calculated to the total bandwidth of the pressure main frequency band. The degree of overlap between the pressure pulsation frequency band and the mechanical vibration frequency band is quantified, providing numerical criteria for evaluating whether conventional filtering methods can separate them without loss. Based on the comparison between the overlapping bandwidth ratio and the preset threshold, the risk level of spectrum aliasing is determined and the subsequent processing path is determined. A rigid quantitative threshold for the degree of spectrum overlap is established, and the test conditions are diverted to the joint decoupling process or the conventional filtering channel to avoid method mismatch. If the overlap bandwidth ratio is greater than or equal to the preset separation difficulty threshold, it is determined that there is spectral aliasing between hydraulic pulsation and mechanical vibration. Conventional filtering methods cannot separate them without loss. The spectral aliasing risk flag of the pressure channel is set to 1. When it is determined that there is spectral aliasing, the coherence function between the measurement points and the cross-spectral phase are calculated to start the pipeline axial coherence-dual-modal projection joint decoupling process. If the overlap bandwidth ratio is less than the preset separation difficulty threshold, the spectrum is determined to be separable. The pressure channel spectrum aliasing risk flag is set to 0, and after error compensation is completed by conventional digital filtering, the separated fluid pulsation signal is directly reconstructed by inverse transformation. The residual error uncertainty is calculated and compared with the accuracy threshold for accuracy verification and output.

4. The hydraulic fitting testing error compensation method based on intelligent sensors according to claim 1, characterized in that: When determining the presence of spectral aliasing, the coherence function and cross-spectral phase between measurement points are calculated, specifically as follows: After clock synchronization and numerical validity verification, the installation position coordinates of all valid pressure measuring points along the pipe fitting axis are obtained based on the comprehensive validity flag and the set of valid measuring points is selected. The geometric distance between each pair of measuring points is calculated, the axial geometric coordinates and distance of the measuring points are obtained, and the physical scale benchmark required for spatial propagation analysis is established to support the subsequent accurate calculation of phase delay. Based on the fluid medium physical properties and pipeline structural parameters, the theoretical pressure wave velocity under the current working condition is calculated, and the expected propagation delay and phase difference between each measuring point are derived accordingly. Based on the fluid-pipeline coupled wave velocity model, the theoretical propagation delay and expected phase difference are derived to provide a physical reference benchmark for coherent matching. Segmented windowing and frequency domain averaging are applied to the effective pressure measurement point signals to estimate the cross-spectral density between each measurement point pair, suppress random noise, and extract stable frequency domain coupling features. The coherence function is calculated from the cross-spectral density and the self-spectral density, and the cross-spectral phase is calculated from the cross-spectral density argument, thus quantifying the degree of linear correlation and phase propagation relationship between the measurement points. Frequency domain analysis is performed on the effective pressure signal and acceleration signal to extract their respective main frequency bands and search for the peak value of the coherent function within the overlapping frequency range determined by the overlap bandwidth ratio. The degree of agreement between the measured phase difference and the theoretical expected phase difference is compared to determine whether the signal belongs to axial coherent fluid pulsation or local incoherent vibration. If the coherence consistency flag of the measuring point pair is 1, it is determined that the signal of the measuring point pair has axial propagation coherence, which is extracted as a candidate fluid pulsation signal and the candidate fluid pulsation component is projected onto the preset acoustic mode base. The modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline. If the coherence consistency flag of the measuring point pair is 0, the signal of the measuring point pair is determined to be local incoherent vibration. It is extracted as a candidate vibration noise signal and the candidate vibration noise component and the component with unclear affiliation are projected onto the preset mechanical mode basis. The energy concentration of the mechanical mode projection is calculated and the physical residual is calculated in combination with the structural dynamics equation.

5. The method for compensating for testing errors of hydraulic pipe fittings based on intelligent sensors according to claim 1, characterized in that: The candidate fluid pulsation components are projected onto a preset acoustic modal basis, the modal projection energy concentration is calculated, and the physical residuals are calculated in conjunction with the one-dimensional wave equation of the pipeline. Specifically: When spectral aliasing is determined, all valid measurement point signals with a coherence consistency flag of 1 in the calculation of the coherence function and cross-spectral phase between measurement points are obtained. Frequency domain averaging is performed in the coherence band to construct the frequency domain representation of the candidate fluid pulsation signal. The candidate signals extracted from the calculation of the coherence function and cross-spectral phase between measurement points when spectral aliasing is determined are reconstructed and represented in the frequency domain to establish the frequency domain input required for subsequent modal projection. Obtain the pre-calibrated acoustic modal basis and verify its normalized state to ensure that the mathematical basis of modal projection is strictly valid; Calculate the projection coefficients of candidate fluid pulsation signals onto the acoustic mode basis, and quantize the coupling strength between the signal and the fluid pulsation mode; Calculate the proportion of the energy projected onto the acoustic mode basis to the total energy of the candidate signal, evaluate the proportion of the projected energy to the total energy of the candidate signal, and determine whether the signal energy is concentrated in the acoustic mode subspace; Substitute the signal reconstructed by modal projection into the one-dimensional wave equation of the pipeline, calculate the physical residual, and verify whether the signal conforms to the fluid propagation law. Based on the joint criterion of projected energy concentration and physical residual, the modal assignment of candidate components is determined. By combining the dual criteria of energy concentration and physical residual, the candidate components are accurately assigned to real fluid pulsation or suspected vibration noise. If the acoustic mode attribution flag is 1, the candidate component is determined to be a real fluid pulsation component and is retained. For components determined to be fluid pulsations, the stability of pressure wave propagation speed is determined based on the temperature change rate and gas content fluctuation. If the acoustic mode attribution determination flag is 0, the attribution of the candidate component is questionable. In this case, the component is preferentially transferred to the preset mechanical mode basis for projecting the candidate vibration noise component and the component with unclear attribution. The energy concentration of the mechanical mode projection is calculated and the physical residual is calculated in combination with the structural dynamics equation as a vibration noise candidate to participate in the mechanical mode projection verification and dual-mode residual arbitration. If the attribution still cannot be determined after arbitration, the process is back to when spectral aliasing is determined, and the coherence function between measurement points and the cross-spectral phase adjustment coherence boundary threshold are calculated for re-screening.

6. The method for compensating for testing errors of hydraulic pipe fittings based on intelligent sensors according to claim 1, characterized in that: The candidate vibration noise components and components with unclear attribution are projected onto a preset mechanical modal basis. The energy concentration of the mechanical modal projection is calculated, and the physical residuals are calculated in conjunction with the structural dynamics equations. Specifically: When spectral aliasing is determined, the coherence function between measurement points and all valid measurement point signals with a coherence consistency flag of 0 in the cross-spectral phase are calculated. The candidate fluid pulsation components are projected onto a preset acoustic mode base, the modal projection energy concentration is calculated, and the fuzzy components with an acoustic mode attribution flag of 0 in the physical residual are calculated in combination with the one-dimensional wave equation of the pipeline. Frequency domain averaging and set aggregation are performed respectively to construct the frequency domain characterization of the candidate vibration noise signal and the fuzzy components to be arbitrated. The incoherent measurement point signals and the acoustic mode attribution doubtful components are aggregated to establish a unified input set for mechanical mode projection and critical arbitration. Call the pre-calibrated mechanical modal basis and verify its normalized state to ensure that the mathematical basis of the structural vibration modal projection is strictly valid; Calculate the projection coefficients of candidate vibration noise signals onto the mechanical mode basis, and quantize the coupling strength between the signals and the structural vibration modes; The energy concentration projected onto the mechanical modal basis is calculated, and the modal projection reconstruction signal is substituted into the structural dynamics equation to calculate the physical residual. The energy proportion of the mechanical modal projection and the fit of the structural dynamics equation are evaluated, providing a dual quantitative criterion for the clear attribution of vibration and noise. Based on the joint criterion of projected energy concentration and structural dynamic residual, the modal attribution of candidate vibration noise components is determined. Mechanical mode dual verification is performed on incoherent measurement point signals. Those with clear attribution are directly eliminated, and those with ambiguous attribution are transferred to the arbitration process. If the mechanical mode attribution flag is 1, then the candidate component is determined to be a mechanical vibration component and is eliminated. If the mechanical mode attribution flag is 0, the candidate component is determined to be of unclear attribution and is projected onto the acoustic mode basis and mechanical mode basis respectively as a critical fuzzy component for reconstruction. The dual-modal physical residual is calculated and minimized through arbitration. The components with unclear attribution and the candidate fluid pulsation components are projected onto the preset acoustic modal basis, the modal projection energy concentration is calculated, and the fuzzy components of the physical residual back are calculated in combination with the one-dimensional wave equation of the pipeline. They are then projected and reconstructed onto the acoustic modal basis and the mechanical modal basis respectively. The dual-modal physical residual is calculated and minimized. For the critical components where the single-modal criterion fails, dual-modal residual minimization arbitration is implemented. The physical equation fit is used as the final attribution basis. If the final attribution flag of the bimodal arbitration is 1, then the determination residual points to the acoustic mode, and this component is retained as fluid pulsation; If the final attribution flag of the dual-modal arbitration is 0, the residual is determined to point to the mechanical mode, and this component is discarded as vibration noise. The final assignment status of all frequency band components is summarized, and a frequency domain characterization of the pure fluid pulsation signal after processing by the axial coherent dual-mode projection decoupling method is constructed. All frequency band components identified as fluid pulsation are aggregated, and the pure pressure pulsation frequency domain signal after three-stage joint decoupling by the axial coherent dual-mode projection decoupling method is reconstructed. For the components identified as fluid pulsation, the stability of the pressure wave propagation speed is determined based on the temperature change rate and the gas content fluctuation.

7. The hydraulic fitting testing error compensation method based on intelligent sensors according to claim 1, characterized in that: For components identified as fluid pulsations, the stability of the pressure wave propagation velocity is determined based on the rate of temperature change and the fluctuation of gas content, specifically as follows: Acquire current and historical values ​​of fluid medium temperature, gas content, and pipe material elastic modulus, constructing a length of [missing information]. The sliding analysis window provides a continuous operating condition data basis for dynamic stability analysis; The average rate of temperature change is calculated within a sliding window to quantify the severity of temperature drift and identify time-varying operating conditions that exceed the steady-state tolerance. The standard deviation of gas content is calculated within a sliding window as a volatility index to quantify the degree of medium inhomogeneity caused by gas precipitation or mixing. The volumetric elastic modulus of hydraulic oil is dynamically corrected based on real-time temperature to establish a temperature-property coupling relationship, ensuring that the wave velocity calculation reflects the true state of the medium. Based on the corrected elastic modulus, oil modulus and density, the theoretical pressure wave velocity under the current working condition is updated to provide a dynamic benchmark for the expected value of phase delay. By comparing the steady-state tolerance thresholds corresponding to the rate of temperature change and the fluctuation of gas content, it is independently determined whether the temperature and gas content are in the steady-state range. If the temperature steady-state flag is output as 1, then the temperature is determined to be in a steady state. If the steady-state temperature flag output is 0, it is determined that the temperature is changing drastically. If the steady-state flag for gas content is output as 1, then the gas content is determined to be in a steady state. If the steady-state gas content flag output is 0, then the gas content fluctuation is determined to be excessive. The steady-state temperature flag and the steady-state gas content flag are logically ANDed to determine the wave velocity stability level and subsequent processing path. The wave velocity stability level is determined by combining the steady-state temperature and gas content flags. If it is stable, the separated fluid pulsation signal is directly reconstructed by inverse transformation. The residual error uncertainty is calculated and compared with the accuracy threshold. If it is time-varying, the expected value of phase delay is updated and the coherence function and cross-spectral phase are recalculated within the adaptive window. If the comprehensive flag output of wave velocity stability is 1, it is determined that the temperature and gas content are both in the steady state range and the pressure wave propagation speed is constant. When spectral aliasing is determined, the fixed phase delay assumption of the coherence function and cross-spectral phase between measurement points is valid. When spectral aliasing is determined, the coherence function and cross-spectral phase between measurement points are calculated to project the candidate vibration noise components and the components with unclear attribution onto the preset mechanical mode basis. The separation result of calculating the energy concentration of the mechanical mode projection and combining it with the structural dynamics equation to calculate the physical residual is used as the intermediate result of error compensation. The fluid pulsation signal obtained after separation is directly reconstructed by inverse transformation. The residual error uncertainty is calculated and compared with the accuracy threshold for final accuracy verification. If the output of the wave velocity stability comprehensive flag is 0, it is determined that the temperature change rate or gas content fluctuation exceeds the steady-state tolerance threshold, and the pressure wave propagation speed has undergone significant time variation. When spectral aliasing is determined to exist, the fixed phase delay assumption of the coherence function and cross-spectral phase between measurement points is assumed to have a mismatch risk. The expected value of the phase delay is updated, and the coherence function and cross-spectral phase are recalculated within the adaptive window to start the time-varying propagation speed adaptive coherent tracking correction.

8. The method for compensating for testing errors of hydraulic pipe fittings based on intelligent sensors according to claim 1, characterized in that: Update the expected phase delay and recalculate the coherence function and cross-spectral phase within the adaptive window, specifically as follows: Based on the current real-time temperature and gas content, interpolation is performed on the pre-established three-dimensional calibration surface of wave velocity-temperature-gas content to obtain the actual pressure wave velocity under the current working conditions and establish a dynamic mapping relationship between temperature, gas content and wave velocity. Based on the actual wave speed obtained by interpolation, the theoretical propagation delay and expected phase difference between each measuring point are recalculated. The theoretical propagation delay and expected phase difference are then recalculated using the actual wave speed, replacing the outdated expected value under the fixed wave speed assumption. The adaptive length of the coherence analysis window is determined based on the relative rate of change of wave velocity. When the wave velocity changes drastically, the window is shortened to track quickly, and when it is stable, the window is lengthened to improve frequency resolution. Within a sliding adaptive window, the cross-spectral density is re-estimated, and the updated coherence function and cross-spectral phase are calculated to obtain the measured frequency domain coupling characteristics that match the current wave velocity. Search for the peak value of the adaptive coherence function within the overlapping frequency range, and compare the degree of agreement between the adaptive cross-spectral phase and the updated expected phase difference to determine whether the signal attribution has recovered after the wave velocity correction. If the adaptive coherence consistency flag is 1, it is determined that the corrected coherence peak value and the updated expected phase delay are rematched. The corrected fluid pulsation component is extracted and the candidate fluid pulsation component is projected onto the preset acoustic mode base. The modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline to verify the acoustic mode attribution. If the adaptive coherence consistency flag is output, it is determined that the adaptive tracking still cannot match. The adaptive window length is expanded by a preset step size, and the updated coherence function and cross-spectral phase are recalculated and the degree of match between the adaptive cross-spectral phase and the updated expected phase difference is compared. At the same time, an expansion upper limit is set, and the matching is restored through the window iteration expansion strategy. If the upper limit is reached and the matching still fails, the array verification is triggered to avoid misjudging it as fluid pulsation. If the sensor array status verification trigger flag is 1, then adaptive tracking is determined to have failed, and the sensor array status verification process is triggered. If the sensor array status verification trigger flag is 0, the match is successful after the window is expanded, and the corrected fluid pulsation component is extracted and returned to project the candidate fluid pulsation component onto the preset acoustic mode base. The modal projection energy concentration is calculated, and the physical residual is calculated in combination with the one-dimensional wave equation of the pipeline.

9. The method for compensating for testing errors of hydraulic pipe fittings based on intelligent sensors according to claim 1, characterized in that: The separated fluid pulsation signal is reconstructed by inverse transformation, and the residual error uncertainty is calculated and compared with the accuracy threshold. Specifically: Candidate vibration noise components and components with unclear attribution are projected onto a preset mechanical mode basis. The energy concentration of the mechanical mode projection is calculated, and the physical residual or the expected value of the updated phase delay is calculated in combination with the structural dynamics equation. The pure fluid pulsation frequency domain characterization of the coherence function and cross-spectral phase output is recalculated within an adaptive window and inverse discrete Fourier transform is performed to reconstruct the time domain waveform. The pure fluid pulsation components separated in the frequency domain are mapped back to the time domain through inverse transform to reconstruct the physical pressure waveform that can be used for pipe performance evaluation. The arithmetic mean of the original signals from all valid measurement points is used to construct a mixed signal benchmark. The residual error sequence is obtained by point-by-point difference with the pure signal. The original mixed signal arithmetic mean benchmark is constructed, and the compensation residual sequence is extracted by point-by-point difference with the pure signal to provide error samples for accuracy assessment. Calculate the mean, standard deviation, and maximum absolute deviation of the residual error sequence; synthesize the standard uncertainty of the compensated signal; statistically analyze the residual error distribution characteristics and synthesize the standard uncertainty; quantify the degree of deviation between the compensated signal and the original mixed signal. The synthesized standard uncertainty is compared with the accuracy threshold specified in the test procedure to establish a rigid comparison between the residual uncertainty and the procedure threshold, and to determine whether the compensation result reaches the accuracy level required for the performance evaluation of the pipe fittings. If the accuracy compliance judgment flag is output as 1, then the judgment error compensation is effective; If the accuracy compliance judgment flag output is 0, then the judgment error compensation has not converged. The final processing path is determined based on the status of the accuracy compliance judgment mark. If it is qualified, the final pure pressure signal is output; if it is not qualified, the working condition re-evaluation or benchmark recalibration process is initiated. If the error compensation is deemed effective, the pure fluid pressure pulsation signal will be output as the final error compensation result for evaluating the sealing performance and pressure-bearing deformation performance of the pipe fittings. If the error compensation fails to converge, the process will backtrack to the component identified as fluid pulsation, and reassess the stability of the operating condition based on the temperature change rate and gas content fluctuation to determine the stability of the pressure wave propagation speed, or trigger the recalibration process of the reference pipeline.

Images