A method for calibrating an outlet flow meter using a clear water tank water level

Abstract

This invention relates to the field of flow measurement and calibration technology, and discloses a method for calibrating an outlet flow meter using the water level of a clear water tank. The method includes determining a safe frequency to avoid water body resonance through spectral analysis; controlling a regulating valve to generate periodic disturbances based on this frequency and simultaneously collecting water level, valve position, and flow data; using water level changes to correct the valve position signal for hydraulic gain decoupling, eliminating the influence of outflow nonlinearity; reconstructing the water level change into a flow reference in the frequency domain, comparing it with the measured flow rate to calculate the calibration coefficient, and constructing a model to complete online compensation. This invention, by applying dynamic disturbances in the frequency domain to avoid water body fluctuation interference and combining a hydraulic model for nonlinear decoupling to overcome the influence of head changes, achieves online automatic calibration of the outlet flow meter without interrupting water supply. This helps solve the problem of insufficient accuracy in traditional volumetric methods, improving the long-term measurement accuracy of large-diameter flow meters and the reliability of system operation.

Description

Technical Field

[0001] This invention relates to the field of flow measurement and calibration technology, and in particular to a method for calibrating an outlet flow meter using the water level of a clear water tank. Background Technology

[0002] In urban water supply systems, the outlet flow meter, located downstream of the clear water tank, is the core equipment for water volume measurement, cost accounting, and production-sales difference analysis. The accuracy of the measurement data is the foundation for the refined scheduling and management of the water supply system. Therefore, regular calibration of the outlet flow meter is a key step in ensuring its measurement performance and data reliability.

[0003] For online calibration of large-diameter pipeline flow meters, a common technique is the volumetric method. The volumetric method typically treats the clear water tank in the water plant as a standard container with a known volume. During the calibration operation, the inlet valve of the clear water tank is first closed to cut off the water supply. Then, water is continuously supplied through the outlet flow meter for a period of time, and the drop in water level in the clear water tank during this period is accurately measured. Finally, the measured drop in water level is multiplied by the effective cross-sectional area of ​​the clear water tank to calculate the theoretical total outflow during this period. This theoretical value is then compared with the cumulative flow reading of the outlet flow meter during the same period to assess the metering error.

[0004] However, the aforementioned volumetric method has inherent limitations in practical applications. As a large open water body, the clear water tank is susceptible to continuous surface fluctuations caused by factors such as the residual waves of incoming water disturbance or wind force, and may even induce water resonance under certain conditions. At the same time, during the calibration of the outflow, the water level in the clear water tank continuously decreases. This change in water level directly leads to a reduction in the static pressure head acting on the outflow pipe, which in turn makes the instantaneous outflow rate through the regulating valve not a constant value. The instability of the water body makes it difficult to accurately measure minute changes in water level, and the nonlinear outflow process introduces a fundamental error into the practice of using the average flow rate as a comparison benchmark. These factors work together to make it difficult to guarantee the accuracy and repeatability of the calibration results of the traditional volumetric method, thereby affecting the long-term reliability of the metering data of the entire water supply system. Summary of the Invention

[0005] The purpose of this invention is to provide a method for calibrating an outlet flow meter using the water level of a clear water tank. This method solves the problem that the traditional volumetric method is easily affected by water fluctuations during calibration and cannot effectively handle nonlinear outflow caused by water level changes, resulting in insufficient calibration accuracy.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The first aspect of the present invention provides a method for calibrating an outlet flow meter using the water level of a clear water tank, comprising the following steps:

[0008] Spectral analysis was performed on the water level response data of the clear water tank. By identifying the water resonance peak and searching for the minimum amplitude point in the water level amplitude-frequency characteristic curve, a safe frequency to avoid water resonance was determined, and the safe frequency was used as the frequency domain reference for calibration.

[0009] The safe frequency control feedback execution adjustment unit generates periodic disturbances and simultaneously collects the water level data of the clear water tank, the valve position signal of the adjustment unit, and the instantaneous flow data of the outlet flow meter.

[0010] The water level data is low-pass filtered to separate the quasi-static water level trend. Based on the orifice outflow principle of fluid mechanics, the valve position signal is weighted and corrected using the reciprocal of the square root of the quasi-static water level trend to eliminate the nonlinear influence of water level changes on the valve flow characteristics and generate an equivalent linear excitation signal. Simultaneously, the physical lag time between the equivalent linear excitation signal and the instantaneous flow data is calculated using a cross-correlation analysis algorithm, and the instantaneous flow data is time-aligned accordingly to obtain the aligned flow signal.

[0011] Based on the principle of fluid continuity and frequency domain differential characteristics, the spectral amplitude of the water level data at a safe frequency is converted into a flow reference amplitude; by calculating the ratio of the flow reference amplitude to the measured spectral amplitude of the aligned flow signal at a safe frequency, the flowmeter calibration coefficient under the current operating conditions is obtained.

[0012] Repeat the above steps under different operating conditions to summarize multiple sets of flow meter calibration coefficients, and use polynomial function fitting to generate a flow meter gain error calibration model; during normal operation, use the model to perform online compensation on the flow data of the outlet flow meter to complete the calibration.

[0013] A second aspect of the present invention provides a system for calibrating an outlet flow meter using the water level of a clear water tank, the system comprising physical components and a central processing control unit.

[0014] The physical components are coupled along the fluid flow direction and include a clear water tank as a volume reference, a liquid level sensing unit for capturing water level changes, an actuation and regulation unit with feedback configured with a position feedback sensor, and an outlet flow meter as the object to be calibrated.

[0015] The central processing control unit is connected to the physical components via an industrial fieldbus and has embedded several logic function modules to execute the method. The logic function modules include:

[0016] The frequency scanning and optimization module is configured to send a frequency conversion scanning command to the execution adjustment unit and analyze the water level response to determine the safe frequency.

[0017] The multi-source excitation and synchronous acquisition module is configured to drive the execution adjustment unit according to the safe frequency and control the multi-channel sampler to ensure the synchronous acquisition of three signals: liquid level, valve position and flow rate.

[0018] The hydraulic gain decoupling module is configured to execute a hydraulic correction algorithm, receive valve position and water level data, output an equivalent linear excitation signal that eliminates the influence of nonlinear hydraulic gain, and calculate the transmission delay between signals.

[0019] The frequency domain calibration calculation module is configured to reconstruct the flow reference amplitude, extract the spectral amplitude of each signal at the characteristic frequency, calculate the flow meter calibration coefficient by comparison, and finally generate and apply the flow meter gain error calibration model.

[0020] In summary, the present invention has at least one of the following beneficial technical effects:

[0021] 1. This invention uses a frequency optimization method based on spectrum analysis, which helps to improve the accuracy of calibration. This method actively identifies and avoids the water resonance frequency of the clear water pool and selects a safe frequency with stable system response as the disturbance reference. This helps to eliminate the fluctuation of the metrological reference caused by water sloshing or standing wave effect from the source, and provides a reliable premise for the accuracy of subsequent calibration.

[0022] 2. This invention utilizes the quasi-static component in the water level data to perform real-time weighted correction on the valve position signal, transforming the nonlinear valve outflow characteristics into an equivalent linear excitation signal. This ensures that the system input and output meet the analytical premise of a linear system, improving the method's adaptability to changes in actual operating conditions and helping to solve the nonlinear interference problems caused by valve characteristics and head changes.

[0023] 3. By constructing a full-condition gain error calibration model, this invention enables online and continuous automatic calibration of the outlet flow meter. This method can automatically compensate based on the real-time valve opening during normal operation of the water plant without water outages or additional manual intervention. This online calibration capability is of positive significance for ensuring the metering accuracy of the outlet flow meter in long-term operation and reducing system maintenance costs. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0025] Figure 2 This is a system framework diagram of the present invention. Detailed Implementation

[0026] The following is in conjunction with the appendix Figure 1 - Appendix Figure 2 The present invention will be further described in detail below.

[0027] Reference Appendix Figure 1 This invention provides a method for calibrating an outlet flow meter using the water level of a clear water tank, comprising the following steps:

[0028] S1: In order to establish a safe physical boundary for online calibration and avoid the inherent fluid resonance interference of large containers, frequency sweep identification and excitation frequency optimization are performed on the fluid dynamic characteristics of the clear water tank water supply system.

[0029] Specifically, the central processing control unit sends a scanning command with a frequency that changes linearly with time to the outlet regulating valve, driving the valve to superimpose a slight frequency-varying oscillation action on the current working opening; at the same time, it monitors the dynamic response of the clear water tank water level to the frequency-varying action in real time, and constructs a relationship reflecting the amplitude-frequency characteristics of the clear water tank to the pipeline system under the current operating conditions.

[0030] Based on this amplitude-frequency characteristic relationship, the inherent fluid resonance frequency band of the system is identified. Under the constraint of avoiding the resonance frequency band and the main fluctuation frequency band of the influent flow, a frequency value in the flat region of amplitude-frequency response is adaptively selected and locked as the target calibration frequency, which serves as the driving reference for subsequent active excitation steps.

[0031] S2: In response to the target calibration frequency determined in step S1, apply a fixed-frequency micro-perturbation excitation to the system and simultaneously collect three data points reflecting the system's input, response, and output characteristics to establish a calibration data source.

[0032] Specifically, the frequency of the outlet regulating valve is controlled to be a fixed-frequency sinusoidal micro-motion at the target calibration frequency, thereby introducing an active disturbance signal with a specific frequency fingerprint into the outlet flow.

[0033] While applying the excitation, a unified clock reference is established, and the actual valve opening feedback data representing the actual action of the actuator, the clear water tank water level data representing the system volume response, and the instantaneous flow data of the outlet flow meter representing the reading of the device under test are collected synchronously. The above three sets of data streams that maintain strict time synchronization are transmitted to the data processing unit as the input basis for subsequent decoupling and feature extraction.

[0034] S3: Based on the principle of orifice outflow in fluid mechanics, the raw data collected in step S2 is subjected to hydraulic gain normalization to eliminate nonlinear errors introduced by changes in water level potential energy and to complete the time-series decoupling of multi-source signals.

[0035] Specifically, the water level data of the clear water tank is first filtered to separate the quasi-static water level trend that represents pressure potential energy.

[0036] Secondly, based on the physical principle that the flow rate is positively correlated with the square root of the pressure head before and after the valve, the square root value of the quasi-static water level trend is used to dynamically weight and correct the actual valve opening feedback data, thereby generating an equivalent linear excitation signal that is linearly related to the flow rate and eliminates the influence of nonlinear hydraulic gain.

[0037] Finally, cross-correlation analysis is performed on the equivalent linear excitation signal and the instantaneous flow data of the outlet flow meter to calculate the physical lag time caused by fluid transmission in the pipeline, and the signals are time-aligned accordingly to provide a linear and synchronous data basis for frequency domain analysis.

[0038] S4: Parallel processing of the water level data in the clear water tank in step S2, extracting the flow reference features unaffected by inflow through liquid level differential reconstruction and frequency domain transformation.

[0039] Specifically, the instantaneous rate of change of water level is calculated using a numerical differential algorithm, and combined with the known effective geometric cross-sectional area of ​​the clear water tank, the rate of change of water level is converted into a reference flow fluctuation signal. In this process, the constant inflow rate and background leakage in the system are converted into DC bias components.

[0040] Subsequently, a discrete Fourier transform is performed on the reference flow fluctuation signal to accurately extract the spectral amplitude at the target calibration frequency in step S1 in the frequency domain, forming the reference frequency domain feature, thereby physically shielding the background noise interference of different frequencies.

[0041] S5: Combine the aligned instrument signal obtained in step S3 with the reference feature obtained in step S4, perform frequency domain feature comparison and verification, and complete the closed-loop correction of the instrument parameters.

[0042] Specifically, the instantaneous flow data of the outlet flow meter after time alignment in step S3 is transformed to the frequency domain, and the spectral amplitude at the target calibration frequency is extracted to form the frequency domain characteristics of the instrument; then, the dual verification logic is executed:

[0043] First, compare the phase difference between the reference signal and the instrument signal to see if it conforms to the physical laws of hydraulic transmission, in order to verify the validity of this calibration data;

[0044] Secondly, after successful verification, the amplitude ratio of the reference frequency domain characteristics to the instrument frequency domain characteristics is calculated to obtain the correction coefficient of the flow meter; finally, the measurement parameters of the outlet flow meter are updated online using the correction coefficient to complete the calibration closed loop.

[0045] The various steps of the present invention will now be described in detail and fully.

[0046] The specific implementation process of step S1, which involves frequency sweep identification of the system's hydrodynamic characteristics and selection of the safe excitation frequency, mainly includes:

[0047] The sub-steps are as follows: S101, constructing the frequency conversion scanning excitation signal; S102, constructing the amplitude-frequency characteristic relationship; S103, identifying the interference frequency band; and S104, locking the target calibration frequency.

[0048] Detailed explanation is as follows:

[0049] Sub-step S101: Construct and apply the frequency conversion scanning excitation signal

[0050] When the clear water tank supply system is in normal operation, the central processing control unit acquires the average opening command of the current outlet regulating valve. This average opening command is used as the reference operating point for maintaining normal water supply flow. The central processing control unit generates a scanning signal whose frequency varies linearly with time. This scanning signal is superimposed on the average opening command to form the final control command sent to the valve positioner. This control command drives the outlet regulating valve to perform a small-amplitude frequency scanning action, thereby introducing test excitation into the fluid system. The scanning signal is preferably a linear frequency modulated signal, and the time-domain expression of the control command is as follows:

[0051] ;

[0052] In the formula, For a moment The opening control command applied to the outlet regulating valve; This is the current average valve opening reference value; The amplitude of the scanning excitation; This is the scan start frequency; This is the scan termination frequency; The duration of a single frequency sweep cycle.

[0053] The specific implementation of parameter settings includes:

[0054] The amplitude of the above scanning excitation The value must be less than 2% (e.g., 1%) of the total valve stroke.

[0055] Amplitude of scanning excitation Limits are used to ensure that pipeline pressure fluctuations during calibration testing remain within safe thresholds.

[0056] The range of the scan start frequency and scan end frequency settings must cover the natural resonant frequency range of the clear water pool.

[0057] For conventional large-scale clear water tank water supply systems, the scanning start frequency The scan termination frequency can be set to 0.001Hz. It can be set to 0.1Hz.

[0058] Sub-step S102: Acquire the dynamic response of the liquid level and construct the amplitude-frequency characteristic relationship.

[0059] During the frequency scanning signal execution process of the outlet regulating valve, the high-precision level gauge continuously collects water level data from the clear water tank.

[0060] The data processing unit first performs linear detrending processing on the collected water level time series data. The specific linear detrending processing method is as follows:

[0061] The DC trend line of the water level data was fitted using the least squares method, and the original water level data was then subtracted from the DC trend line. Linear detrending processing eliminated the monotonic rise and fall components of the liquid level caused by the imbalance of inflow and outflow rates, while retaining the liquid level fluctuation components caused by the sweep frequency action of the outflow regulating valve.

[0062] Subsequently, the data processing unit uses the Fast Fourier Transform algorithm to map the liquid level fluctuation component from the time domain to the frequency domain. Combining the spectrum of the input frequency scanning signal, the data processing unit calculates the frequency response function of the clear water tank supply system. The amplitude-frequency characteristics of the clear water tank supply system are calculated based on the following formula:

[0063] ;

[0064] In the formula, For frequency The amplitude-frequency response modulus at that point; The amplitude of the spectrum after Fourier transform of the liquid level fluctuation signal; The amplitude of the frequency spectrum of the opening control command signal after Fourier transform.

[0065] The data processing unit calculates point by point within the scanning frequency range to construct an amplitude-frequency characteristic curve describing the relationship between valve action and water level response. The amplitude-frequency characteristic curve reflects the feasibility and sensitivity of using water level changes to infer flow rate changes at different frequencies.

[0066] Sub-step S103: Identify the fluid resonance frequency band and the influent interference frequency band

[0067] Based on the constructed amplitude-frequency characteristic curve, the system identifies the interference frequency bands that affect the calibration accuracy of the flow meter. For the fluid resonance frequency band, the central processing and control unit searches for local maxima on the amplitude-frequency characteristic curve using the sliding window extremum search method.

[0068] The central processing and control unit defines frequency points where the amplitude exceeds the reference amplitude by more than 3 dB as resonance peaks, and marks the frequency corresponding to the resonance peak and its neighborhood as the fluid resonance frequency band. Within the fluid resonance frequency band, water level fluctuations include amplitude distortion caused by resonance, making them unsuitable as a measurement reference.

[0069] For the inlet interference frequency band, before executing sub-step S101, the system pre-collects a section of static water level data of the clear water tank when no disturbance is applied, and performs spectrum analysis on it to obtain the background noise spectrum.

[0070] The system identifies low-frequency regions with concentrated energy within the background noise spectrum. The system marks low-frequency regions (e.g., below 0.005 Hz) where the power spectral density is significantly higher than the average noise floor of the high-frequency band (e.g., more than 5 times the average noise floor of the high-frequency band) as the inlet interference frequency band. Within this frequency band, random fluctuations in the inlet flow rate can mask the calibration signal.

[0071] Sub-step S104: Adaptively select and lock the target calibration frequency

[0072] The system iterates through all frequency points within the scanning range and selects the optimal excitation frequency. The frequency selection logic must simultaneously satisfy the following three conditions:

[0073] Condition 1: The selected frequency must be outside the fluid resonance frequency band to prevent water level data distortion caused by violent water sloshing.

[0074] Condition 2: The selected frequency must be higher than the upper limit of the inlet interference frequency band to ensure that the inlet fluctuations are represented as a filterable DC component.

[0075] Condition 3: The selected frequency should be located in the flat region of the amplitude-frequency characteristic curve, that is, the amplitude change rate near the selected frequency is less than the preset threshold.

[0076] The central processing unit selects the frequency with the highest signal-to-noise ratio (SNR) from the set of frequencies that meet the above conditions. The SNR is defined as the magnitude of the amplitude-frequency response at that frequency point. The ratio of the amplitude of the same frequency in the background noise spectrum obtained in sub-step S103. The central processing control unit locks the selected frequency value as the target calibration frequency, denoted as... The target calibration frequency is stored in the system register and used as the driving parameter for the fixed-frequency micro-perturbation excitation in step S2.

[0077] This implementation transforms the complex fluid dynamics environment into quantifiable spectral characteristics through frequency sweep identification and frequency optimization. By identifying and avoiding swell resonance and inflow interference, it helps overcome the shortcomings of the traditional water level-volume method, which cannot accurately obtain the volume change benchmark under dynamic inflow and water sloshing conditions. The process of frequency sweep identification and frequency optimization provides physical stability support for the subsequent reconstruction of a high-precision flow benchmark using water level differential.

[0078] For step S2, which involves the synchronous acquisition of three-dimensional data based on the target calibration frequency and the fixed-frequency micro-perturbation excitation, the specific implementation process of step S2 mainly includes: sub-step S201 of generating fixed-frequency excitation control instructions, sub-step S202 of executing valve perturbation control, sub-step S203 of constructing the timing sequence for synchronous acquisition of multi-source data, and sub-step S204 of verifying data validity and storage.

[0079] Detailed explanation is as follows:

[0080] Sub-step S201: Generate fixed-frequency excitation control command

[0081] The central processing control unit retrieves the target calibration frequency locked in step S1. The central processing control unit uses the target calibration frequency as a reference and combines it with the current average opening degree of the outlet regulating valve to generate a sinusoidal waveform control sequence. This control sequence is used to drive the valve to oscillate at a constant amplitude near its stable operating point, thereby introducing a single-frequency component into the outlet flow rate. The time-domain mathematical expression of the fixed-frequency excitation control command is as follows:

[0082] ;

[0083] In the formula, For a moment The opening command value sent to the actuator of the outlet water regulating valve; This is the current average valve opening reference value; To calibrate the excitation amplitude; Calibrate the target frequency; The initial phase of the excitation signal is usually set to 0.

[0084] Regarding the setting of the calibration excitation amplitude:

[0085] Calibrate excitation amplitude The settings must adhere to the principle of local linearization. Since the opening-flow characteristics of industrial valves are typically nonlinear, to ensure that subsequent signal processing is compatible with linear system theory, the calibration excitation amplitude needs to be adjusted. The amplitude is controlled within the linear region near the valve's current operating point (i.e., the flow characteristic curve within this range is approximately a straight line). Typically, this value is taken as 1% to 3% of the valve's current opening. Too small an amplitude will result in insufficient signal-to-noise ratio, while too large an amplitude introduces nonlinear high-order harmonic distortion, affecting the accuracy of subsequent frequency domain analysis.

[0086] Sub-step S202: Perform valve disturbance control

[0087] The central processing and control unit sends the generated fixed-frequency excitation control command to the positioner of the outlet regulating valve in real time via an analog output channel or fieldbus interface. The actuator of the outlet regulating valve responds to the fixed-frequency excitation control command, driving the valve core to perform reciprocating motion. This reciprocating motion of the outlet regulating valve changes the flow cross-sectional area of ​​the outlet pipeline, thereby forcing the fluid flowing through the outlet pipeline to generate flow fluctuations at the same frequency as the target calibration frequency. These flow fluctuations cause periodic changes in the water level of the clear water tank and directly affect the downstream outlet flow meter.

[0088] In this process, to eliminate the impact of mechanical lag and dead zone of the actuator on control accuracy, a displacement sensor inside the outlet regulating valve monitors the actual position of the valve core in real time. The displacement sensor converts the actual position into an electrical signal, which is then uploaded to the central processing and control unit as feedback data of the actual valve opening, providing a true actuator basis for the nonlinear decoupling in the subsequent step S3.

[0089] Sub-step S203: Construct a multi-source data synchronous acquisition time sequence

[0090] To ensure that valve actuation, water level response, and flow meter readings are aligned on the timeline, the central processing unit (CPU) establishes a unified system clock reference. The CPU uses hardware trigger signals or synchronous polling commands to ensure that these three sets of critical data are captured at the same sampling moment.

[0091] The three sets of key data specifically include:

[0092] The first set of data: feedback data on the actual valve opening, representing the excitation source of the system, denoted as... ;

[0093] The second set of data: water level data in the clear water tank characterizing the system's volumetric reference response, denoted as... ;

[0094] The third set of data: instantaneous flow rate data of the outlet flowmeter characterizing the reading of the object to be calibrated, denoted as... .

[0095] Sampling frequency of synchronous acquisition The Nyquist sampling theorem must be satisfied, and to ensure the accuracy of subsequent differential operations and phase analysis, the sampling frequency must be... It needs to be set as the target calibration frequency. More than 10 times (i.e.) ≥10 For standard applications, the sampling frequency is typically set to 1Hz to 10Hz. The central processing unit assigns a unique timestamp to each frame of acquired data, forming a time-series dataset.

[0096] Sub-step S204: Verify data validity and store it

[0097] The central processing and control unit continuously executes excitation and acquisition actions and monitors the length of the acquired data sequence in real time. To ensure sufficient frequency resolution of the Discrete Fourier Transform in steps S4 and S5, the acquisition duration must include an integer number of excitation cycles, and the total number of cycles is recommended to be no less than 10. The constraints on the acquisition duration are as follows:

[0098] ;

[0099] In the formula, This represents the total duration of data collection. The number of complete excitation cycles, taken as an integer. ; The target calibration frequency is set.

[0100] Once the acquisition duration meets the above conditions, the central processing control unit stops the excitation and restores the outlet regulating valve to the constant opening control mode. The central processing control unit stores the three sets of time-series data with timestamps into the data storage unit, which serves as the original data source for signal processing in step S3 and reference reconstruction in step S4.

[0101] If a step change or loss of valve opening feedback data is detected during the data acquisition process, the central processing control unit will mark the data as invalid and trigger a re-acquisition process.

[0102] This implementation method establishes a reference signal of known frequency in the fluid system by applying a precisely controlled fixed-frequency disturbance and synchronously acquiring data from valves, water levels, and flow meters. This helps to eliminate time phase errors caused by equipment communication delays, ensuring the correspondence between water level changes and flow meter readings in the time domain, and providing a reliable data foundation for subsequent calibration based on frequency domain feature comparison.

[0103] Regarding the hydraulic gain normalization and multi-source signal timing decoupling based on the orifice outflow principle of fluid mechanics in step S3, the specific implementation process of step S3 mainly includes: sub-step S301 of extracting quasi-static water level trend, sub-step S302 of constructing equivalent linear excitation signal, sub-step S303 of calculating fluid transmission lag time, and sub-step S304 of performing signal timing alignment.

[0104] Detailed explanation is as follows:

[0105] Sub-step S301: Extract quasi-static water level trend

[0106] The data processing unit receives the water level data of the clear water tank stored in step S2. Since the water level data in the clear water tank contains both slow trend changes caused by the difference in inflow and outflow rates and minute fluctuations caused by valve sweep frequency actions, the data processing unit needs to separate the quasi-static component representing pressure potential energy.

[0107] The data processing unit uses a low-pass filtering algorithm to process the water level data of the clear water tank. The preferred low-pass filtering algorithm is a moving average filter, with the filter window length set to an integer multiple of the excitation period. The formula for calculating the quasi-static water level trend is as follows:

[0108] ;

[0109] In the formula, For a moment The quasi-static water level trend value; (Time) The previous Original water level data of the clear water pool at each sampling point; The number of sampling points contained in the sliding window, and satisfy (in It is a positive integer. Sampling frequency, (for target calibration frequency). This represents the sampling time interval.

[0110] The data processing unit, through the above calculations, filters out high-frequency fluctuation components to obtain the quasi-static water level trend reflecting the static pressure difference across the valve at the current moment. Quasi-static water level trend value. It is used as a time-varying parameter in subsequent hydraulic model corrections.

[0111] Sub-step S302: Constructing an equivalent linear excitation signal

[0112] Based on the orifice outflow principle of fluid mechanics, the data processing unit feeds back the actual valve opening data collected in step S2. Nonlinear corrections are performed. In the orifice outflow model, the flow rate is proportional to the valve's flow area and the square root of the pressure difference across the valve. To eliminate the influence of water level changes on the system flow gain, the data processing unit dynamically weights the actual valve opening feedback data using the square root of the quasi-static water level trend. The data processing unit calculates the equivalent linear excitation signal using the following formula:

[0113] ;

[0114] In the formula, For a moment The equivalent linear excitation signal; This refers to the feedback data of the actual valve opening collected in step S2; This refers to the quasi-static water level trend calculated in sub-step S301.

[0115] Through this step, the data processing unit transforms the valve opening signal, which is affected by water level, into a normalized signal that is linearly related to the theoretical flow rate, thus correcting the... The readings of the outlet flow meter meet the prerequisites for linear system analysis.

[0116] Sub-step S303: Calculate fluid transport lag time

[0117] The data processing unit calculates the physical transmission delay between the equivalent linear excitation signal and the instantaneous flow data from the outlet flow meter. Because fluid travels through the pipeline network, the outlet flow meter's response lags behind the valve's action. The data processing unit processes the equivalent linear excitation signal... Instantaneous flow data from the outlet flow meter Perform cross-correlation analysis. The formula for calculating the cross-correlation function is as follows:

[0118] ;

[0119] In the formula, The lag time is Cross-correlation coefficients at different times; For a moment The equivalent linear excitation signal; This refers to the instantaneous flow rate data from the outlet flow meter. This represents the total duration of the data sequence.

[0120] The data processing unit searches for the cross-correlation coefficient within a preset lag search interval (e.g., 0 to 10 seconds). Reaching the maximum value The data processing unit will maximize the cross-correlation coefficient. The value is labeled as the physical lag time of the system, denoted as . Physical lag time It reflects the average time required for the pressure wave to travel from the outlet regulating valve to the outlet flow meter.

[0121] Sub-step S304: Perform signal timing alignment

[0122] The data processing unit uses the calculated physical lag time The multi-source signals are time-series reassembled. To eliminate the impact of phase errors on subsequent frequency domain comparison, the data processing unit processes the instantaneous flow data from the outlet flowmeter. Shift forward on the timeline So that it is similar to the equivalent linear excitation signal Alignment is performed in phase. The expression for the flow meter signal after timing alignment is as follows:

[0123] ;

[0124] In the formula, This is the time-aligned outlet flow meter flow data; This is the instantaneous flow rate data from the original outlet flow meter; This refers to the physical lag time determined in sub-step S303.

[0125] The data processing unit will convert the equivalent linear excitation signal Outlet flow meter flow data aligned with time sequence The data are combined into synchronized data pairs. These synchronized data pairs remove the nonlinear effects of water level changes and the time delays of fluid transport, providing a synchronized and linearized data foundation for the accurate extraction of frequency amplitude features in steps S4 and S5.

[0126] For step S4, which involves reconstructing the flow reference and calculating the calibration coefficient based on the frequency domain differential principle, the specific implementation process of step S4 mainly includes: sub-step S401 of obtaining the geometric parameters of the clear water tank, sub-step S402 of extracting the fundamental amplitude of the water level, sub-step S403 of reconstructing the amplitude of the flow reference, and sub-step S404 of calculating the calibration coefficient of the flow meter.

[0127] Detailed explanation is as follows:

[0128] Sub-step S401: Obtain the geometric parameters of the clear water tank

[0129] The data processing unit retrieves pre-stored geometric dimension data of the clear water tank. This data is used to establish a mapping relationship between changes in water level and changes in water storage capacity.

[0130] For conventional rectangular or cylindrical clear water tanks, the data processing unit reads the effective cross-sectional area parameter of the tank. If the cross-sectional area of ​​the clear water tank changes with the water level, the data processing unit needs to calculate the quasi-static water level trend obtained in step S301. The instantaneous cross-sectional area corresponding to the current working water level is determined by a lookup table method or interpolation method. In this embodiment, the cross-sectional area of ​​the clear water tank within the working water level range is set as a constant, denoted as . (Unit: m2)

[0131] Sub-step S402: Extract the water level baseline wave amplitude value

[0132] The data processing unit processes the detrended water level data of the clear water tank stored in step S2. Perform a Discrete Fourier Transform (DFT) to obtain the water level spectrum. The data processing unit accurately extracts the target calibration frequency from the spectrum. The amplitude of the single-sided spectrum at that point. The formula for calculating the amplitude of the fundamental wave at water level is as follows:

[0133] ;

[0134] In the formula, Target calibration frequency The physical amplitude of water level fluctuation at the location; The total number of data points involved in the transformation; For the first Water level data of the clear water pool at each sampling point; The spectral line index number in the spectrum corresponding to the target calibration frequency; It is the imaginary unit.

[0135] Sub-step S403: Reconstruct the flow reference amplitude

[0136] The data processing unit calculates the flow rate reference amplitude based on the principle of fluid continuity and the frequency domain differential characteristics. In the frequency domain, differentiating with respect to time is equivalent to multiplying by the angular frequency factor. The data processing unit utilizes the physical amplitude of water level fluctuations and the cross-sectional area of ​​the clear water tank. and target calibration frequency The actual flow rate fluctuation amplitude caused by valve disturbance is reconstructed. The formula for calculating the flow rate reference amplitude is as follows:

[0137] ;

[0138] In the formula, The reconstructed flow reference amplitude (unit: m) 3 / s, which is the physical true value of the actual flow fluctuation through the outlet regulating valve. Calibrate the target frequency; This represents the effective cross-sectional area of ​​the clear water pool. The physical amplitude of water level fluctuation is extracted in sub-step S402.

[0139] Sub-step S404: Calculate the flow meter calibration coefficient

[0140] The data processing unit processes the time-aligned outlet flow meter flow data output in step S3. Perform spectrum analysis. The data processing unit uses the same discrete Fourier transform algorithm as sub-step S402 to extract... At the target calibration frequency The single-sided spectral amplitude at point is denoted as .

[0141] The data processing unit compares the flow reference amplitude with the measured amplitude of the outlet flow meter to calculate the calibration coefficient of the flow meter. The calibration coefficient reflects the gain error of the flow meter under calibration at the current operating point. The formula for calculating the calibration coefficient is as follows:

[0142] ;

[0143] In the formula, For flow meter calibration coefficient; The reconstructed flow reference amplitude in sub-step S403; This represents the physical amplitude of the outlet flow meter reading after timing alignment.

[0144] like If the value is not equal to 1, the data processing unit will generate a correction command. The correction command is used to update the internal coefficient of the flow meter or to compensate for subsequent flow measurement values ​​in the host computer system.

[0145] This implementation combines frequency domain differentiation with the principle of fluid continuity to establish a mathematical tracing link from water level fluctuations to flow rate fluctuations. Extracting the physical amplitude at specific frequency points helps avoid noise amplification problems caused by direct time-domain differentiation and enables online quantification of the outlet flowmeter gain error. The final calibration coefficients are... It can be used to correct the metering deviation of flow meters, thereby achieving in-situ high-precision calibration of large-diameter flow meters without interrupting water supply services and without the need for external standard flow meters.

[0146] The specific implementation process of step S5, which involves establishing a gain error calibration model based on frequency sweep characteristics and optimizing parameters, mainly includes: sub-step S501 of establishing a calibration coefficient mapping model, sub-step S502 of performing model parameter fitting, and sub-step S503 of storing calibration model parameters.

[0147] Detailed explanation is as follows:

[0148] Sub-step S501: Establish a calibration coefficient mapping model

[0149] The data processing unit constructs a functional relationship between the calibration coefficient and the valve opening reference value based on the flow characteristics of the outlet regulating valve. The valve opening reference value is the sweep frequency center opening value set in step S2. The data processing unit uses a polynomial model as the mapping model.

[0150] This embodiment uses a third-order polynomial model to fit the calibration coefficients. Valve opening reference value The relationship between them. The expression for the calibration coefficient mapping model is as follows:

[0151] ;

[0152] In the formula, The valve opening reference value is The predicted value of the calibration coefficient at that time; This is the reference value for valve opening. , , , These are the model coefficients to be determined.

[0153] Sub-step S502: Perform model parameter fitting

[0154] The data processing unit uses multiple sets of data pairs containing valve opening reference values ​​and corresponding calibration coefficients obtained under different sweep frequency opening reference values ​​in steps S2 to S4 to perform model parameter fitting.

[0155] The data processing unit employs the least squares method as the parameter fitting algorithm. The goal of parameter fitting is to find an optimal set of polynomial model coefficients that minimizes the sum of squared deviations between the actual calculated values ​​of the calibration coefficients at all test points and the model predictions calculated based on this set of coefficients. The calculation formula is as follows:

[0156] ;

[0157] In the formula, The total squared error of the model error function; This represents the total number of calibration coefficient sets under different aperture reference values ​​that have been collected. For the first The calibration coefficients actually calculated for each test point; For the first The calibration coefficients for each test point are predicted by the mapping model.

[0158] The data processing unit minimizes Solve for the optimal model parameter set. .

[0159] Sub-step S503: Store calibration model parameters

[0160] The data processing unit will fit the optimal model parameter set. It is stored in the non-volatile memory of the data processing unit.

[0161] Optimal model parameter set This constitutes the flowmeter gain error calibration model. When the system is in operation, the data processing unit reads the current valve opening command or feedback value as... Calculate the calibration coefficient at the current moment. It also performs real-time compensation on the instantaneous measurement value of the outlet flow meter. The real-time corrected flow measurement value... The calculation formula is as follows:

[0162] ;

[0163] In the formula, For a moment The real-time corrected flow measurement value; These are the calibration coefficients calculated from the flowmeter gain error calibration model; This is the instantaneous measurement value of the outlet flow meter.

[0164] This implementation fits discrete calibration coefficient data points into a continuous mathematical model. By establishing a functional relationship between the calibration coefficients and the valve opening reference value, the data processing unit can calculate the flow meter's gain error under any operating conditions, achieving full-range, online calibration of large-diameter flow meters.

[0165] Reference Appendix Figure 2 To support the logical execution of the above method flow, this invention also provides a system for calibrating an outlet flow meter using a clear water tank level, comprising:

[0166] Physical layer architecture: The components of the physical layer are coupled sequentially along the fluid flow direction, forming the hardware carrier of the calibration method.

[0167] Clear water tank (reference source): Serves as the physical reference for volumetric measurement. Its inlet is connected to a continuously operating inlet pipeline, allowing for dynamic water intake during calibration.

[0168] High-precision liquid level sensing unit: configured as a liquid level measuring device with high-frequency sampling capability, used to capture the sweep frequency response in step S1 and the minute volume change in step S2;

[0169] The feedback-enabled control unit is configured as an electric control valve installed on the outlet pipeline. It integrates an opening feedback sensor to output the valve core physical position signal in real time, providing real execution end data for the nonlinear decoupling in step S3.

[0170] The metering unit to be calibrated is the outlet flow meter installed downstream of the regulating valve, which outputs instantaneous flow readings as the monitored object.

[0171] Central processing and control unit: Connected to the above units via industrial fieldbus, serving as the core of calibration logic calculation.

[0172] Logical control layer architecture: The central processing and control unit internally operates several logical function modules, achieving a closed loop of method steps S1-S5 through data flow transmission and processing.

[0173] Frequency scanning and optimization module: Configured to execute the characteristic identification logic in step S1. This module includes a signal generation unit and a spectrum analysis unit, used to send frequency conversion commands to the regulating valve, analyze the water level response feedback, automatically lock in a safe frequency that avoids resonance peaks, and pass this frequency as a global reference parameter to the next level module.

[0174] Multi-source excitation and synchronous acquisition module: Configured to execute the data acquisition logic of step S2. This module drives the regulating valve according to a locked safety frequency and controls the multi-channel sampler to ensure that the liquid level, valve feedback, and instrument flow signals are acquired and timestamped at the same clock cycle.

[0175] Hydraulic gain decoupling module: configured to execute the physical error correction logic in step S3. This module embeds a fluid dynamics correction algorithm, receives valve feedback and water level data, uses the square root characteristic of the water level to correct the valve opening, outputs an equivalent linear excitation signal that eliminates nonlinear hydraulic gain error, and calculates the transmission delay between signals.

[0176] Frequency domain calibration calculation module: Configured to execute the feature extraction and comparison logic of steps S4 and S5. This module is responsible for reconstructing the reference flow signal, extracting the spectral amplitude at the characteristic frequency using the FFT algorithm, calculating the calibration coefficient through comparison, and finally sending the correction command to the outlet flow meter through the communication interface.

Claims

1. A method for calibrating an outlet flow meter using the water level of a clear water tank, characterized in that, include: By analyzing the water level response data of the clear water tank through spectrum analysis, a safe frequency that avoids water resonance is automatically locked, and the safe frequency is used as the frequency domain reference for calibration. Based on the safety frequency, the actuator with feedback is driven to perform periodic disturbances, and the water level data of the clear water tank, the valve position signal of the actuator, and the instantaneous flow data of the outlet flow meter are collected simultaneously. The valve position signal is subjected to hydraulic gain nonlinear decoupling, and the nonlinear effect of the orifice outflow gain is eliminated by the quasi-static component in the water level data, thereby generating an equivalent linear excitation signal. Based on the physical time lag between the equivalent linear excitation signal and the instantaneous flow data, the instantaneous flow data is time-aligned to obtain the aligned flow signal. By utilizing the principle of fluid continuity and the frequency domain differential characteristics, the spectral amplitude of the water level data is reconstructed into the flow reference amplitude, and the flow meter calibration coefficient under the current operating conditions is calculated by comparing the flow reference amplitude with the measured spectral amplitude of the aligned flow signal. Based on the flow meter calibration coefficients under multiple operating conditions, a flow meter gain error calibration model is generated by fitting and online compensation is performed on the flow data of the outlet flow meter to complete the calibration.

2. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 1, characterized in that, The automatic locking of a safe frequency to avoid water resonance specifically includes: Send a frequency conversion scanning command to the feedback-enabled control unit to drive the valve to perform a frequency sweeping action within a preset frequency range; Perform a discrete Fourier transform on the collected water level data of the clear water pool to obtain the water level amplitude-frequency characteristic curve; Search for the minimum amplitude point in the water level amplitude-frequency characteristic curve, and determine the frequency corresponding to the minimum point as the safe frequency.

3. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 1, characterized in that, The generation of the equivalent linear excitation signal and the acquisition of the aligned flow signal specifically include: The water level data is subjected to low-pass filtering to separate the quasi-static water level trend that reflects changes in water storage. Based on the orifice outflow principle of fluid mechanics, the valve position signal is weighted and corrected using the reciprocal of the square root of the quasi-static water level trend to eliminate the nonlinear influence of water level changes on the valve flow characteristics and generate the equivalent linear excitation signal. The time delay between the equivalent linear excitation signal and the instantaneous flow data is calculated using a cross-correlation analysis algorithm, and the time delay is used as the physical lag time.

4. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 3, characterized in that, The specific calculation logic for generating the equivalent linear excitation signal is as follows: Obtain the square root value of the quasi-static water level trend, divide the valve position signal output by the feedback-enabled execution control unit by the square root value, and define the resulting quotient as the equivalent linear excitation signal.

5. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 1, characterized in that, The calculation of the flow meter calibration coefficient under the current operating conditions specifically includes: Extract the physical amplitude of the water level fluctuation at the safe frequency from the water level data; Using the frequency domain differential principle, the product of the angular frequency corresponding to the safe frequency, the effective cross-sectional area of ​​the clear water pool, and the physical amplitude of the water level fluctuation is calculated, and the product is defined as the flow reference amplitude. Extract the measured physical amplitude of the aligned flow signal at the safe frequency; Calculate the ratio of the flow reference amplitude to the measured physical amplitude, and determine the ratio as the flow meter calibration coefficient.

6. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 5, characterized in that, The specific process for reconstructing the flow reference amplitude is as follows: The four parameters—twice the value of pi, the safe frequency, the effective cross-sectional area of ​​the clear water pool, and the physical amplitude of the water level fluctuation—are multiplied together, and the resulting product is used as the baseline amplitude of the flow rate.

7. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 1, characterized in that, The fitted flowmeter gain error calibration model includes: A polynomial function is used as the basis function to establish the mapping relationship between the flow meter calibration coefficient and the valve opening reference value; The least squares method was used to perform regression analysis on multiple sets of measured data to find the optimal model parameter set that minimizes the sum of squared prediction residuals of the flowmeter gain error calibration model. The optimal model parameter set is stored in the controller as core parameters.

8. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 7, characterized in that, The valve opening reference value is defined as the center average opening of the feedback-enabled actuator when the periodic disturbance is performed; The flow meter gain error calibration model is used to cover the full stroke opening range of the feedback-enabled actuator, providing continuous error correction coefficients to complete the full-condition calibration of the outlet flow meter.

9. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 1, characterized in that, The implementation of online compensation specifically includes: In normal water supply mode, the current opening value of the execution regulation unit with feedback is collected in real time; Substitute the current opening value into the flow meter gain error calibration model to calculate the instantaneous calibration coefficient; The original flow reading output by the outlet flow meter is multiplicatively corrected using the instantaneous calibration coefficient to generate the final metered flow data.

10. The method for calibrating an outlet flow meter using the water level of a clear water tank according to claim 1, characterized in that, The synchronous acquisition of water level data from the clear water tank, valve position signals from the regulating unit, and instantaneous flow data from the outlet flow meter specifically involves: The central processing unit controls the multi-channel sampler to ensure that the three signals can be collected and timestamped at the same clock beat.

Images