An intelligent pulse ultrasonic fluid measurement method and system fusing multi-sensor collaborative collection

Abstract

The application provides a kind of intelligent pulse ultrasonic fluid measurement method and system of fusion multi-sensor cooperative collection, and relates to intelligent pulse ultrasonic fluid measurement technical field.The application realizes cross-channel synchronous acquisition and quality marking by configuring ultrasonic and non-ultrasonic channel and establishing unified time base;Based on cyclic stationary and coherent joint analysis, vortex characteristic frequency is extracted and uncertainty index is generated;Strouhal number is online corrected by combining medium parameters and flow state criterion, and installation geometry and convection effect are compensated, to obtain corrected flow velocity estimation and its uncertainty;In time window, time series fusion is performed with uncertainty as weight, to obtain fused flow velocity and uncertainty;Then, volume flow and its uncertainty are calculated by combining pipe diameter, to realize full-link optimized measurement under complex working conditions, and to improve precision, range ratio and traceability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent pulse ultrasonic fluid measurement, in particular to an intelligent pulse ultrasonic fluid measurement method and system fusing multi-sensor cooperative collection. BACKGROUND

[0002] Ultrasonic flow measurement mainly includes Doppler method, time-of-flight (ToF) method and vortex street method. For the common solid-containing / gas-containing and pulsating working conditions in actual industrial scenes, pulse ultrasonic is usually used as a carrier in engineering, which is modulated by flow downstream of a vortex generator, and then the frequency characteristics related to vortex shedding are extracted through amplification, frequency selection, synchronous rectification (synchronous demodulation) and low-pass filtering system, and the flow rate and flow are converted by combining the Strouhal number (St) and geometric parameters. Under the conditions of homogeneous medium and stable working condition, this route has the advantages of simple implementation and convenient maintenance; but under the conditions of low signal-to-noise ratio (SNR), pipe wall vibration, particle / gas bubble scattering and obvious change of medium physical properties with temperature, frequency doubling, sidelobe and aliasing are easily generated in the signal, which leads to unstable vortex frequency identification, limited turndown ratio (TDR) and decreased cross-condition consistency.

[0003] With the expansion of the measurement object from single-phase steady state to multiphase flow (MPF), strong disturbance and periodic pulsating working condition, the instrument is evolving towards multi-source information fusion (DF) and time reference unification: on the one hand, by introducing pressure pulsation, pipe wall vibration, temperature and conductivity as auxiliary channels outside the ultrasonic link, and realizing clock / time stamp alignment of all channels by phase-locked loop (PLL); on the other hand, using cyclostationary analysis (CSA), cross power spectral density (CPSD) and coherence function (coherence function) and other cross-channel coherence criteria to improve the robust extraction ability of the main peak of vortex shedding. At the same time, based on the Reynolds number (Re) and the on-site disturbance index, the Strouhal number is adaptively corrected online, and the extended Kalman filter (EKF) or particle filter (PF) and other time series estimation methods are introduced, which has become a consensus direction to improve the cross-condition stability and realize the output of traceable measurement uncertainty (MU).

[0004] There are still a large number of schemes relying on a single ultrasonic receiving link and a fixed Strouhal number, and lacking online compensation for the 'vortex frequency-flow velocity' deviation caused by the change of medium viscosity / sound velocity with temperature and composition; in the strong scattering and mechanical noise background, the traditional 'frequency selection-rectification-low pass' serial process is insufficient to suppress the frequency doubling, sidelobe and aliasing, and the cross-channel information is not effectively utilized, resulting in weak coherence discrimination and easy drift of peak value under low SNR; at the same time, lacking adaptive switching strategy of estimator and closed-loop calibration mechanism of unified time base linked with working condition identification, it is difficult to balance the range ratio and uncertainty consistency, and the consistency of on-site reproducibility and bus / pulse output is also difficult to maintain for a long time. SUMMARY

[0005] In order to overcome the shortcomings of the prior art, the purpose of the present application is to provide an intelligent pulse ultrasonic fluid measurement method and system fusing multi-sensor collaborative acquisition, which realizes the full-link optimization from signal acquisition to flow output under complex working conditions through multi-sensor collaborative acquisition, cyclic stationary and coherent joint extraction, online Strouhal number correction, geometric and convection compensation, time sequence fusion weighting and uncertainty propagation, overcomes the systematic deviation, insufficient anti-interference and uncertainty of the prior art, and significantly improves the measurement accuracy, range ratio and traceability.

[0006] To achieve the above purpose, the present application provides the following scheme:

[0007] An intelligent pulse ultrasonic fluid measurement method fusing multi-sensor collaborative acquisition, comprising:

[0008] Establishing a unified time base to align the acquisition time of the configured ultrasonic measurement channel and non-ultrasonic measurement channel, and registering the installation geometric parameters; the ultrasonic measurement channel includes a transmitting transducer and a receiving transducer, and the non-ultrasonic measurement channel includes at least two types of sensors for representing working conditions;

[0009] Under the unified time base, the transmitting transducer is driven to emit ultrasonic pulses according to a preset pulse sequence, the receiving transducer receives echoes to obtain flow-modulated ultrasonic representations, the non-ultrasonic measurement channel data is collected, and cross-channel alignment and quality labeling are completed;

[0010] Based on the joint extraction model of cyclic stationary characteristics and cross-channel coherence, the ultrasonic representations and the non-ultrasonic measurement channel data are jointly analyzed to determine the characteristic frequency corresponding to vortex shedding, and generate the frequency uncertainty and coherence consistency index corresponding to the characteristic frequency;

[0011] The medium parameters and flow state criteria obtained from the non-ultrasonic measurement channel data are used to correct the Strouhal number online, and systematic deviations caused by installation geometry and convection effects are compensated to obtain a corrected flow rate real-time estimate, and a flow rate real-time uncertainty is synthesized according to the frequency uncertainty;

[0012] The flow rate real-time estimate is time-series fused with the flow rate real-time uncertainty as a weight within a given time window to obtain a fused flow rate estimate and a fused flow rate uncertainty;

[0013] The cross-sectional area is calculated based on the pipe inner diameter, and the volume flow rate and volume flow rate uncertainty are obtained based on the fused flow rate estimate; the volume flow rate uncertainty is obtained by propagating the fused flow rate uncertainty and the geometric size uncertainty.

[0014] Preferably, the installation geometry parameters include the center distance of the transmitting transducer and the receiving transducer, the incident angle or the included angle, the characteristic width of the flow blocking member, and the pipe inner diameter.

[0015] Preferably, it further comprises:

[0016] When the coherence consistency index does not meet the threshold value, a robust estimation strategy is switched to before time-series fusion is performed to limit the weight and suppress outliers.

[0017] Preferably, a unified time base is established to align the acquisition times of the configured ultrasonic measurement channels and non-ultrasonic measurement channels, and installation geometry parameters are recorded, including:

[0018] A phase-locked time reference is provided to generate a reference clock and a common trigger signal, and the reference clock and the common trigger signal are distributed to the acquisition ends of the ultrasonic measurement channels and the non-ultrasonic measurement channels;

[0019] A unified sampling rate, timestamp format, and trigger rule are set at each acquisition end to complete timestamp alignment verification and record the alignment results;

[0020] A reference pulse is emitted by the transmitting transducer under no-flow conditions, and each channel synchronously records a reference response, and a fixed delay of each channel relative to the reference is calculated based on a time correlation alignment method to generate a delay compensation table;

[0021] The reference response is repeatedly acquired within a set time period, the time difference fluctuation and drift amount are counted, the resynchronization period and drift compensation amount are set, and the drift compensation amount is deducted online during the acquisition process;

[0022] The transmitting transducer and the receiving transducer are installed and fixed using a positioning jig, the center distance of the transmitting transducer and the receiving transducer, the incident angle or the included angle, the characteristic width of the flow blocking member, and the pipe inner diameter are measured and recorded, and corresponding tolerance ranges are given.

[0023] Preferably, under a unified time base, the transmitting transducer is driven to emit ultrasonic pulses according to a preset pulse sequence, and the receiving transducer receives the echo to obtain the flow-modulated ultrasonic characterization. Simultaneously, non-ultrasonic measurement channel data is acquired, and cross-channel alignment and quality marking are completed, including:

[0024] A trigger pulse signal is generated based on the unified time base and sent to the drive end of the transmitting transducer, so that the transmitting transducer emits ultrasonic pulses according to a preset pulse sequence;

[0025] The echo signal is acquired by the receiving transducer under the unified time base to obtain the raw ultrasonic data containing the fluid modulation effect;

[0026] During each pulse sequence emitted by the transmitting transducer, the non-ultrasonic measurement channel is simultaneously triggered to start sampling using the unified time base, thereby obtaining temperature data, pressure data, vibration data, and conductivity data.

[0027] The raw ultrasound data and the non-ultrasound measurement channel data are aligned with a unified timestamp to form a cross-channel synchronous dataset;

[0028] The quality of the cross-channel synchronous dataset is evaluated, including amplitude saturation detection, noise level detection, and missing segment detection, and quality tags are added to unqualified data segments.

[0029] Preferably, based on a joint extraction model of cyclic stationarity and cross-channel coherence, the ultrasonic characterization and the non-ultrasonic measurement channel data are jointly analyzed to determine the characteristic frequencies corresponding to vortex shedding, and frequency uncertainty and coherence consistency indices corresponding to the characteristic frequencies are generated, including:

[0030] The ultrasonic characterization is subjected to Hilbert transform to obtain an analytical signal, and the ultrasonic envelope sequence is extracted. The ultrasonic envelope sequence and the non-ultrasonic measurement channel data are divided into time slices of the same length according to the unified time base. The time slices are spliced ​​and verified with a fixed overlap ratio and an aligned set of segments is output.

[0031] For each time slice, the second-order cyclic autopower spectrum of the ultrasonic envelope sequence is calculated to obtain a joint spectrum of frequency and cyclic frequency, which is used to characterize the cyclic stationary characteristics caused by vortex shedding. The formula is as follows: , ;in, The ultrasound envelope sequence; The ultrasound envelope in the ultrasound envelope sequence at frequency With cycle frequency Cyclic self-power spectral density at; the cyclic cross-correlation function of the ultrasonic envelope; is the time delay; is the time; is the expectation operator; is the conjugate operation;

[0032] select at least one category from the non-ultrasound measurement channel data as an auxiliary channel, calculate the cross-power spectral density and the amplitude squared coherence function of the ultrasonic envelope and the auxiliary channel in the frequency domain based on the unified time base, and extract the phase consistency quantity, the formula being: , ; wherein, is the sequence of the auxiliary channel; is the cross-power spectral density of the ultrasonic envelope and the auxiliary channel at the frequency ; and are the respective power spectral densities; is the amplitude squared coherence function; is the root mean square deviation of the cross-spectrum phase over the window set; is the number of time slices; is the phase operator; the overline indicates the average in the time slice dimension;

[0033] search for the maximum cyclic frequency on the cyclic spectrum for each frequency and perform robust normalization, construct a weightless joint scoring function and determine the characteristic frequency, the formula being: , , , ; wherein, is the cyclic frequency that makes the cyclic spectrum reach the maximum at the frequency ; is the cyclic spectrum amplitude based on median normalization, used to suppress background noise and sidelobes; is the median operator; is the joint scoring function; is the characteristic frequency corresponding to vortex shedding;

[0034] statistical estimation of the characteristic frequency in the time slice is performed by resampling with the delete-one method, to obtain the frequency uncertainty, the formula being: ; wherein, is the characteristic frequency estimate obtained after the first time slice is removed; is the average of the characteristic frequency estimates of all time slices; is the frequency uncertainty;

[0035] combine the coherence intensity and the phase consistency into a coherence consistency index with a value between zero and one, used to provide triggering and weighting basis for subsequent steps, the formula being: ; wherein, is the coherence consistency index; is the amplitude squared coherence function at the characteristic frequency; is the inter-spectral phase root mean square deviation at the characteristic frequency; is the circular constant.

[0036] Preferably, the medium parameters and flow regime criteria obtained from the non-ultrasonic measurement channel data are used to correct the Strouhal number online and compensate for systematic deviations caused by installation geometry and convection effects, to obtain a corrected flow velocity instantaneous estimate, and the flow velocity instantaneous uncertainty is synthesized according to the frequency uncertainty, including:

[0037] The medium temperature, pressure and conductivity are calculated from the non-ultrasonic measurement channel data to obtain the kinematic viscosity and density, and the Reynolds number and flow regime criteria are determined in combination with the pipe diameter and the calibrated initial value of the previous time window, and the corresponding reference Reynolds number and reference coherence index are selected as the correction reference;

[0038] The reference Strouhal number is adaptively corrected based on the Reynolds number and the coherence consistency index to obtain a corrected Strouhal number for the current time slice, and the formula is: ; wherein, is the corrected Strouhal number; is the reference Strouhal number; is the Reynolds number of the current time slice; is the reference Reynolds number; is the coherence consistency index; is the reference coherence consistency index; is the lower limit constant used to prevent division by zero and excessive amplification;

[0039] The geometric compensation coefficient is calculated according to the registered installation geometry parameters to make the vortex shedding conversion model consistent with the perceived path projection, and the formula is: ; wherein, is the geometric compensation coefficient; is the angle of incidence or the included angle between the acoustic beam formed by the transmitting transducer and the receiving transducer and the pipe axis; is the reference angle at the calibration time;

[0040] The ultrasonic envelope and selected non-ultrasonic channel are narrowband filtered around the characteristic frequency and the cross-correlation is calculated to obtain the convection time delay, the instantaneous estimate based on the frequency model and the convection model is constructed, and the unweighted fusion is used to suppress single-source deviation, and the formula is: , , ; wherein, is the reference flow velocity obtained based on the characteristic frequency and the corrected Strouhal number; is the characteristic frequency; is the characteristic width of the flow blocking member; is the convection flow velocity obtained by the convection time delay inversion; is the axial distance between the ultrasonic sensing position and the auxiliary sensing position; is the convection time delay obtained by the cross-correlation in the characteristic frequency narrow band; is the flow velocity instantaneous estimation;

[0041] The frequency uncertainty, the Strouhal correction uncertainty and the convection delay uncertainty are integrated into the flow velocity instantaneous uncertainty by using the linear uncertainty propagation, and the formula is: ; wherein, is the flow velocity instantaneous uncertainty; is the frequency uncertainty; is the uncertainty corresponding to the Strouhal correction; is the uncertainty of the convection time delay.

[0042] Preferably, the cross-sectional area is calculated based on the pipe inner diameter, and the volume flow rate and the volume flow rate uncertainty are obtained according to the fusion flow rate estimation, including:

[0043] The temperature data in the non-ultrasonic measurement channel data is read under the unified time base, and the current inner diameter is obtained by temperature compensation on the calibrated inner diameter, and the formula is: ; wherein, is the current inner diameter of the pipe under the current temperature; is the calibrated inner diameter under the reference temperature; is the linear expansion coefficient of the pipe material; is the current medium temperature; is the reference temperature;

[0044] The cross-sectional area of the circle is calculated according to the current inner diameter, and the formula is: ; wherein, is the pipe cross-sectional area; is the constant pi;

[0045] The cross-sectional area is multiplied by the fusion flow rate estimation to obtain the volume flow rate, and the formula is: ; wherein, is the volume flow rate; is the fusion flow rate estimation;

[0046] The calibrated inner diameter uncertainty and the size uncertainty caused by temperature are linearly synthesized to obtain the current inner diameter uncertainty, and the formula is: ; wherein, is the current inner diameter uncertainty; is the uncertainty of the calibrated inner diameter; is a linear expansion coefficient; is a calibrated inner diameter; is a temperature measurement uncertainty;

[0047] According to the first-order uncertainty propagation law, the fusion flow rate uncertainty and the inner diameter uncertainty are propagated to the volume flow rate uncertainty, and the formula is: , ; wherein, is a volume flow rate uncertainty; is a fusion flow rate uncertainty;

[0048] The obtained volume flow rate and volume flow rate uncertainty are subjected to threshold verification, and when the preset limit value is exceeded, re-measurement or calibration prompting is triggered, otherwise the volume flow rate, volume flow rate uncertainty and corresponding time stamp are archived and output.

[0049] An intelligent pulse ultrasonic fluid measurement system for fusion multi-sensor cooperative collection, characterized in that it comprises:

[0050] A unified time base and installation geometry registration unit is used to establish a unified time base, align the collection time of the configured ultrasonic measurement channel and non-ultrasonic measurement channel, and register the installation geometry parameters; the ultrasonic measurement channel includes a transmitting transducer and a receiving transducer, and the non-ultrasonic measurement channel includes at least two types of sensors for representing working conditions;

[0051] A cooperative collection and quality marking unit is used to drive the transmitting transducer to emit ultrasonic pulses according to a preset pulse sequence under the unified time base, receive echoes by the receiving transducer to obtain flow-modulated ultrasonic representations, simultaneously collect non-ultrasonic measurement channel data, and complete cross-channel alignment and quality marking;

[0052] A vortex feature extraction and uncertainty generation unit is used to jointly analyze the ultrasonic representations and the non-ultrasonic measurement channel data based on a joint extraction model of cyclostationary characteristics and cross-channel coherence, determine a characteristic frequency corresponding to vortex shedding, and generate a frequency uncertainty corresponding to the characteristic frequency and a coherence consistency index;

[0053] A Strouhal correction and geometry compensation unit is used to correct the Strouhal number online according to medium parameters and flow state criteria obtained from the non-ultrasonic measurement channel data, compensate for systematic deviations caused by installation geometry and convection effects, obtain a corrected flow rate instantaneous estimate, and synthesize a flow rate instantaneous uncertainty according to the frequency uncertainty;

[0054] A time series fusion and flow rate uncertainty unit is used to perform time series fusion on the flow rate instantaneous estimate with the flow rate instantaneous uncertainty as the weight within a given time window, to obtain a fusion flow rate estimate and a fusion flow rate uncertainty.

[0055] A cross-sectional area calculation and volume flow uncertainty unit is used to calculate the cross-sectional area based on the pipe inner diameter, and estimate the volume flow and volume flow uncertainty according to the fusion flow rate; the volume flow uncertainty is obtained by propagation law from the fusion flow rate uncertainty and the geometric size uncertainty.

[0056] The present application discloses the following technical effects:

[0057] The present application solves the problem of insufficient working condition information caused by the prior art relying on only a single ultrasonic signal by introducing ultrasonic measurement channels and at least two types of non-ultrasonic measurement channels and establishing a unified time base to realize cross-channel synchronous acquisition, enabling stable and rich data sources to be obtained under multiphase flow, pulsating flow and disturbance background, laying a foundation for feature extraction.

[0058] The present application adopts a joint model of cyclostationarity and cross-channel coherence in vortex feature extraction, avoids the problem that traditional single-channel filtering is easily disturbed by frequency multiplication, sidelobes and noise, realizes robust identification of vortex shedding characteristic frequency under low signal-to-noise ratio conditions, and can output frequency uncertainty and coherence consistency index, enhancing the feature credibility.

[0059] The present application inverses medium parameters and flow state criteria using non-ultrasonic measurement channel data, and accordingly corrects the Strouhal number online, and at the same time combines installation geometric parameters and convection effect compensation, solving the systematic deviation caused by the fixed Strouhal number and static installation assumption in the traditional method, and improving the accuracy and adaptability of flow rate estimation.

[0060] The present application introduces frequency uncertainty propagation in the flow rate estimation link to form real-time flow rate uncertainty, and uses it as a weighted basis to perform time sequence fusion, which can suppress incidental abnormalities in continuous measurement at multiple times, improving the smoothness and time sequence consistency of flow rate estimation, and expanding the range ratio of the instrument.

[0061] The present application considers temperature compensation and geometric uncertainty of the pipe inner diameter in volume flow calculation, constructs a volume flow uncertainty evaluation method based on propagation law, effectively makes up for the defect that the existing technology lacks a complete uncertainty chain in flow measurement, and guarantees the traceability and reliability of the measurement results.

[0062] The coherence consistency index and uncertainty linkage mechanism provided by the present application enable robust estimation strategies to be triggered when the working condition is abnormal or the signal is distorted, ensuring that the output flow rate and flow results still have stability and consistency under extreme conditions, thereby significantly improving the applicability and engineering value of the system in complex industrial sites. BRIEF DESCRIPTION OF DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings described below only illustrate some of the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0064] Figure 1 The method flowchart provided for the embodiments of the present application is as shown in

[0065] Figure 2 The system structure schematic diagram provided for the embodiments of the present application is as shown in DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0067] The purpose of the present application is to provide an intelligent pulse ultrasonic fluid measurement method and system fusing multi-sensor cooperative collection, which improves the stability and reliability of fluid measurement under complex working conditions.

[0068] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0069] Figure 1 The method flowchart provided for the embodiments of the present application is as shown in Figure 1 The present application provides an intelligent pulse ultrasonic fluid measurement method fusing multi-sensor cooperative collection, which comprises the following steps:

[0070] Step 100: establishing a unified time base, aligning the collection time of the configured ultrasonic measurement channel and non-ultrasonic measurement channel, and registering the installation geometric parameters; the ultrasonic measurement channel comprises a transmitting transducer and a receiving transducer, and the non-ultrasonic measurement channel comprises at least two types of sensors for representing working conditions;

[0071] Step 200: under the unified time base, driving the transmitting transducer to emit ultrasonic pulses according to a preset pulse sequence, receiving the echoes by the receiving transducer to obtain the flow-modulated ultrasonic representation, simultaneously collecting the non-ultrasonic measurement channel data, and completing the cross-channel alignment and quality marking;

[0072] Step 300: Based on the joint extraction model of cyclostationary characteristics and cross-channel coherence, the ultrasonic representation and non-ultrasonic measurement channel data are jointly analyzed to determine the characteristic frequency corresponding to vortex shedding, and the frequency uncertainty and coherence consistency index corresponding to the characteristic frequency are generated;

[0073] Step 400: According to the medium parameters and flow state criteria obtained from the non-ultrasonic measurement channel data, the Strouhal number is corrected online, and the systematic deviation caused by installation geometry and convection effect is compensated to obtain the corrected flow rate real-time estimate, and the flow rate real-time uncertainty is obtained according to the frequency uncertainty synthesis;

[0074] Step 500: In a given time window, the flow rate real-time estimate is weighted by the flow rate real-time uncertainty to obtain the fused flow rate estimate and the fused flow rate uncertainty;

[0075] Step 600: Based on the pipe diameter, the cross-sectional area is calculated, and the volume flow rate and volume flow rate uncertainty are obtained according to the fused flow rate estimate; the volume flow rate uncertainty is obtained by propagating the fused flow rate uncertainty and the geometric size uncertainty.

[0076] Preferably, it further comprises:

[0077] When the coherence consistency index does not meet the threshold value, switch to a robust estimation strategy before performing temporal fusion to limit the weight and suppress outliers.

[0078] Specifically, step 100 of the embodiment provides a phase-locked time reference, generates a reference clock and a common trigger signal, and sends them to the acquisition end of the ultrasonic measurement channel and the non-ultrasonic measurement channel through an active distributor. To ensure the comparability of cross-channel data in time, the embodiment sets consistent sampling rate, timestamp format and trigger rule at each acquisition end; the trigger rule uses the edge mode of "trigger first, sample later" and records the trigger sequence number, which is used for subsequent alignment verification. After the reference clock enters each acquisition end, a sampling clock is generated through the phase-locked process, and the sampling start time is bound to the common trigger signal and written into the timestamp field. To avoid time jitter caused by power disturbance, the embodiment sets shielding and ground isolation in the distribution network, and enables jitter filtering and buffer caching at the acquisition end to complete the establishment of a unified time base.

[0079] The embodiment emits a reference pulse from the transmitting transducer under no-flow conditions, synchronously acquires the reference response of each channel, and records to the same time axis. Then, the time correlation alignment method is called to calculate the fixed delay of each channel relative to the common trigger, form a delay compensation table, and store in the non-volatile memory. To evaluate the stability of the unified time base under long-term operation in the field, the embodiment repeatedly acquires the reference response within a set time period, calculates the time difference fluctuation and long-term drift, gives the upper limit of jitter and drift, and sets the resynchronization period and online drift deduction amount accordingly. After completing the above process, the reference acquisition is triggered again to verify whether the compensation table is effective; if the verification residual exceeds the threshold, the embodiment automatically prompts to recalibrate until the verification is passed.

[0080] The embodiment uses a positioning jig to fix the transmitting transducer and the receiving transducer at predetermined positions, measures and records the center distance and the incident angle or the included angle between them; uses a measuring tool to measure the characteristic width of the flow resistance member and the inner diameter of the pipe, and sets the allowable tolerance range for each parameter. After all parameters are entered, the embodiment performs a short-time acquisition under a unified time base, and comprehensively judges the time alignment residual and the consistency of the reference response; when the center distance, the incident angle or the included angle, the characteristic width, the pipe inner diameter, and the time residual all meet the preset threshold, it is confirmed that the installation geometry registration is effective and the parameter file is locked for subsequent calculation and calling; if any of them exceeds the threshold, the embodiment prompts to adjust the installation or re-tighten and re-measure until all items are qualified.

[0081] The exemplary "unified time base" refers to a time reference generated by a phase-locked mechanism and consistent across all acquisition ends, used to ensure that the data of each channel is aligned on the same time axis; the "common trigger signal" refers to a trigger pulse that is bound to the unified time base and used to simultaneously start the sampling of each channel; the "time correlation alignment method" refers to a processing flow that compares the reference response waveforms within the same reference time window and calculates the relative delay of the channel based on the principle of minimum time difference; the "delay compensation table" refers to a list of compensation amounts for the fixed delay of each channel, which is deducted in the time stamp writing and data buffer reading links during operation; the "positioning jig" refers to a special fixture used to constrain the installation position and angle of the transducer, which includes angle markings and positioning surfaces; the "resynchronization period" refers to the time interval for re-executing the trigger alignment and delay verification under continuous operation conditions; the "drift compensation amount" refers to the time amount deducted online to offset the time drift caused by long-term temperature drift or mechanical loosening.

[0082] In this embodiment, the reference clock frequency is 1000000 Hz to 10000000 Hz, preferably 1000000 Hz; the sampling rate is 50000 Hz to 2000000 Hz, preferably 500000 Hz; the fixed delay control from triggering to sampling is controlled within 1 μs, the time alignment residual error is controlled in the range of 20 ns to 100 ns; the upper limit of the allowed short-term jitter is not more than 200 ns, and the upper limit of the allowed long-term drift is not more than 1 μs; the resynchronization period is set to 1800 s to 7200 s, preferably 3600 s; the reference pulse width is set to 5 μs to 20 μs, and the reference response recording window is set to 2 ms; the center distance allowed tolerance is not more than 0.5 mm, the incidence angle or the included angle allowed tolerance is not more than 0.5°, the blocking member feature width measurement resolution is not less than 0.1 mm, and the pipe inner diameter measurement uncertainty is not higher than 0.2 mm; the time alignment verification threshold is set to 100 ns, and the installation geometry comprehensive judgment threshold is set to 95% pass rate.

[0083] Specifically, in step 200, under a unified time base, the embodiment generates a trigger pulse according to a preset pulse sequence, and sends the trigger pulse to the driving end of the transmitting transducer, so that the transmitting transducer transmits ultrasonic pulses according to a preset rhythm; the receiving transducer synchronously starts collection under the same unified time base, and records the echo to form ultrasonic raw data. To ensure that the details modulated by flow are not distorted, the embodiment completes gain setting and pre-filter passband verification before collection, and immediately writes the trigger sequence number and collection start time after each trigger, which is used for subsequent cross-channel alignment. To avoid saturation caused by sudden impact, the embodiment enables amplitude limiting protection and buffer caching at the receiving end, and the ultrasonic raw data is written into the storage area in data blocks and attached with time stamps. During the effective period of each pulse sequence, the embodiment simultaneously triggers the non-ultrasonic measurement channels to start sampling through the unified time base, and obtains temperature data, pressure data, vibration data and conductivity data; each channel writes a unified time stamp and trigger sequence number at the start of sampling. After completing data collection in the same time window, the embodiment calls the alignment process to pair the ultrasonic raw data and the non-ultrasonic measurement channel data with the time stamp and trigger sequence number as the primary key, to trim the start and end boundaries of the pairing result, and to output the cross-channel synchronous data set. To control the alignment error, the embodiment calculates the start point difference and the end point difference for each alignment segment, and when the difference exceeds the preset threshold, it is marked as suspicious and enters the resampling or realignment process; the data segments successfully aligned enter the quality evaluation link.

[0084] The embodiment performs quality evaluation on the cross-channel synchronized data set, including amplitude saturation detection, noise level detection, and missing segment detection. The amplitude saturation detection is used to identify front-end overdrive or quantization top-end crushing. If a high proportion of full-scale adjacent samples appear in a specified window in any channel, it is determined to be saturated and the reliability level of the segment is downgraded. The noise level detection is used to identify environmental electromagnetic interference or mechanical noise rise. If the root mean square of the silent interval exceeds the set threshold, it is determined to be too noisy. The missing segment detection is used to identify buffer overflow or communication jitter. If the time stamp is found to be discontinuous or the sample loss exceeds the set upper limit, it is determined to be missing. The results of the above three detections are combined to generate a quality label. The quality label is divided into three categories: qualified, suspicious, and invalid. Qualified segments directly enter subsequent feature extraction. Suspicious segments are retained but are downweighted in subsequent processing. Invalid segments are rejected and a re-sampling prompt is triggered. Exemplarily, the "preset pulse sequence" refers to a pulse group used to drive the transmitting transducer within one acquisition period, which includes parameters such as pulse number, pulse interval, and group interval. The "cross-channel synchronized data set" refers to a data set obtained by merging ultrasonic raw data and non-ultrasonic measurement channel data after aligning them with a unified time base and trigger number within the same time window. The "quality label" refers to a classification identifier given for various detection results within the same time window, which is used to guide the selection and weighting of subsequent feature extraction and fusion. The "silent interval" refers to a sampling section with no valid echo or no strong excitation, which is used for noise level estimation. "Suspicious" refers to a segment whose quality does not meet the qualified threshold but still has reference value. "Invalid" refers to a segment whose quality does not meet the minimum requirement and cannot be used for subsequent processing.

[0085] Specifically, the embodiment supplements the following numerical constraints to step 200: each pulse sequence group contains 16 to 32 pulses, the pulse interval is 1000 μs, and the group interval is 50 ms; the single-block buffer capacity of each channel at the receiving end is not less than 2048 samples; the sampling rate of the temperature channel is 10 Hz, the sampling rate of the pressure channel is 2000 Hz, the sampling rate of the vibration channel is 10000 Hz, and the sampling rate of the conductivity channel is 100 Hz; the determination condition for amplitude saturation detection is that the proportion of samples reaching more than 90% of the full scale within a 5 ms window exceeds 5%; the determination condition for noise level detection is that the root mean square of the silent interval exceeds 0.5% of the full scale; the determination condition for missing segment detection is that the time stamp is discontinuous for more than 1 ms or there are more than 128 consecutive missing samples; when the alignment difference reaches 300 nanoseconds, it is marked as suspicious, and when the alignment difference exceeds 500 nanoseconds, it is marked as invalid and a re-sampling prompt is triggered.

[0086] Optionally, step 300 of the embodiment includes:

[0087] The embodiment performs analytic signal processing on the ultrasonic feature and extracts an envelope sequence; according to a unified time base, the envelope sequence and non-ultrasonic measurement channel data are divided into time slices with consistent lengths according to a fixed time length, and a fixed overlap ratio is set for overlapping and splicing verification. To ensure the effectiveness of subsequent feature extraction, the embodiment calls quality marking results before time slice division, removes segments marked as invalid, performs weight reduction processing on segments marked as suspicious, and only segments marked as qualified are directly entered into subsequent calculation. To suppress direct current drift and wideband interference, the embodiment applies bandpass processing and endpoint detrending processing on the envelope sequence, and archives the processing parameters and timestamps together, ensuring that the same batch of data can reproduce experimental results when repeatedly calculated.

[0088] The embodiment estimates the second-order cyclic autocorrelation power spectrum of the ultrasonic envelope in each time slice according to the cyclic stationary idea, obtains a joint spectrum diagram with frequency as the horizontal axis and cyclic frequency as the vertical axis, and uses it to characterize the periodic modulation caused by vortex shedding; at the same time, at least one type of non-ultrasonic measurement channel data is selected as an auxiliary channel, the amplitude squared coherence and phase consistency between the ultrasonic envelope and the auxiliary channel are calculated under the unified time base, and robust statistics are performed in the time slice dimension. To improve robustness, the embodiment uses fixed bandwidth smoothing on the joint spectrum diagram, uses sliding window median operation on the coherence and phase consistency, and performs removal and resampling strategies on abnormal time slices; if resampling is insufficient, set low weight processing for the time slice in subsequent joint scoring to avoid individual abnormalities affecting global judgment.

[0089] The embodiment searches for the maximum cyclic frequency at each frequency position on the joint spectrum diagram, and performs normalization based on the median of the full frequency band on the maximum value, then geometrically synthesizes the amplitude squared coherence at the corresponding frequency position to obtain a joint scoring curve without weight; the frequency corresponding to the global maximum position of the joint score is the feature frequency related to vortex shedding. To give the frequency uncertainty, the embodiment uses the delete-one method for resampling on the time slice set: remove one time slice and recalculate the feature frequency each time, and the dispersion of all results is taken as the frequency uncertainty; at the same time, the amplitude squared coherence and phase consistency at the feature frequency form a coherence consistency index with a value between zero and one, which is used to provide triggering and weighting basis for subsequent steps. When the coherence consistency index is lower than the threshold, the embodiment triggers the robust estimation strategy before entering the subsequent fusion link, limits the weight of the time window and records the triggering reason and time.

[0090] wherein, "analytic signal processing" refers to a process flow of converting a real signal into a complex representation with instantaneous amplitude and instantaneous phase description in time domain and extracting an amplitude sequence therefrom; "time slice" refers to a minimum analysis unit obtained by dividing continuous data with a fixed time length and a fixed overlap ratio under a uniform time base; "cyclic self-power spectrum" refers to a joint energy distribution of frequency and cyclic frequency corresponding to a signal with periodic statistical characteristics on the time axis, which is used to reveal the periodic modulation caused by vortex shedding; "amplitude squared coherence" refers to a normalized quantity of amplitude correlation of two channels at the same frequency, with a value range of zero to one; "phase consistency quantity" refers to a normalized quantity of the dispersion degree of the phase difference of two channels at the same frequency with respect to time slices, and a smaller value indicates a more consistent phase; "joint scoring curve" refers to a frequency score obtained by non-weighted synthesis of the normalized cyclic spectrum amplitude and the amplitude squared coherence, which is used to locate the characteristic frequency; "one-deletion resampling" refers to a method of repeatedly calculating a target quantity by removing one time slice at a time in a set containing a plurality of time slices, and evaluating the uncertainty by the dispersion degree of the results. Exemplarily, the time slice length of the present embodiment is 2.0 s, and the time slice overlap ratio is 0.50; the envelope bandpass lower limit is 20 Hz, and the envelope bandpass upper limit is 2000 Hz; the frequency search range of the joint spectrum is 5 Hz to 500 Hz, the cyclic frequency search range is 0.5 Hz to 200 Hz, and the cyclic frequency step is 0.5 Hz; the smoothing bandwidth is 5 Hz; the threshold value for triggering robust estimation of the coherence consistency index is 0.60, and the hysteresis threshold value for canceling robust estimation is 0.70; the number of time slices for one-deletion resampling is not less than 12, and when the number of time slices is insufficient, the frequency uncertainty is not calculated and a supplementary sampling prompt is given; the frequency resolution of joint scoring is 0.1 Hz; the allowable proportion of abnormal time slices in a single window is not higher than 0.20, and if the proportion exceeds the value, the window is determined to be invalid and a supplementary sampling instruction is recorded.

[0091] Specifically, the "median normalization" in the present embodiment is derived from the normalization processing of the cyclic spectrum amplitude with the median as the reference in the full frequency range, and its application is to suppress false peaks caused by background noise and local peaks; the "one-deletion resampling" is derived from the recalculation method of removing each time slice in a set of time slices in statistics, and its application is to evaluate the uncertainty of the characteristic frequency without external standards; the "coherence consistency index" is derived from the combined calculation of the amplitude squared coherence and the phase consistency quantity, and its application is to trigger a robust estimation strategy when the index is below a threshold value, thereby ensuring the reliability of subsequent fusion; the "joint scoring function" is derived from the geometric synthesis of the cyclic spectrum amplitude and the coherence function, and its application is to locate the vortex shedding characteristic frequency and enhance the robustness of identification.

[0092] Further, the step 400 of the present embodiment comprises:

[0093] The embodiment reads temperature, pressure and conductivity from non-ultrasonic measurement channel data under unified time base, and obtains kinematic viscosity and density according to the table lookup relationship of materials and media; combined with the pipe diameter and the calibration initial value of the previous time window, the dimensionless flow state index is calculated, which is used to judge whether the current time window belongs to laminar flow, transition or turbulent flow. In order to form a traceable correction reference, the embodiment writes the reference flow state index and the reference coherence consistency index at the initial calibration, and compares the current value with the reference value in each time window to obtain the scaling factor and the quality factor used for subsequent correction.

[0094] The embodiment takes the reference Strouhal number obtained at the calibration as the starting point, and introduces two types of adaptive correction: one type of open square correction based on the flow state index, which is used to describe the inherent frequency bandwidth shift caused by the change of viscous effect and inertial effect; the other type of normalization correction based on the coherence consistency index, which is used to automatically reduce the weight of the speed derived from the characteristic frequency when the signal quality decreases. The above two types of correction are realized in the firmware with closed mapping, which does not depend on artificial weight, and the lower limit constant is set to prevent excessive amplification caused by the denominator approaching zero. The embodiment applies the two types of correction to the reference Strouhal number in sequence to obtain the corrected Strouhal number of the current time window, and archives the intermediate quantities of the correction process with time stamp.

[0095] The embodiment calculates the geometric compensation coefficient according to the registered installation geometric parameters, corrects according to the cosine projection relationship between the included angle of the sound beam relative to the pipe axis and the calibration reference angle, and sets upper and lower limits for the correction coefficient to prevent unreasonable amplification caused by large angle error. Subsequently, the embodiment performs narrowband filtering and cross-correlation on the ultrasonic envelope and selected non-ultrasonic channels near the characteristic frequency, locates the convection time delay, and then obtains the convection flow velocity combined with the axial distance between the ultrasonic sensing position and the auxiliary sensing position; at the same time, the reference flow velocity is obtained by using the characteristic frequency and the corrected Strouhal number, the characteristic width of the flow blocking member and the geometric compensation coefficient. In order to suppress single-source bias, the embodiment uses harmonic fusion to synthesize the reference flow velocity and the convection flow velocity to estimate the flow velocity in real time, and records each intermediate result involved in the fusion together with the quality label for subsequent tracing and diagnosis.

[0096] The embodiment adopts one-time linear uncertainty propagation to transfer the frequency uncertainty, the Strouhal correction uncertainty and the convection time delay uncertainty to the flow velocity instantaneous uncertainty. The frequency uncertainty is derived from the resampling result of the previous step; the Strouhal correction uncertainty is derived from the sensitivity of the correction mapping to the flow state indicator and the coherence consistency indicator, and the sensitivity is obtained from the calibration data and the small disturbance experiment in the field; and the convection time delay uncertainty is derived from the half-width of the cross-correlation peak and the variance of repeated measurements. The embodiment calculates the three contributions in each time window, and then combines them into the flow velocity instantaneous uncertainty by square root of the sum of squares, and writes the flow velocity instantaneous uncertainty and the flow velocity instantaneous estimation into the record corresponding to the time stamp as the weight basis for subsequent time series fusion.

[0097] Exemplarily, the "closed adaptive correction" refers to converting the flow state indicator and the coherence consistency indicator into a multiplicative correction factor of the Strouhal number through a predefined analytical mapping without introducing an artificially adjustable weight; the "geometric compensation coefficient" refers to an amplitude correction quantity obtained through the cosine projection relationship according to the recorded angle and the reference angle during installation, and is used to correct the inconsistency between the sensing path and the vortex shedding path; the "convection time delay" refers to the time difference corresponding to the cross-correlation peak of the ultrasonic envelope and the auxiliary channel in the characteristic frequency neighborhood, and is used to invert the convection velocity propagating along the axial direction; the "one-time linear uncertainty propagation" refers to transferring the uncertainty of the input link to the uncertainty of the target quantity according to the sum of the products of the partial derivative sensitivity and the input uncertainty and then taking the square root under the assumption of small disturbance; and the "harmonic fusion" refers to fusing the estimation values of two positive quantities by taking the reciprocal average, so as to naturally limit the result when any source is too small or too large.

[0098] As an optional implementation, the embodiment sets the reference Strouhal number to 0.20, allows the offset range to be 0.15 to 0.25; sets the reference flow state indicator to 50000, and the online effective range to 10000 to 300000; sets the coherence consistency reference value to 0.80, and the lower constant for correction to 0.30; sets the limiting range of the geometric compensation coefficient to 0.85 to 1.15; sets the allowable range of the difference between the angle of the sound beam relative to the pipe axis and the reference angle to -10.0° to +10.0°; sets the axial spacing between the ultrasonic sensing position and the auxiliary sensing position to 0.30m to 1.20m; sets the narrowband filtering bandwidth of the cross-correlation to 5Hz to 20Hz, and sets the cross-correlation search window to 0.50s to 0.80s; sets the acceptable range of the convection time delay to 0.005s to 0.200s, and when the range is exceeded, the system is judged to be invalid and triggers resampling; and sets the target upper limit of the flow velocity instantaneous uncertainty to 5% of the fused flow velocity, and when the upper limit is exceeded, the system automatically limits the weight in subsequent time series fusion.

[0099] Specifically, before entering the time sequence fusion, the coherence consistency index generated by the reading step 300 is compared with the set threshold value. When the coherence consistency index is less than the lower threshold value, it is determined that the correlation between the ultrasonic representation and the non-ultrasonic measurement channel data in the time window is insufficient, which may cause the time sequence fusion result to be distorted, and thus the robust estimation strategy is triggered. When the coherence consistency index is restored to be higher than the hysteresis threshold value, the robust strategy is released and the normal weighting mode is restored. In the robust estimation strategy, the weight of the flow rate instantaneous estimation is limited to avoid individual abnormal fragments dominating the fusion result due to the small uncertainty. Specifically, the uncertainty weight of all fragments is truncated within a preset interval, and the median weighted or truncated average method is used to synthesize the fragments that deviate abnormally, reducing the interference of abnormal data on the whole. In this way, even under the condition of signal degradation or channel inconsistency, a smooth and reliable fusion result can be output.

[0100] In the time window triggered by the robust estimation strategy, the embodiment additionally adds a quality label to the fused flow rate and the fused flow rate uncertainty, which is used to represent that the result is obtained in the robust mode. The label will be published together at the output interface for subsequent data users to identify and distinguish; at the same time, the embodiment writes the trigger time window number, the coherence consistency index value and the trigger reason into the log file, supporting the diagnosis and tracing of long-term running conditions.

[0101] Optionally, the "robust estimation strategy" refers to a strategy of using an estimation method insensitive to abnormal values for weighting or synthesis when the coherence consistency index does not meet the set threshold value, specifically including weight truncation, median weighted and truncated average methods, which is used to ensure that reliable fusion flow rate estimation can be obtained even in the presence of abnormal or low-quality data. Demonstratively, the lower threshold value of the coherence consistency index of the embodiment is set to 0.60, and the hysteresis release threshold value is set to 0.70; the weight truncation interval is 0.10 minimum and 0.50 maximum; the median weighted uses at least 5 fragments in the window, and less than that does not perform fusion and triggers the prompt for re-sampling; when the truncated average is used, the fragments deviating from the median by more than 25% in the window are removed.

[0102] Specifically, the step 500 in the embodiment includes:

[0103] The embodiment first sets a fixed-length time window, and the flow rate instantaneous estimation and the corresponding uncertainty obtained continuously in the time window are taken as the fusion input. Each time window is defined with the trigger serial number as the starting point and the ending point under the unified time base, so as to ensure that the data in the window are integrated on the same time reference. At the end of the window, the embodiment collects all the flow rate instantaneous estimations in the window, and forms a fusion candidate set together with the corresponding uncertainty, providing input for subsequent weighted fusion.

[0104] The embodiment uses the flow rate instantaneous uncertainty as the inverse of the weight, so that the flow rate instantaneous estimation with smaller uncertainty contributes more in the fusion, and the data with larger uncertainty is automatically reduced in influence. To ensure the stability of the fusion, the embodiment sets a lower threshold for the uncertainty when calculating the weight, to prevent the fusion result from deviating due to unreasonable small individual data. The weighted data is superimposed in an arithmetic way and normalized to obtain the fusion flow rate estimation, and the fusion flow rate uncertainty is generated synchronously.

[0105] The embodiment detects outliers before weighting the input data. If a flow rate instantaneous estimation deviates from the median in the window by more than a preset proportion, it is determined as an outlier and is given a very low weight. If the proportion of outliers exceeds a threshold of the total number in the window, robust processing is triggered, and the fusion method is switched to median weighting or trimmed average, to ensure that the fusion result is not disturbed by extreme interference. The embodiment also records the abnormal trigger event and the corresponding time window number for archiving, for subsequent working condition diagnosis and tracing.

[0106] After the fusion is completed, the embodiment takes the fusion flow rate estimation and the fusion flow rate uncertainty as the representative result of the time window, and timestamps them under a unified time base. To support subsequent volume flow calculation, the embodiment writes the fusion flow rate result and the corresponding uncertainty into the database when storing, and provides output through a digital interface. The fusion flow rate uncertainty is directly input for the propagation of the volume flow uncertainty, laying the foundation for the next uncertainty link. If the fusion flow rate uncertainty exceeds a preset upper limit, the embodiment marks the output quality label as suspicious and prompts the user to supplement or review. The exemplary "time window" refers to a fixed-length continuous time interval divided under a unified time base, used for collecting and fusing multiple flow rate instantaneous estimations; "fusion candidate set" refers to a set of flow rate instantaneous estimations and corresponding uncertainties collected in the same time window; "robust processing" refers to a method of switching from traditional weighted average to median weighting or trimmed average when the proportion of outliers is too high, to reduce the influence of abnormal data on the result; "fusion flow rate uncertainty" refers to the uncertainty propagated and synthesized from each flow rate instantaneous uncertainty in the weighted fusion process, used to represent the credibility of the fusion result.

[0107] Specifically, the embodiment sets the time window length to 10s to 60s, preferably 30s; the window contains at least 10 flow rate instantaneous estimations, otherwise the window is determined to be invalid; the lower threshold of the uncertainty is set to 1% of the fusion flow rate, and only the data exceeding the lower threshold is involved in the weighting; the outlier judgment condition is that the deviation from the median in the window exceeds 20%, and the data exceeding the condition is given a very low weight; if the proportion of outliers exceeds 30% of the total number in the window, robust processing is triggered; the upper limit of the fusion flow rate uncertainty is set to 5% of the fusion flow rate, and the data exceeding the upper limit is marked as suspicious.

[0108] Further, step 600 in the embodiment includes:

[0109] The embodiment reads temperature data from non-ultrasonic measurement channel data under a unified time base, performs linear thermal expansion compensation after associating the calibrated inner diameter with the reference temperature to obtain the inner diameter of the pipeline at the current temperature, and the linear expansion coefficient is derived from the pipeline material specification or the factory warranty of the profile. The compensated current inner diameter is written into the parameter cache together with the corresponding time stamp. To improve repeatability, the embodiment deburrs and short-window smooths the temperature sequence, and limits the compensation result to an upper and lower limit to avoid size abnormalities caused by temperature occasional jumps. When the temperature sensor quality label is suspicious or invalid, the current inner diameter does not participate in the flow update of the current time window, and a temperature channel review prompt is triggered.

[0110] The embodiment calculates the area of the circular cross section according to the aforementioned current inner diameter; multiplies the cross-sectional area by the fusion flow rate estimate obtained in step five to obtain the volumetric flow, and writes the time stamp in a unified time base after the calculation is completed. To ensure traceability, the embodiment archives the current inner diameter, cross-sectional area, fusion flow rate estimate and volumetric flow used in this calculation with the same record number, and records the number of temperature samples, the number of rejected segments and the quality label used in this calculation, so as to trace back the abnormal output in the future. For windows with obvious working condition mutations, the embodiment performs a one-time robust modification on the fusion flow rate according to the window median, and then completes the calculation and archiving of the volumetric flow.

[0111] The embodiment first synthesizes the uncertainty of the current inner diameter: the uncertainty of the calibrated inner diameter and the size uncertainty caused by temperature measurement are synthesized under the assumption of linear propagation to obtain the uncertainty of the current inner diameter; then the fusion flow rate uncertainty and the current inner diameter uncertainty are propagated to the volumetric flow uncertainty to obtain the volumetric flow uncertainty corresponding to the time window. To ensure the quality of the output, the embodiment performs threshold verification on the volumetric flow and the volumetric flow uncertainty: if the volumetric flow uncertainty exceeds the upper limit, or the difference between the volumetric flow and the previous valid time window exceeds the mutation threshold, the output quality label is set to suspicious, and the user is prompted to perform resampling or calibration; when the verification is passed, the volumetric flow, the volumetric flow uncertainty and the time stamp are released as qualified records, and the same numerical value is output on the digital interface, the pulse interface or the analog current interface.

[0112] The "calibrated inner diameter" refers to the inner diameter reference value obtained by field gauges or laboratory measuring devices at a reference temperature and evaluated for uncertainty; the "linear expansion coefficient" refers to the linear dimension change coefficient of the pipe material when the temperature changes, which is derived from the material specification or material certificate; the "current inner diameter uncertainty" refers to the inner diameter size uncertainty at the current temperature obtained by one linear propagation after considering the calibrated inner diameter uncertainty and the temperature measurement uncertainty; the "one linear propagation" refers to a method of synthesizing the output uncertainty according to the sensitivity of the input quantity and the input uncertainty under the assumption of small perturbation; and the "threshold check" refers to comparing the calculated volume flow and the volume flow uncertainty with the preset upper limit and the mutation threshold, and deciding whether to output as a qualified record or trigger the process of supplementary sampling and calibration.

[0113] Specifically, the reference temperature of the embodiment is set to 20 degrees Celsius; the linear expansion coefficient of the stainless steel pipe is 0.000017 per degree Celsius, the linear expansion coefficient of the carbon steel pipe is 0.000012 per degree Celsius, and the linear expansion coefficient of the plastic pipe is 0.000070 per degree Celsius (one of the three is selected according to the actual material); the temperature measurement uncertainty is not greater than 0.2 degrees Celsius; the upper limit of the current inner diameter uncertainty is 0.15 millimeters, and if it is exceeded, it is marked as suspicious; the cross-sectional area archive precision is 0.000001 square meters; the volume flow output unit is cubic meters per hour, and the decimal place is retained to 3 digits; the upper limit of the volume flow uncertainty is 6% of the volume flow, and the mutation threshold is set to 20% of the volume flow difference of adjacent effective time windows; the calibration prompt is triggered when the abnormal mark appears for 3 consecutive time windows.

[0114] Corresponding to the above method, as shown in Figure 2 The embodiment also provides an intelligent pulse ultrasonic fluid measurement system fusing multi-sensor cooperative collection, which is characterized by comprising:

[0115] A unified time base and installation geometry registration unit is configured to establish a unified time base, align the collection time of the configured ultrasonic measurement channel and non-ultrasonic measurement channel, and register installation geometry parameters; the ultrasonic measurement channel comprises a transmitting transducer and a receiving transducer, and the non-ultrasonic measurement channel comprises at least two types of sensors for representing working conditions;

[0116] A cooperative collection and quality marking unit is configured to drive the transmitting transducer to emit ultrasonic pulses according to a preset pulse sequence under the unified time base, receive echoes by the receiving transducer to obtain flow-modulated ultrasonic representation, simultaneously collect non-ultrasonic measurement channel data, and complete cross-channel alignment and quality marking;

[0117] The vortex feature extraction and uncertainty generation unit is configured to jointly analyze the ultrasonic feature and the non-ultrasonic measurement channel data based on a joint extraction model of cyclostationarity and cross-channel coherence, determine a characteristic frequency corresponding to vortex shedding, and generate a frequency uncertainty and a coherence consistency index corresponding to the characteristic frequency.

[0118] The Strouhal correction and geometry compensation unit is configured to correct the Strouhal number online based on medium parameters and flow state criteria obtained from the non-ultrasonic measurement channel data, compensate for systematic deviations caused by installation geometry and convection effects, obtain a corrected flow rate real-time estimate, and synthesize a flow rate real-time uncertainty based on the frequency uncertainty.

[0119] The timing fusion and flow rate uncertainty unit is configured to perform timing fusion on the flow rate real-time estimate in a given time window with the flow rate real-time uncertainty as a weight, obtain a fused flow rate estimate and a fused flow rate uncertainty.

[0120] The cross-sectional area calculation and volume flow uncertainty unit is configured to calculate a cross-sectional area based on a pipe inner diameter, and obtain a volume flow and a volume flow uncertainty based on the fused flow rate estimate; the volume flow uncertainty is obtained by propagating the fused flow rate uncertainty and a geometry size uncertainty.

[0121] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other.

[0122] The principles and implementation manners of the present application are described by using specific examples in the present application. The above description of the embodiments is only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, the specific implementation manners and application ranges can be changed according to the idea of the present application. In conclusion, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A smart pulsed ultrasonic fluid measurement method integrating multi-sensor collaborative acquisition, characterized in that, include: Establish a unified time base to align the acquisition times of the configured ultrasonic measurement channels and non-ultrasonic measurement channels, and register the installation geometric parameters; The ultrasonic measurement channel includes a transmitting transducer and a receiving transducer, and the non-ultrasonic measurement channel includes at least two types of sensors for characterizing the working conditions. Under a unified time base, the transmitting transducer is driven to emit ultrasonic pulses according to a preset pulse sequence, and the receiving transducer receives the echo to obtain the ultrasonic characterization modulated by the flow. At the same time, non-ultrasonic measurement channel data is acquired, and cross-channel alignment and quality marking are completed. Based on the joint extraction model of cyclic stationarity and cross-channel coherence, the ultrasonic characterization and the non-ultrasonic measurement channel data are jointly analyzed to determine the characteristic frequency corresponding to vortex shedding, and generate frequency uncertainty and coherence consistency index corresponding to the characteristic frequency. Based on the medium parameters and flow regime criteria obtained from the non-ultrasonic measurement channel data, the Strouhal number is corrected online, and the systematic deviation caused by installation geometry and convection effects is compensated to obtain the corrected instantaneous flow velocity estimate. The instantaneous flow velocity uncertainty is obtained by synthesizing the frequency uncertainty. Within a given time window, the instantaneous flow velocity estimate is fused with the instantaneous flow velocity uncertainty as a weight to obtain a fused flow velocity estimate and a fused flow velocity uncertainty. The cross-sectional area is calculated based on the pipe's inner diameter, and the volumetric flow rate and its uncertainty are obtained based on the fused flow velocity estimate; the volumetric flow rate uncertainty is obtained from the fused flow velocity uncertainty and the geometric dimension uncertainty through the propagation law. Based on a joint extraction model of cyclic stationarity and cross-channel coherence, the ultrasonic characterization and non-ultrasonic measurement channel data are jointly analyzed to determine the characteristic frequencies corresponding to vortex shedding, and frequency uncertainty and coherence consistency indices corresponding to the characteristic frequencies are generated, including: The ultrasonic characterization is subjected to Hilbert transform to obtain an analytical signal, and the ultrasonic envelope sequence is extracted. The ultrasonic envelope sequence and the non-ultrasonic measurement channel data are divided into time slices of the same length according to the unified time base. The time slices are spliced ​​and verified with a fixed overlap ratio and an aligned set of segments is output. For each time slice, the second-order cyclic autopower spectrum of the ultrasonic envelope sequence is calculated to obtain a joint spectrum of frequency and cyclic frequency, which is used to characterize the cyclic stationary characteristics caused by vortex shedding. The formula is as follows: , ;in, The ultrasound envelope sequence; The ultrasound envelope in the ultrasound envelope sequence at frequency With cycle frequency Cyclic self-power spectral density at; This is the cyclic cross-correlation function of the ultrasound envelope; For time delay; For time; For expectation operators; This is a conjugate operation; Represents the imaginary unit; At least one type of data from the non-ultrasonic measurement channels is selected as an auxiliary channel. Based on the unified time base, the cross-power spectral density and amplitude squared coherence function of the ultrasonic envelope and the auxiliary channel in the frequency domain are calculated, and the phase consistency quantity is extracted. The formula is as follows: ;in, For auxiliary channels; For ultrasound envelope and auxiliary channel at frequency Cross-power spectral density at; and Each represents its respective power spectral density; It is a squared coherence function; The root mean square deviation of the cross-spectral phase over the window set; Number of time slices; This is the phase operator; the overline indicates the average over the time slice dimension. Indicates the first Within a time slice, at frequency The cross-power spectral density function between the ultrasonic envelope and the auxiliary channel at the location is used to characterize the degree of energy coupling between the two signals in the frequency domain. For each frequency in the cyclic spectrum, search for the maximum cyclic frequency and perform robust normalization. Construct an unweighted joint scoring function and determine the feature frequencies, as shown in the formula: , , , ;in, In frequency This allows the cyclic spectrum to achieve an extremely high cyclic frequency; The cyclic spectral amplitude is based on median normalization and is used to suppress background noise and sidelobes; These are the baseline normalization coefficients obtained by performing median operations on the cyclic spectrum amplitude along the frequency dimension; For joint scoring functions; The characteristic frequency corresponding to vortex shedding; The frequency uncertainty is obtained by statistically estimating the characteristic frequencies within the time slice using the one-out-of-a-row resampling method, and the formula is as follows: ;in, To remove the first The feature frequency estimate obtained after one time slice; This is the average value obtained from the characteristic frequency estimates of all time slices; For frequency uncertainty; Coherence intensity and phase consistency are combined into a coherence consistency index with a value between zero and one, which is used to provide triggering and weighting criteria for subsequent steps. The formula is as follows: ;in, As a coherence consistency indicator; Let the amplitude squared coherence function be the coherence function at the characteristic frequency; The root mean square deviation of the cross-spectral phase at the characteristic frequency; Pi is the mathematical constant of a circle.

2. The intelligent pulse ultrasonic fluid measurement method based on multi-sensor collaborative acquisition according to claim 1, characterized in that, The installation geometry parameters include the center distance between the transmitting and receiving transducers, the incident angle or included angle, the characteristic width of the flow-blocking component, and the inner diameter of the pipe.

3. The intelligent pulse ultrasonic fluid measurement method based on multi-sensor collaborative acquisition according to claim 1, characterized in that, Also includes: When the coherence consistency index does not meet the threshold, a robust estimation strategy is switched to before performing time series fusion to limit weights and suppress outliers.

4. The intelligent pulse ultrasonic fluid measurement method based on multi-sensor collaborative acquisition according to claim 1, characterized in that, Establish a unified time base to align the acquisition times of the configured ultrasonic measurement channels and non-ultrasonic measurement channels, and register the installation geometry parameters, including: Provide a phase-locked time reference, generate a reference clock and a common trigger signal, and distribute the reference clock and the common trigger signal to the acquisition terminals of the ultrasonic measurement channel and the non-ultrasonic measurement channel; Set a uniform sampling rate, timestamp format, and triggering rules at each acquisition terminal to complete timestamp alignment verification and record the alignment results; Under no-current conditions, the transmitting transducer emits a reference pulse, each channel synchronously records the reference response, and calculates the fixed delay of each channel relative to the reference based on the time-related alignment method to generate a delay compensation table. The reference response is repeatedly collected within a set time period, and the time difference fluctuation and drift are statistically analyzed. The resynchronization period and drift compensation are set and deducted online during the collection process. The transmitting transducer and the receiving transducer are installed and fixed using a positioning fixture. The center distance, incident angle or included angle of the transmitting transducer and the receiving transducer are measured and recorded. The characteristic width of the flow-blocking component and the inner diameter of the pipe are measured and recorded, and the corresponding tolerance range is given.

5. The intelligent pulse ultrasonic fluid measurement method based on multi-sensor collaborative acquisition according to claim 1, characterized in that, Under a unified time base, the transmitting transducer is driven to emit ultrasonic pulses according to a preset pulse sequence, and the receiving transducer receives the echoes to obtain the flow-modulated ultrasonic characterization. Simultaneously, data from non-ultrasonic measurement channels are acquired, and cross-channel alignment and quality labeling are completed, including: A trigger pulse signal is generated based on the unified time base and sent to the drive end of the transmitting transducer, so that the transmitting transducer emits ultrasonic pulses according to a preset pulse sequence; The echo signal is acquired by the receiving transducer under the unified time base to obtain the raw ultrasonic data containing the fluid modulation effect; During each pulse sequence emitted by the transmitting transducer, the non-ultrasonic measurement channel is simultaneously triggered to start sampling using the unified time base, thereby obtaining temperature data, pressure data, vibration data, and conductivity data. The raw ultrasound data and the non-ultrasound measurement channel data are aligned with a unified timestamp to form a cross-channel synchronous dataset; The quality of the cross-channel synchronous dataset is evaluated, including amplitude saturation detection, noise level detection, and missing segment detection, and quality tags are added to unqualified data segments.

6. The intelligent pulse ultrasonic fluid measurement method based on multi-sensor collaborative acquisition according to claim 1, characterized in that, Based on the medium parameters and flow regime criteria obtained from the non-ultrasonic measurement channel data, the Strouhal number is corrected online, and systematic deviations caused by installation geometry and convection effects are compensated to obtain a corrected instantaneous velocity estimate. The instantaneous velocity uncertainty is then synthesized based on the aforementioned frequency uncertainty, including: The medium temperature, pressure and conductivity are calculated from the non-ultrasonic measurement channel data to obtain the kinematic viscosity and density. The Reynolds number and flow regime criteria are determined by combining the pipe inner diameter and the initial calibration value of the previous time window. The corresponding reference Reynolds number and reference coherence index are selected as correction benchmarks. Based on the Reynolds number and the aforementioned coherence consistency index, a closed-loop adaptive correction is performed on the baseline Strouhal number to obtain the corrected Strouhal number for the current time slice, as shown in the formula: ;in, To correct the Strauhal number; The baseline Strouhal number; The Reynolds number for the current time slice; For reference Reynolds number; As a coherence consistency indicator; For reference coherence consistency index; This is a lower limit constant used to prevent division by zero and over-amplification; The geometric compensation coefficient is calculated based on the registered installation geometric parameters to make the vortex shedding conversion model consistent with the sensing path projection. The formula is as follows: ;in, The geometric compensation coefficient; The angle of incidence or the included angle between the sound beam formed by the transmitting and receiving transducers and the axis of the pipe; This serves as a reference angle during calibration. Narrowband filtering is applied to the ultrasonic envelope and a selected non-ultrasonic channel near the characteristic frequency, and the cross-correlation is calculated to obtain the convection time delay. An instantaneous estimate based on the frequency model and the convection model is constructed, and unweighted fusion is used to suppress single-source bias. The formula is as follows: , , ;in, The baseline flow velocity is obtained based on the characteristic frequency and the corrected Strouhal number; The characteristic frequency; The characteristic width of the flow-blocking component; The convective velocity is obtained by inversion from the convection time delay. The axial distance between the ultrasonic sensing position and the auxiliary sensing position; The convection time delay is obtained by cross-correlation within a narrow band of the characteristic frequency. For real-time flow velocity estimation; Using linear uncertainty propagation, the frequency uncertainty, the Strouhal corrected uncertainty, and the convection delay uncertainty are combined into an instantaneous velocity uncertainty, as shown in the following formula: ;in, The instantaneous uncertainty of the flow velocity; For frequency uncertainty; Correct the uncertainty corresponding to Strauhal; This represents the uncertainty of the convection time delay.

7. The intelligent pulse ultrasonic fluid measurement method based on multi-sensor collaborative acquisition according to claim 1, characterized in that, The cross-sectional area is calculated based on the pipe's inner diameter, and the volumetric flow rate and its uncertainty are obtained based on the fused flow velocity estimate, including: Under the unified time base, the temperature data in the non-ultrasonic measurement channel is read, and the current inner diameter is obtained by temperature compensation of the calibrated inner diameter, using the following formula: ;in, The current inner diameter of the pipe at the current temperature; The calibrated inner diameter at the reference temperature; The coefficient of linear expansion of the pipe material; The current medium temperature; For reference temperature; The area of ​​the circular cross-section is calculated based on the current inner diameter using the following formula: ;in, This refers to the cross-sectional area of ​​the pipe. Pi; The volumetric flow rate is obtained by multiplying the cross-sectional area by the estimated fusion velocity, as shown in the formula: ;in, Volumetric flow rate; For fusion velocity estimation; The current inner diameter uncertainty is obtained by linearly combining the calibrated inner diameter uncertainty with the temperature-induced dimensional uncertainty, using the following formula: ;in, This represents the current uncertainty of the inner diameter. To determine the uncertainty of the inner diameter; The coefficient of linear expansion; To calibrate the inner diameter; For temperature measurement uncertainty; Based on the propagation law of first-order uncertainty, the combined flow velocity uncertainty and inner diameter uncertainty are propagated to the volumetric flow rate uncertainty, as shown in the formula: , ;in, For volumetric flow rate uncertainty; To incorporate the uncertainty of flow velocity.

8. An intelligent pulsed ultrasonic fluid measurement system integrating multi-sensor collaborative acquisition, characterized in that, include: A unified time base and installation geometry registration unit is used to establish a unified time base, align the acquisition times of the configured ultrasonic measurement channels and non-ultrasonic measurement channels, and register the installation geometry parameters; the ultrasonic measurement channel includes a transmitting transducer and a receiving transducer, and the non-ultrasonic measurement channel includes at least two types of sensors for characterizing the operating conditions; The collaborative acquisition and quality labeling unit is used to drive the transmitting transducer to emit ultrasonic pulses according to a preset pulse sequence under a unified time base, and the receiving transducer receives the echo to obtain the flow-modulated ultrasonic characterization. At the same time, it acquires data from non-ultrasonic measurement channels and completes cross-channel alignment and quality labeling. The vortex feature extraction and uncertainty generation unit is used to perform joint analysis on the ultrasonic characterization and non-ultrasonic measurement channel data based on the joint extraction model of cyclic stationary characteristics and cross-channel coherence, determine the characteristic frequency corresponding to vortex shedding, and generate frequency uncertainty and coherence consistency index corresponding to the characteristic frequency. The Strouhal correction and geometric compensation unit is used to correct the Strouhal number online based on the medium parameters and flow criterion obtained from the non-ultrasonic measurement channel data, and to compensate for the systematic deviation caused by the installation geometry and convection effect, so as to obtain the corrected instantaneous flow velocity estimate, and to obtain the instantaneous flow velocity uncertainty based on the frequency uncertainty. The time series fusion and flow velocity uncertainty unit is used to perform time series fusion of the instantaneous flow velocity estimate with the instantaneous flow velocity uncertainty as a weight within a given time window, so as to obtain the fused flow velocity estimate and the fused flow velocity uncertainty. The cross-sectional area calculation and volumetric flow rate uncertainty unit is used to calculate the cross-sectional area based on the pipe's inner diameter and obtain the volumetric flow rate and its uncertainty based on the fused flow velocity estimate; the volumetric flow rate uncertainty is obtained from the fused flow velocity uncertainty and the geometric dimension uncertainty through the propagation law. Based on a joint extraction model of cyclic stationarity and cross-channel coherence, the ultrasonic characterization and non-ultrasonic measurement channel data are jointly analyzed to determine the characteristic frequencies corresponding to vortex shedding, and frequency uncertainty and coherence consistency indices corresponding to the characteristic frequencies are generated, including: The ultrasonic characterization is subjected to Hilbert transform to obtain an analytical signal, and the ultrasonic envelope sequence is extracted. The ultrasonic envelope sequence and the non-ultrasonic measurement channel data are divided into time slices of the same length according to the unified time base. The time slices are spliced ​​and verified with a fixed overlap ratio and an aligned set of segments is output. For each time slice, the second-order cyclic autopower spectrum of the ultrasonic envelope sequence is calculated to obtain a joint spectrum of frequency and cyclic frequency, which is used to characterize the cyclic stationary characteristics caused by vortex shedding. The formula is as follows: , ;in, The ultrasound envelope sequence; The ultrasound envelope in the ultrasound envelope sequence at frequency With cycle frequency Cyclic self-power spectral density at; This is the cyclic cross-correlation function of the ultrasound envelope; For time delay; For time; For expectation operators; This is a conjugate operation; Represents the imaginary unit; At least one type of data from the non-ultrasonic measurement channels is selected as an auxiliary channel. Based on the unified time base, the cross-power spectral density and amplitude squared coherence function of the ultrasonic envelope and the auxiliary channel in the frequency domain are calculated, and the phase consistency quantity is extracted. The formula is as follows: , ;in, For auxiliary channels; For ultrasound envelope and auxiliary channel at frequency Cross-power spectral density at; and Each represents its respective power spectral density; It is a squared coherence function; The root mean square deviation of the cross-spectral phase over the window set; Number of time slices; This is the phase operator; the overline indicates the average over the time slice dimension. This indicates that in the m-th time slice, at frequency The cross-power spectral density function between the ultrasonic envelope and the auxiliary channel at the location is used to characterize the degree of energy coupling between the two signals in the frequency domain. For each frequency in the cyclic spectrum, search for the maximum cyclic frequency and perform robust normalization. Construct an unweighted joint scoring function and determine the feature frequencies, as shown in the formula: , , , ;in, In frequency This allows the cyclic spectrum to achieve an extremely high cyclic frequency; The cyclic spectral amplitude is based on median normalization and is used to suppress background noise and sidelobes; These are the baseline normalization coefficients obtained by performing median operations on the cyclic spectrum amplitude along the frequency dimension; For joint scoring functions; The characteristic frequency corresponding to vortex shedding; The frequency uncertainty is obtained by statistically estimating the characteristic frequencies within the time slice using the one-to-one deletion resampling method, as shown in the formula. ;in, To remove the first The feature frequency estimate obtained after one time slice; This is the average value obtained from the characteristic frequency estimates of all time slices; For frequency uncertainty; Coherence intensity and phase consistency are combined into a coherence consistency index with a value between zero and one, which is used to provide triggering and weighting criteria for subsequent steps. The formula is as follows: ;in, As a coherence consistency indicator; Let the amplitude squared coherence function be the coherence function at the characteristic frequency; The root mean square deviation of the cross-spectral phase at the characteristic frequency; Pi is the mathematical constant of a circle.

Images