Carbon can venting resistance test data compensation method based on dynamic flow characteristics

Abstract

The present application relates to the field of carbon tank ventilation detection, and particularly relates to a carbon tank ventilation resistance test data compensation method based on dynamic flow characteristics, a dynamic baseline of a fixture link is collected first, and a fixed electrical delay is separated; after a carbon tank to be tested is installed, a dynamic analysis segment is intercepted with a flow change event as an index, and homologous frequency domain characteristics of flow and pressure difference are extracted to obtain an effective excitation frequency band; a dynamic response is decomposed into a same-phase resistance component and a quadrature lag component by aerodynamic-electricity equivalence, then a flow-pressure difference amplitude square coherence degree is used for two-channel consistency arbitration, and only a pressure difference component driven by stable flow excitation is received; then local incremental resistance of each flow bucket is reconstructed into a steady-state pressure difference-flow relationship, and after viscosity and inertia coefficients are inversely solved, a compensation secant ventilation resistance is output. The present application can use loading transition section information, eliminate cavity charging and discharging and external disturbance interference, and obtain a ventilation resistance detection result with high repeatability and few misjudgments under fast rhythm.

Description

Technical Field

[0001] The present invention relates to the field of canister ventilation detection, and specifically relates to a canister ventilation resistance test data compensation method based on dynamic flow characteristics. Background Art

[0002] The canister is a core component in the automotive fuel evaporative emission control system, with activated carbon filled in its housing, which is used to adsorb the oil-gas volatilized in the fuel tank during parking and refueling processes, and desorb the adsorbed oil-gas with the help of fresh air when the engine is running, then deliver the desorbed oil-gas to the air intake system for combustion, so as to inhibit the direct escape of oil-gas into the atmosphere. The pressure drop formed when the airflow passes through the canister, that is, the ventilation resistance, is directly related to the desorption efficiency of the fuel evaporation system and the vehicle emission performance, and is a key quality indicator that must be assessed for each canister before leaving the factory.

[0003] Since the canister is a mass-produced automotive component, its ventilation resistance usually needs to be detected online piece by piece on the production line by special testing equipment. Such testing equipment generally includes a sealing fixture, a proportional flow valve, an air pump, a flow sensor, a differential pressure sensor and a data processor. The proportional flow valve ventilates the canister at a set target flow rate, and then converts the measured differential pressure into the ventilation resistance to determine whether the canister is qualified, and the whole process needs to be completed within a short production line tact time.

[0004] However, the air pump inevitably has periodic pulsation during operation, and the activated carbon cavity inside the canister needs to go through a transition process of gas charging and discharging during ventilation before it tends to be stable. In the existing testing method, the equipment mostly adopts a fixed steady-state waiting time, and after ventilating for a period of time, takes the average value of the terminal differential pressure to calculate the ventilation resistance. To take into account the production line tact time, this waiting time is often set relatively short, and sampling is performed as if the gas has not really stabilized and the cavity charging and discharging process has not ended, so that the hysteresis pressure drop that should belong to dynamic energy storage is also included in the ventilation resistance. As a result, the ventilation resistance readings of canisters of the same specification fluctuate significantly when repeatedly tested on the production line, and some canisters that should be qualified are misjudged as having excessive resistance and rejected, which not only causes an artificially low actual yield of the production line, but also makes the test conclusion difficult to be stably reproduced, bringing obvious troubles to quality control in mass production.

[0005] Chinese patent application CN121521405A discloses a method for monitoring the local resistance of a heat storage body and controlling valve opening. Under a fixed air / gas flow rate, it fits the monthly average pressure difference across the heat storage body with the cumulative number of months of use to obtain a curve showing the pressure difference changing with the cumulative months. Based on fluid mechanics principles, it establishes a mapping relationship between the local resistance coefficient and the pressure difference, thereby predicting the pressure difference for the target month and determining the valve opening adjustment scheme. However, this method averages the pressure difference on a monthly basis, essentially representing long-term steady-state trend monitoring. This makes it unsuitable for rapid, piece-by-piece ventilation resistance detection under production line cycle times, and it also fails to separate the dynamic hysteresis pressure drop introduced during the cavity filling and discharging transition process.

[0006] Chinese patent application CN121561848A discloses a method for correlation analysis and coupling management of pressure pulsation and vibration response in pumping stations. This method synchronously collects pressure and vibration signals from multiple sources, extracts time-frequency domain features after preprocessing, calculates the coherence coefficients of pressure pulsation and vibration response using wavelet coherence analysis, and establishes the transfer function between the two through subspace identification, thereby evaluating the coupling state and dynamically controlling the system. Although this method involves coherence analysis and frequency domain processing, it focuses on the coupling relationship between two physical quantities: pressure pulsation and vibration. Its purpose is to suppress vibration and ensure equipment safety. However, it does not separate the dissipative components that can be factored into ventilation resistance from the flow excitation and differential pressure response, nor does it address the problem of misjudgment of ventilation resistance caused by the lag in cavity energy storage. Summary of the Invention

[0007] To address the issue of significant fluctuations in ventilation resistance readings during repeated testing of carbon canisters on the production line due to the fixed steady-state waiting time of the aforementioned equipment, this invention proposes a carbon canister ventilation resistance test data compensation method based on dynamic flow characteristics. The method includes: acquiring the flow rate sequence and pressure difference sequence of the carbon canister under test; S1, acquiring the dynamic baseline of the fixture link, linearly fitting the expanded phase of the coherent effective frequency points, identifying the phase slope that is linear with frequency and does not change with average flow rate as a fixed electrical delay, and using the flow-related phase generated by the fixture cavity and pressure tapping pipe as the residual fixture baseline, storing the two separately; S2, using the moment when the flow rate change exceeds a set threshold as the starting point of dynamic loading, extracting the synchronous flow rate and pressure difference sequences before and after it to form a dynamic test segment, dividing it into overlapping pulses, each with complete coverage. S3. Extract the flow autopower spectrum, pressure differential autopower spectrum, and cross-power spectrum using the flow channel as a reference to determine the effective excitation frequency band covering the low-frequency transition band and pump pulsation frequency band; S4. Calculate the complex impedance after performing a fixed electrical delay correction only once for each window, deduct the residual fixture baseline interpolated according to the average flow rate of the window, take the real part as the resistance component in phase with the flow rate, and the imaginary part as the dynamic hysteresis component approximately orthogonal to the flow rate; S5. Arbitrate the reliable frequency point based on the coherence of the flow rate and pressure differential channel, and obtain the local incremental resistance of each window by weighting the coherence and flow spectrum power; S6. Reconstruct the local incremental resistance of each window into a steady-state pressure differential-flow rate relationship according to the average flow rate, inversely calculate the constitutive coefficients and convert them into compensated secant ventilation resistance before uploading via the communication interface.

[0008] Compared to the commonly used method of "taking the end average after waiting for steady state" in existing carbon canister ventilation resistance testing, this invention fundamentally changes the measurement paradigm. It allows readings to be taken without waiting for the flow rate and pressure difference to fully stabilize, thus overcoming long-standing pain points of traditional methods such as long test cycles per unit, direct discarding of massive amounts of data during transition and pump pulsation phases, and poor repeatability due to the randomness of the initial phase of pump pulsation. This invention utilizes the flow rate step transition and the pump pulsation inversely as natural excitation signals, extracting effective information from dynamic data discarded by existing technologies, significantly improving testing efficiency. Crucially, this invention quantitatively separates the real resistance component in phase with the flow rate and the dynamic hysteresis component, which is approximately orthogonal to the flow rate and generated by the energy stored in the clamp cavity, by decomposing the complex impedance into real and imaginary parts. This invention solves the problem of existing methods being unable to distinguish between true ventilation resistance and cavity energy storage lag, leading to contamination of measurement results, thus giving the obtained resistance a clear physical meaning. Simultaneously, this invention pre-collects the dynamic baseline of the fixture link, separates and stores the fixed electrical delay linear with frequency and the residual fixture baseline related to flow rate, and subtracts it window by window, effectively eliminating the systematic phase error introduced by the measurement link itself. Furthermore, it arbitrates the reliable frequency point using the coherence of the flow rate and differential pressure channels and weights it to further suppress noise interference. Finally, it uses constitutive inverse calculation to convert it into a compensating secant ventilation resistance, making the dynamic measurement results equivalent to the steady-state resistance consistent with existing specifications. Therefore, while significantly shortening the cycle time, it also considers measurement accuracy, repeatability, and production line compatibility, demonstrating outstanding engineering application value.

[0009] Furthermore, the dynamic baseline of the acquisition fixture link includes: installing a standard straight-through component on the fixture that has the same sealing port and installation posture as the carbon canister but does not contain an activated carbon cavity, or installing a standard throttling component whose own impedance has been pre-calibrated by the metering bench; dividing the differential pressure and flow cross power spectrum obtained synchronously in the baseline acquisition by the flow self power spectrum to obtain the total complex impedance, and then subtracting the known complex impedance of the standard component to obtain the residual fixture baseline.

[0010] Compared to methods that rely solely on experience or approximate formulas to deduct the influence of the fixture, this invention installs a standard straight-through component that matches the sealing port and installation posture of the carbon canister but does not contain the activated carbon cavity, or installs a standard throttling component whose impedance has been pre-calibrated on a metrological bench. By using the known complex impedance as a reference for actual measurement and deducting the contribution of the fixture itself, the calibration of the residual fixture baseline has a traceable metrological basis. This avoids the miscalculation of the fixture impedance into the carbon canister resistance. When changing fixtures or retesting across equipment, the consistency and accuracy of the baseline can still be maintained, thereby improving the reliability of the compensation results.

[0011] Furthermore, identifying the phase slope as a fixed electrical delay also includes: the linear fitting is performed at a frequency point where both coherence and flow spectrum power meet the effective conditions, and the fixed electrical delay is accepted when the range and retest deviation of the candidate delays in each flow interval do not exceed the preset tolerance.

[0012] Compared to directly using single-point phase or coarsely estimating the delay, this invention performs linear fitting on the expanded phase only at frequency points where both coherence and flow spectrum power meet the effective conditions. It uses the candidate delay range and whether the retest deviation exceeds the tolerance as the acceptance criterion, effectively avoiding baseline separation distortion caused by misjudging the flow-related phase as a fixed delay. This ensures the reliability and reproducibility of fixed electrical delay identification, and makes subsequent delay correction stable and consistent in different batch tests.

[0013] Furthermore, the process of extracting and forming dynamic test segments also includes: converting the collected flow rate into standard state volumetric flow rate according to calibration parameters; the dynamic loading start point is the moment when the flow rate change rate exceeds a set multiple of the static noise change rate or exceeds a set ratio of the target flow rate within a continuously preset time period; and identifying the main pulsation frequency corresponding to the pulsation period by autocorrelation or short-time spectrum tracking.

[0014] Compared to methods that rely on manual judgment of loading time or fixed window length for data extraction, this invention first converts the flow rate into standard state volumetric flow rate to eliminate the caliber deviation caused by temperature and pressure differences. Then, it automatically captures the dynamic loading start point using dual criteria: the flow rate change rate exceeds the static noise multiple or the target flow rate ratio. Furthermore, it adaptively identifies the main pulsation frequency using autocorrelation or short-time spectrum tracking, making dynamic segment extraction and window division fully automated. This allows it to adapt to different pump sources and operating conditions, improving the automation level and universality of the testing process.

[0015] Furthermore, it also includes anomaly monitoring: when the flow sensor or differential pressure sensor fails to update for a set number of consecutive sampling periods, the sampling exceeds the range, or the dual-channel timestamp deviation exceeds the set number of sampling periods, the generation of the current dynamic test segment is stopped, and the proportional flow valve is controlled to return to zero flow or minimum safe flow, the clamp remains locked, and the sampling anomaly is marked; until the sampling data is continuously restored to validity and the dual-channel timestamps are resynchronized, the identification of the dynamic loading start point is retried.

[0016] Furthermore, the extraction of the flow rate auto-power spectrum, pressure difference auto-power spectrum, and cross-power spectrum includes: removing the DC mean from the flow rate subsequence and pressure difference subsequence of each local analysis window and applying a Hanning window, and then estimating the above power spectra using Welch spectrum; the pump pulsation frequency band covers the main pulsation frequency and its second harmonic; and the frequency points where the flow rate spectrum power is higher than the static noise floor are identified as frequency points with effective flow excitation.

[0017] Compared to the simple transformation of the original time-domain signal, this invention removes the DC mean from the subsequence and applies a Hanning window before using Welch spectrum estimation to obtain each power spectrum. This effectively suppresses spectral leakage and estimation variance, ensuring that the effective excitation frequency band covers the low-frequency transition band and the pump pulsation frequency band containing the second harmonic. This fully utilizes the effective information in the dynamic excitation, and uses the flow spectrum power being higher than the static noise floor as the criterion for effective frequency points to exclude interference without excitation. This enhances the adaptability to different signal characteristics and ensures the reliability of frequency domain analysis.

[0018] Furthermore, the complex impedance is calculated for the first... A window at frequency The procedure is as follows: ; In the formula, For the first A local analysis window at frequency The corrected complex impedance, The frequency within the effective excitation frequency band; For the first The cross-power spectrum of pressure difference versus flow rate within a window. For the first The flow rate of each window is determined by the power spectrum, and each of the above power spectra has been corrected for the fixed electrical delay. For the first Average traffic per window To be based on the average flow rate The complex impedance value obtained by interpolating the baseline of the residual fixture; and the complex impedance value is taken. The real part is used as the resistance component, and the imaginary part is used as the dynamic hysteresis component.

[0019] Furthermore, the coherence-based arbitration of reliable frequency points includes: calculating the squared coherence of the flow-pressure difference amplitude at each frequency point for each window, where the squared amplitude coherence is the square of the modulus of the cross-power spectrum divided by the product of the pressure difference auto-power spectrum and the flow auto-power spectrum; when a frequency point simultaneously satisfies the following conditions: the flow spectrum power is higher than the static noise floor, the squared amplitude coherence is not lower than a set threshold, the drag component is positive, and the change in the corresponding drag component of adjacent windows does not exceed the continuity threshold, it is received as a reliable frequency point; and using coherence and flow spectrum power as weights, the drag components of multiple reliable frequency points within the window are weighted and averaged to obtain the local incremental drag.

[0020] Furthermore, reconstructing the steady-state pressure difference-flow rate relationship and converting it into a compensating secant ventilation resistance includes: assigning the local incremental resistance of each window to a set number of flow rate buckets according to their average flow rate, and weighting and summing them into a compensating incremental resistance for each bucket; based on the carbon canister steady-state pressure difference-flow rate constitutive relationship... and its derivative with respect to flow rate Using the incremental resistance of each barrel as a sample, the non-negativity coefficients are inversely calculated using weighted least squares with non-negativity constraints. and ,in Characterizing viscous laminar flow resistance, Characterize inertial or local turbulent drag; then integrate the incremental drag curve from zero flow to the target flow to obtain the compensated steady-state pressure drop, and convert the compensated secant ventilation drag according to the following formula: ;in For target traffic.

[0021] Furthermore, the compensation cleavage ventilation resistance, compensation steady-state pressure drop, difference before and after compensation, main pulsation frequency, average coherence, proportion of dynamic hysteresis component, and compensation confidence are uploaded via the communication interface; when the PLC or MES does not return confirmation within the set time limit or the result exceeds the upper limit of the specification project, the result is written to the local traceability cache and the retest or manual confirmation status is output, and the qualified judgment is not released.

[0022] The technical effects of this invention are as follows: This invention transforms the carbon canister ventilation resistance test from a traditional static measurement that takes the end-point average after a steady-state wait to a frequency-domain complex impedance identification and compensation method based on dynamic flow characteristics. Its key technical concept utilizes the inverse relationship between flow transition and pump pulsation as a natural excitation signal. This is achieved by separating the fixed electrical delay and residual fixture baseline, decomposing the complex impedance into real and imaginary parts to distinguish between the true ventilation resistance and the dynamic hysteresis of the cavity energy storage, arbitrating reliable frequency points based on the coherence of flow and pressure difference and weighting them, and then calculating the secant ventilation resistance by inversely converting coefficients based on a physical constitutive model. This eliminates system errors and dynamic hysteresis contamination without waiting for a steady state, yielding results consistent with existing specifications. Its main application value lies in significantly shortening the testing cycle, improving measurement repeatability and accuracy, and enabling closed-loop traceability of production line quality via a communication interface. Attached Figure Description

[0023] Figure 1 This is a flowchart of the carbon canister ventilation resistance test data compensation method based on dynamic flow characteristics according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the dynamic analysis segmentation of the traffic loading event index according to an embodiment of the present invention. Detailed Implementation

[0024] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0025] Example of a method for compensating carbon canister ventilation resistance test data based on dynamic flow characteristics: like Figure 1 As shown, the carbon canister ventilation resistance test data compensation method based on dynamic flow characteristics of the present invention includes: S1. Acquire the dynamic baseline of the fixture link and separate the fixed electrical delay.

[0026] Before batch testing, it is necessary to obtain the dynamic response introduced by the test fixture, upstream and downstream pipelines, pressure taps, sensor links, and data acquisition links. Otherwise, the differential pressure changes in the formal test will be mixed with the fixture link response, making it difficult to determine whether the phase lag originates from the carbon canister under test. Therefore, the data processor acquires the dynamic baseline under the same sampling frequency, loading method, sensor installation, and fixture locking conditions as in the formal test.

[0027] Dynamic baseline This is complex lookup table data that characterizes only the residual dynamic impedance and phase response introduced by the fixture interface, upstream and downstream pipelines, pressure taps, sensor links, and acquisition links, excluding the steady-state ventilation resistance, activated carbon pore resistance, and cavity effect of the carbon canister under test. In this embodiment, a standard straight-through component with the same sealing port and installation posture as the carbon canister but without an activated carbon cavity is preferred. Alternatively, a standard throttling component whose impedance has been pre-calibrated on the metering bench can also be installed. The data processor uses the flow rate sequence and differential pressure sequence synchronously obtained during baseline acquisition to calculate the complex response of the differential pressure relative to the flow rate excitation at various frequencies and near each average flow rate. ; ; In the formula The cross-power spectrum of pressure difference and flow rate Divide by the flow-power spectrum The total complex impedance obtained, in units ; Frequency, unit ; Average window flow, in units ; The complex impedance of the standard component is known. This represents the baseline of the residual dynamic impedance of the fixture link after deducting the standard components. For low-resistance through-hole components, their impedance relative to the carbon canister test impedance is less than... Press when deal with.

[0028] To avoid performing fixed delay correction later, and in By repeatedly subtracting the same phase, this embodiment fixes the electrical transmission delay between the flow channel and the differential pressure channel, which can be stably identified from the baseline. Extract separately and with Separate storage. Fixed electrical delays exhibit a phase slope that is linear with frequency and essentially unaffected by average flow rate, while nonlinear or flow-dependent phases generated by fixture cavities, pressure taps, etc., are retained in the... During extraction, the data processor operates at a coherence level of no less than [value missing]. The flow spectrum power is higher than the static noise floor. A linear fit is made to the expanded phase at the effective frequency points: ; In the formula For the first The average flow interval, the first Phase expansion at each frequency point Take the product of the coherence at that frequency and the spectral power of the flux. , This is the corresponding weighted average. Candidate delay for each flow interval. By weight Weighted average fixed electrical delay When each The range and the deviation after retesting with replaced standard parts do not exceed [a certain value]. When accept It can also be used for the baseline of standard through parts. and Perform a cross-correlation peak parabolic interpolation check; if the two values ​​match, record them in the baseline table. The corresponding number of sampling points is... The integer part is achieved by shifting the sampling points, and the fractional part is achieved by decimal delay using FIR. After separating the fixed delay, the residual fixture baseline for formal storage is: ; in The original total complex impedance without delay correction. The time degenerates to the previous formula. If the fitting cannot stably distinguish between the fixed delay and the link complex impedance phase, and the phase error it causes in the effective frequency band exceeds... If the baseline is not used for dynamic compensation, the system will re-acquire the baseline or revert to the original steady-state process.

[0029] During baseline acquisition, the proportional flow valve was tested according to the formal testing procedure. Step or ramp loading to target flow rate Retain the air pump to Pulsation can also be superimposed with an amplitude value equal to the target flow rate. to Small amplitude modulation; at least one sample per flow range Each pulse cycle, frequency domain storage coverage to . Stored in a complex number table, each entry containing frequency, average flow rate, real part, and imaginary part. Temperature, pressure, and timestamps are used, interpolated based on the current window's average flow rate and frequency during formal testing. After baseline acquisition, the throttling device is installed and verified: This is done first using the formal testing method. Correction and Subtraction If the impedance of the verification component deviates from the measured value in the real part of the effective frequency band by more than [a certain amount], Or phase deviation exceeds The baseline is deemed invalid, and the proportional flow valve remains in place. The device is in a safe state and should not be used for carbon canister determination until it is re-collected and re-verified.

[0030] S2. Identify traffic loading events and generate dynamic analysis segments.

[0031] After baseline acquisition, the carbon canister to be tested is installed. The data processor reads the flow rate, differential pressure, and temperature and pressure sampling data, and converts the flow rate into standard state volumetric flow rate according to the calibration parameters. If the current method is still used to wait for a fixed steady-state time before taking the average terminal differential pressure, the pulsation and step transition information at the initial loading stage will be discarded, and the remaining average value may still be affected by the initial phase of pulsation. Therefore, this step constructs synchronous data segments using flow rate change events as indices.

[0032] The data processor receives a one-dimensional time-domain flow sequence. (unit ) and pressure difference sequence (unit After confirming that the clamp is locked and the seal is effective, record the target flow rate setting and timestamp; when continuous to The rate of change of internal flow exceeds the rate of change of static noise. Double or more than the target traffic At that time, record that moment as the starting point of dynamic loading, and extract the data before that point. to After the starting point to The synchronized data forms a dynamic test segment. The periodic pulsations of a piston pump or diaphragm pump are identified using autocorrelation or short-time spectral tracking to determine the main pulsation frequency (typically...). to The fragments were then divided into overlapping segments. Local analysis windows, each window should cover at least One pulse cycle; in this embodiment, the window length is preferably selected. Point (sampling) Main pulsation (Preferred values ​​below), when the time requirement is higher or the main pulsation is higher, can be used. Click to When adjusting the pump speed, the window boundaries are readjusted according to the current window's main pulsation frequency, and this adjustment does not change the core feature of synchronizing data based on flow dynamic events.

[0033] Specifically, when the sensor continuously One sampling period not updated, sampling exceeded range The above, or the dual-channel timestamp deviation exceeds During each sampling period, the data processor stops generating the current segment, and the proportional flow valve returns. Alternatively, maintain minimum safe flow rate, keep the clamp locked, and mark sampling anomalies until data is continuous. It can only be retried after the recovery is effective and the timestamp is synchronized. As a result, the originally discarded traffic transition segment is converted into a dynamic analysis segment with timestamps, main pulsation frequency, and average traffic.

[0034] like Figure 2 As shown in the diagram, this schematic diagram uses a synchronous flow sequence. and pressure difference sequence For each object, the segment range is marked with the starting point of dynamic loading, 100ms to 200ms before the starting point, and 0.8s to 2.5s after the starting point, as well as local analysis windows with 50% overlap and 512 points or 256 points to 512 points, to reflect the effect of the main pulsation frequency on the adjustment of the window boundaries.

[0035] S3. Extract the same source frequency domain features to obtain the effective excitation frequency band.

[0036] The flow channel in the dynamic window contains pump pulsation, valve transitions, and random noise, while the differential pressure channel simultaneously contains the actual ventilation resistance response, the cavity filling and discharging response, and sensor and environmental disturbances. Directly averaging the differential pressure cannot determine whether its changes are driven by stable flow. Therefore, this step uses the flow channel as a reference to extract the same source frequency domain features.

[0037] The data processor takes the DC mean of the flow rate subsequence and the pressure difference subsequence for each window (to focus the calculation on the dynamic changes near the current average flow rate), applies a Hanning window, and then estimates the flow rate power spectrum using the Welch spectrum. Pressure differential power spectrum With cross-power spectrum This processing does not involve applying a fixed-cutoff low-pass filter to the pressure difference, but rather searches for components in the pressure difference that originate from the flow loading in the frequency domain. Frequency domain analysis focuses on two types of frequency bands: the low-frequency transition band formed by the flow step transition (typically...). to ) and pump pulsation frequency band (main pulsation frequency and its (Second harmonics). When the power of the flow spectrum at a certain frequency is higher than the static noise floor... It is determined that it has effective flow excitation; power frequency interference, sensor mechanical resonance, and frequencies without effective excitation are not included in the compensation. This step outputs the values ​​in each window. , , The effective excitation frequency band set, window average flow, and timestamp are used. In other embodiments, Welch spectrum estimation can be replaced by short-time Fourier, multi-window spectrum estimation, or cross-spectral methods that meet the same input and output requirements.

[0038] S4. Complex impedance decomposition to distinguish between drag components and hysteresis components.

[0039] In actual testing, the pressure difference includes not only the in-phase pressure drop caused by ventilation resistance, but also the phase lag component caused by the filling and discharging of the pipeline cavity, gas compression, and sensor hysteresis. If these two are not distinguished, the dynamic energy storage response is easily included in the steady-state ventilation resistance when the fixed waiting time is insufficient. Therefore, this step uses the aerodynamic-electrical equivalence concept to represent the pressure difference as voltage and the flow rate as current, and expresses the dynamic response within the local flow range of the carbon canister as complex aerodynamic impedance.

[0040] Before decomposition, the data processor calls the fixed electrical delay stored in S1. With residual baseline .like If valid, then only execute. Secondary delay correction, that is, in the time domain according to Apply an integer shift and fractional delay (FIR) to the pressure differential or flow rate sequence, or multiply the original transfer function by a phase factor in the frequency domain. Both are equivalent and will not be executed repeatedly; if S1 has already been judged... If negligible, no further correction is needed. Delayed correction and The subtraction functions differ: the former only corrects the dual-channel pure time shift once, while the latter only subtracts the fixture link complex response that still exists after correction, thus avoiding repeated subtraction of the same phase. Subsequently, the... A window at frequency Calculation at point: ; ; ; In the formula , The cross-power spectrum and flow auto-power spectrum have been completed with delay correction; To average flow rate by window With frequency The residual baseline of the interpolation; The resistance component is in phase with the flow rate, corresponding to the dissipative response that can take into account the ventilation resistance. This is a dynamic hysteresis component that is approximately orthogonal to the flow rate, primarily characterizing the hysteresis between the cavity filling / discharging and the sensor, and is not directly used as the steady-state ventilation resistance output. Since the DC average has been removed from S3, Reflecting the small-signal incremental resistance near the current average flow rate is not equivalent to the resistance from... arrive The secant air resistance. Any frequency domain correlation technique that can obtain phase-distinguishable resistance components from synchronous dual-channel data can be used in this step.

[0041] S5, dual-channel consistency arbitration to obtain local incremental resistance.

[0042] The resistance component output in step S4 may still be affected by valve vibration, power ripple, fixture collision, or random pressure disturbances; if compensation is based solely on the differential pressure amplitude or single-frequency resistance component, external disturbances may be misinterpreted as changes in the carbon canister resistance. Therefore, this step arbitrates candidate resistance components based on the consistency between the flow channel and the differential pressure channel, accepting only the differential pressure component driven by stable flow excitation. The data processor calculates the... A window at frequency Coherence of flow rate versus squared pressure difference amplitude at the location: ; Its value to This is used to determine the proportion of the pressure difference spectrum energy at the same frequency that can be stably explained by the flow spectrum, which is equivalent to establishing a flow reference gate in pressure difference processing: this component only enters the drag compensation when the pressure difference and flow rate are stably synchronized in frequency and phase relationship. In this embodiment, the flow spectrum power is simultaneously satisfied to be higher than the static noise floor. , Not less than Preferred , For the positive and adjacent window corresponding resistance component changes not to exceed the continuity threshold, this embodiment can take... to The frequency points are received as trusted frequency points. The data processor uses coherence and flow spectral power as weights to process multiple trusted frequency points within the window. The weighted average yields the local incremental resistance of the window. (scalar, unit) ).

[0043] When the number of reliable frequency points within the window is less than The coherence of one or more major frequency bands is at to When the boundary is reached, the window is marked as a boundary window and extended sampling is triggered. to Alternatively, a small dynamic loading can be applied, during which the proportional flow valve maintains a safe flow rate, only recording traceability data without outputting a pass / fail judgment. The pump pulsation amplitude is lower than the target flow rate. Furthermore, when the effective frequency points are insufficient for a long period, the fixed coherence threshold can be replaced with high confidence, boundary confidence, or unconfidentiality. The discrete criterion is used, with only the high-confidence window directly participating in the output. This eliminates isolated differential pressure noise, asynchronous impulses, and electromagnetic interference. In addition.

[0044] S6. Reconstruct the steady-state pressure drop from the local incremental resistance and output the compensation ventilation resistance.

[0045] These are the local incremental resistances of each window near its average flow rate, while production line determination typically requires a target flow rate. The steady-state pressure drop or secant ventilation resistance. Directly multiplying the local incremental resistance by the target flow rate will misinterpret the small-signal slope as the flow rate from... arrive The secant resistance can cause systematic deviations in medium-to-high flow rates or large-capacity carbon canisters. Therefore, this step reconstructs the incremental resistance sample into a steady-state pressure difference-flow relationship before outputting it.

[0046] The data processor will process each trusted window's... Its average flow Correspondingly, based on target traffic Will to Divided into a set number of flow buckets, in this embodiment the bucket width , can to (Adjustment). For the central flow rate... The set of trusted windows is The Buckets, by window weight Summary: ; ; in To compensate for the increased resistance of the barrel, For bucket weights, Average coherence of effective frequency points in the window The product of the normalized flow spectrum power and the normalized flow spectrum power results in a higher contribution from windows with strong excitation and good synchronization. The steady-state pressure difference-flow constitutive relationship of the carbon tank is as follows: ; In the formula For traffic steady-state pressure difference (unit) ), , These are the non-negative coefficients for the viscous laminar flow drag term and the inertial or local turbulent flow drag term, respectively. At low flow rates, the viscous term dominates, while the inertial or local turbulent flow term gradually increases with increasing flow rate, thus covering the nonlinear drag characteristics at medium to high flow rates. Because... , These are all small-signal incremental resistances, corresponding to the local slope of the constitutive relation with respect to flow rate rather than the secant resistance. Therefore, for... Taking the derivative yields the regression relationship: ; Data processor bucket samples Inverse calculation using nonnegativity-constrained weighted least squares , : ; In the formula The set of effective traffic buckets to satisfy the requirement of a reliable window size. Since the noise tends to be negative, it is limited by the non-negativity constraint. And it degenerates into a linear relationship. Determine. , Then, the incremental resistance curve from Points to To compensate for the steady-state pressure drop, we then calculate the secant resistance to compensate for the pressure drop. ; ; The above formula clearly distinguishes between local incremental resistance. With secant air resistance This avoids directly using the dynamic slope as the final result. In this embodiment, we use... Taking a small cavity sample as an example, the measurements were... The effective flow bucket samples are shown in Table 1 below:

[0047] Substitution Nonnegative constrained weighted least squares yields , ,but ,and , It can be seen that even if the dynamic incremental resistance near the target flow rate is The output judgment value after reconstruction is still the same. Therefore, it cannot be directly used as or Output secant ventilation resistance.

[0048] If the effective flow bucket is less than If the coverage area is insufficient to support the inverse calculation of coefficients, a high-confidence compensation value is not directly output. Instead, supplementary dynamic loading, extended sampling, or regression to the original steady-state average value is triggered, and the confidence level is marked as insufficient. Alternatively, the quadratic coefficients may not be inversely calculated. Instead, monotonically interpolating the incremental resistance of each bucket and accumulating it using the trapezoidal surfaces of adjacent buckets directly yields the compensation value. to The steady-state pressure drop is then converted into secant resistance, suitable for scenarios with many specifications and where it is inconvenient to uniformly preset the quadratic form. When testing a single flow point, the average flow rate falling into the range is selected. of or The conversion is completed by combining the window with multiple flow buckets in the transition section; during multi-flow point testing, the compensation incremental resistance is calculated separately for each flow bucket to generate the pressure difference-flow compensation curve.

[0049] The final data processor transmits data to the PLC or MES via industrial Ethernet, RS485, or CAN bus, including compensated steady-state voltage drop, compensated secant ventilation resistance, difference before and after compensation, main pulsation frequency, average coherence, proportion of dynamic hysteresis components, and coefficients. and and compensation confidence level; when PLC or MES is in If no confirmation is returned or the result exceeds the upper limit of the specification project, the result is written to the local traceability cache and the retest or manual confirmation status is output without releasing the pass / fail judgment.

Claims

1. A method for compensating for carbon canister ventilation resistance test data based on dynamic flow characteristics, characterized in that, The method includes: acquiring the flow rate sequence and differential pressure sequence of the carbon canister under test; S1, acquiring the dynamic baseline of the fixture link, linearly fitting the expanded phase of the coherent effective frequency point, identifying the phase slope that is linear with frequency and does not change with average flow rate as a fixed electrical delay, and using the flow-related phase generated by the fixture cavity and pressure tapping pipe as the residual fixture baseline, storing the two separately; S2, taking the moment when the flow rate change exceeds a set threshold as the starting point of dynamic loading, extracting the synchronous flow rate and differential pressure sequences before and after it to form a dynamic test segment, dividing it into overlapping local analysis windows that each cover a complete pulsation cycle; S3, extracting the flow self-power spectrum and differential pressure self-power spectrum using the flow channel as a reference. S4. After performing a fixed electrical delay correction only once for each window, calculate the complex impedance, subtract the residual fixture baseline interpolated according to the average flow rate of the window, take the real part as the resistance component in phase with the flow rate, and take the imaginary part as the dynamic hysteresis component approximately orthogonal to the flow rate; S5. Arbitrate the reliable frequency point with the coherence of the flow rate and pressure difference channel, and obtain the local incremental resistance of each window by weighting the coherence and flow rate spectrum power; S6. Reconstruct the local incremental resistance of each window into a steady-state pressure difference-flow rate relationship according to the average flow rate, inversely calculate the constitutive coefficient and convert it into the compensation secant ventilation resistance, and then upload it through the communication interface.

2. The method for compensating carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 1, characterized in that, The dynamic baseline of the acquisition fixture link includes: Install a standard straight-through component on the fixture that has the same sealing port and installation posture as the carbon canister but does not contain an activated carbon cavity, or install a standard throttling component whose own impedance has been pre-calibrated by the metering bench; The total complex impedance is obtained by dividing the differential pressure and flow rate cross power spectrum obtained synchronously in the baseline acquisition by the flow rate autopower spectrum. The known complex impedance of the standard straight-through component or standard throttling component is then subtracted to obtain the residual fixture baseline.

3. The method for compensating carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 1, characterized in that, Identifying phase slope as a fixed electrical delay also includes: The linear fitting is performed at frequency points where both coherence and flow spectrum power meet the effective conditions, and the fixed electrical delay is accepted when the range of candidate delays and retest deviations in each flow interval do not exceed the preset tolerance.

4. The method for compensating carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 1, characterized in that, Extracting and forming dynamic test segments also includes: The collected flow rate is converted into standard state volumetric flow rate according to the calibration parameters; The dynamic loading start point is the moment when the rate of change of flow exceeds a set multiple of the rate of change of static noise or a set ratio of the target flow within a continuously preset time period. The main pulsation frequency corresponding to the pulsation period is identified by autocorrelation or short-time spectrum tracking.

5. The method for compensating for carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 4, characterized in that, It also includes anomaly monitoring: When the flow sensor or differential pressure sensor fails to update for a set number of consecutive sampling periods, exceeds the sampling range, or the dual-channel timestamp deviation exceeds the set number of sampling periods, the generation of the current dynamic test segment is stopped, and the proportional flow valve is controlled to return to zero flow or minimum safe flow, the clamp remains locked, and the sampling abnormality is marked. The identification of the dynamic loading start point will be triggered again once the sampled data is continuously restored to validity and the dual-channel timestamps are resynchronized.

6. The method for compensating carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 1, characterized in that, The extracted flow rate autopower spectrum, pressure difference autopower spectrum, and cross-power spectrum include: The DC mean was removed from the flow rate subsequence and pressure difference subsequence of each local analysis window, and a Hanning window was applied. The power spectra were then estimated using Welch spectrum. The pump pulsation frequency band covers the main pulsation frequency and its second harmonic; Frequency points with flow spectrum power higher than the static noise floor are identified as having effective flow excitation.

7. The method for compensating for carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 1, characterized in that, Calculate the complex impedance for the first A window at frequency The procedure is as follows: ； In the formula, For the first A local analysis window at frequency The corrected complex impedance, The frequency within the effective excitation frequency band; For the first The cross-power spectrum of pressure difference versus flow rate within a window. For the first The flow rate of each window is determined by the power spectrum, and each of the above power spectra has been corrected for the fixed electrical delay. For the first Average traffic per window To be based on the average flow rate The complex impedance value obtained by interpolating the baseline of the residual fixture; and the complex impedance value is taken. The real part is used as the resistance component, and the imaginary part is used as the dynamic hysteresis component.

8. The method for compensating carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 7, characterized in that, The reliable frequency points for coherence arbitration include: Calculate the squared coherence of the flow-pressure difference amplitude at each frequency point for each window. The squared coherence is the square of the modulus of the cross power spectrum divided by the product of the pressure difference power spectrum and the flow rate power spectrum. When a frequency point simultaneously satisfies the following conditions: the power of the flow spectrum is higher than the static noise floor, the square coherence of the amplitude is not lower than a set threshold, the drag component is positive, and the change of the drag component corresponding to the adjacent window does not exceed the continuity threshold, it is received as a reliable frequency point. The local incremental drag is obtained by weighting the drag components at multiple reliable frequency points within the window with coherence and flow spectral power as weights.

9. The method for compensating carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 8, characterized in that, Reconstructing the steady-state pressure difference-flow rate relationship and converting it to compensated secant ventilation resistance includes: The local incremental resistance of each window is assigned to a set number of flow buckets according to its average flow rate, and the weighted sum is used to compensate for the incremental resistance of each bucket. Based on the constitutive relationship of steady-state pressure difference-flow rate in carbon canisters and its derivative with respect to flow rate Using the incremental resistance of each barrel as a sample, the non-negativity coefficients are inversely calculated using weighted least squares with non-negativity constraints. and ,in Characterizing viscous laminar flow resistance, Characterizing inertial or local turbulent drag, For traffic steady-state pressure difference, in units of ; The incremental resistance curve is then integrated from zero flow to the target flow to obtain the compensated steady-state pressure drop, and the compensated secant ventilation resistance is calculated using the following formula: ;in For target traffic, To compensate for the air resistance of the cleavage.

10. The method for compensating carbon canister ventilation resistance test data based on dynamic flow characteristics according to claim 1, characterized in that, The compensation secant ventilation resistance, compensation steady-state pressure drop, difference before and after compensation, main pulsation frequency, average coherence, proportion of dynamic hysteresis component, and compensation confidence level are uploaded via the communication interface. When the PLC or MES fails to return confirmation within the set time limit or the result exceeds the upper limit of the specification project, the result is written to the local traceability cache and the retest or manual confirmation status is output, and the pass / fail judgment is not released.

Images