A flow velocity measuring system for a propeller type hydraulic current meter

Abstract

This invention discloses a flow velocity measurement system for a turbine-type hydraulic current meter, belonging to the field of hydrological monitoring technology. The system includes: acquiring multi-source sensor data through a multi-source signal acquisition module; fusing the multi-source sensor data through a feature fusion processing module to generate fused flow velocity features; adaptively adjusting the fused flow velocity features based on a historical flow velocity dataset through an adaptive calibration module to generate a calibrated flow velocity value; dynamically compensating for errors in the calibrated flow velocity value based on the multi-source sensor data through an error dynamic compensation module to generate a final flow velocity value; and finally, appending the final flow velocity value to the historical flow velocity dataset by a dataset update module to generate an updated historical flow velocity dataset for the next adjustment. This invention employs a combined technical solution of multi-source signal acquisition, feature fusion processing, adaptive calibration, and dynamic error compensation, enabling high-precision, adaptive, and robust flow velocity measurement.

Description

Technical Field

[0001] This invention relates to the field of hydrological monitoring technology, and in particular to a flow velocity measurement system for a turbine-type hydraulic flow meter. Background Technology

[0002] The impeller-type current meter is a fundamental measuring instrument widely used in hydrology, hydraulic engineering, and environmental monitoring. Its core function is to determine the flow velocity of water by measuring the speed at which the impeller rotates. This instrument converts the kinetic energy of the water flow into the rotational mechanical energy of the impeller, and then into a measurable electrical signal, providing crucial basic data for water resource assessment, flood control and disaster reduction, and water environment research.

[0003] In existing technologies, traditional impeller-type hydraulic velocity meters typically contain only one impeller sensor, which converts the impeller's rotational motion into a series of electrical pulse signals. The subsequent measurement circuit counts the number of pulses per unit time to obtain the impeller's rotational frequency. Finally, the system directly converts the measured rotational frequency into a flow velocity value based on a fixed functional relationship calibrated in a standardized experimental water tank. Once determined, this calibration relationship usually remains unchanged throughout the instrument's service life.

[0004] However, the aforementioned existing technical solutions have significant limitations in practical applications. First, their fixed calibration formulas are based on ideal, stable laminar flow conditions, while actual water flow in nature often involves complex flow states such as turbulence and vortices. This causes the impeller's response characteristics to deviate from the ideal model, introducing measurement errors. Second, this solution neglects the influence of water physical properties such as density and viscosity on the impeller's hydrodynamic characteristics due to temperature variations. Furthermore, during long-term use, wear on the impeller bearings and the adhesion of aquatic organisms or sediment can alter the instrument's mechanical properties, causing the initial calibration relationship to fail and resulting in measurement drift. Existing systems lack the ability to adaptively adjust to such changes. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a flow velocity measurement system for impeller-type hydraulic flow meters. It employs a technical solution that combines multi-source signal acquisition, feature fusion processing, adaptive calibration, and dynamic error compensation, enabling high-precision, adaptive, and robust measurement of flow velocity.

[0006] The above objectives can be achieved through the following approach:

[0007] A flow velocity measurement system for a turbine-type hydraulic current meter includes: a multi-source signal acquisition module for acquiring pulse signals generated by a turbine sensor, pressure signals generated by a pressure sensor, and temperature signals generated by a temperature sensor, generating multi-source sensor data; a feature fusion processing module for fusing the multi-source sensor data to generate fused flow velocity features; an adaptive calibration module for adaptively adjusting the fused flow velocity features based on a historical flow velocity dataset to generate a calibrated flow velocity value; an error dynamic compensation module for dynamically compensating for errors in the calibrated flow velocity value based on the multi-source sensor data, generating a final flow velocity value; and a dataset update module for appending the final flow velocity value to the historical flow velocity dataset to generate an updated historical flow velocity dataset for the next adjustment.

[0008] Optionally, the multi-source signal acquisition module includes: a signal sampling unit for synchronously acquiring pulse signals, pressure signals, and temperature signals to generate an original signal set; an analog-to-digital conversion unit for performing analog-to-digital conversion on the original signal set to generate a digital signal set; and a digital signal filtering unit for filtering the digital signal set to generate multi-source sensor data.

[0009] Optionally, the feature fusion processing module includes: a multi-source signal decoupling unit, used to extract filtered pressure signal, filtered temperature signal, and filtered pulse signal from the multi-source sensor data respectively; a turbulence feature analysis unit, used to calculate the turbulence intensity of the water flow based on the filtered pressure signal and generate turbulence intensity parameters; a fluid property correction unit, used to calculate the water density correction factor based on the filtered temperature signal and generate density correction parameters; a pulse signal compensation unit, used to combine the turbulence intensity parameters and the density correction parameters to compensate the filtered pulse signal and generate compensated pulse data; and a velocity feature extraction unit, used to extract features from the compensated pulse data and generate fused velocity features.

[0010] Optionally, the feature extraction of the compensation pulse data includes: calculating the average value of the compensation pulse data to generate an average pulse value; calculating the variance of the compensation pulse data to generate a pulse variance; and combining the average pulse value and the pulse variance to generate a fused flow velocity feature.

[0011] Optionally, the adaptive calibration module includes: an online learning calibration unit, used to calculate adaptive calibration parameters based on the historical flow velocity dataset and the fused flow velocity features; and a flow velocity value calculation unit, used to apply the adaptive calibration parameters to adjust the fused flow velocity features and generate a calibrated flow velocity value.

[0012] Optionally, the calculation of adaptive calibration parameters includes: generating a feature deviation by calculating the difference between the historical flow velocity dataset and the fused flow velocity features; and adjusting the initial calibration coefficient proportionally based on the feature deviation to generate adaptive calibration parameters.

[0013] Optionally, the error dynamic compensation module includes: a working condition compensation unit, used to perform weighted summation and polynomial transformation on the filtered temperature signal and the filtered pressure signal to generate an error compensation value; a dynamic correction unit, used to perform arithmetic superposition of the error compensation value and the calibrated flow rate value to generate a compensated flow rate value; and a flow rate filtering unit, used to perform moving average filtering on the compensated flow rate value to generate a final flow rate value.

[0014] Optionally, the dataset update module includes: a validity verification unit, used to verify the validity of the final flow rate value and generate valid flow rate data; a time-series association storage unit, used to associate the valid flow rate data with the corresponding collection timestamp and store it in the historical flow rate dataset to generate a historical flow rate dataset with time-series tags; and a timeliness filtering unit, used to perform timeliness filtering processing on the historical flow rate dataset with time-series tags to generate an updated historical flow rate dataset.

[0015] Optionally, the validity verification of the final flow velocity value includes: calculating the deviation between the final flow velocity value and the statistical feature value of the historical flow velocity dataset to generate a historical deviation; comparing the historical deviation with a preset confidence interval threshold; and determining the flow velocity data as valid when the historical deviation is within the range of the confidence interval threshold.

[0016] Based on the same inventive concept, this invention also provides a flow velocity measurement method for a turbine-type hydraulic flow meter, comprising: acquiring pulse signals generated by a turbine sensor, pressure signals generated by a pressure sensor, and temperature signals generated by a temperature sensor to generate multi-source sensor data; performing fusion processing on the multi-source sensor data to generate fused flow velocity features; adaptively adjusting the fused flow velocity features based on a historical flow velocity dataset to generate a calibrated flow velocity value; performing dynamic error compensation on the calibrated flow velocity value based on the multi-source sensor data to generate a final flow velocity value; and appending the final flow velocity value to the historical flow velocity dataset to generate an updated historical flow velocity dataset for the next adjustment.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] This invention improves the overall performance of flow velocity measurement by fusing information from multiple sensors and combining it with intelligent algorithms. The system incorporates real-time pressure and temperature data into the flow velocity calculation model, providing physical-level compensation for nonlinear errors caused by changes in water flow turbulence characteristics and fluid physical properties, thereby improving the accuracy of measurement results in complex and variable hydrological environments.

[0019] The adaptive calibration and dynamic dataset update mechanism designed in this invention endows the system with the ability to learn and optimize online. Based on long-term accumulated valid measurement data, the system can continuously adjust its internal calibration parameters, effectively combating systematic drift caused by factors such as instrument aging, mechanical wear, or gradual environmental changes. This ensures the long-term stability and reliability of the measurement system throughout its entire lifecycle, reducing the frequency of manual maintenance and recalibration.

[0020] This invention constructs a multi-level error correction system integrating feature fusion, adaptive calibration, and dynamic error compensation, and ensures the quality of learning samples through a data validity verification mechanism. This multi-layered protection and closed-loop feedback design enhances the robustness of the entire measurement system, enabling it to effectively suppress various systematic errors and random interferences, and ensuring the output of stable and reliable flow velocity data under various operating conditions.

[0021] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a framework diagram of a flow velocity measurement system for an impeller-type hydraulic flow meter according to an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram of the structure of a flow velocity measurement system for an impeller-type hydraulic flow meter according to an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram illustrating the principle of validating the final flow rate value in an embodiment of the present invention.

[0026] Figure 4This is a schematic flowchart of a flow velocity measurement method for an impeller-type hydraulic flow meter according to an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Reference Figure 1 One embodiment of the present invention proposes a flow velocity measurement system for a turbine-type hydraulic flow meter, which adopts a technical solution combining multi-source signal acquisition, feature fusion processing, adaptive calibration and dynamic error compensation, and can achieve high-precision, adaptive and robust measurement of flow velocity.

[0029] The system described in this embodiment specifically includes:

[0030] S1, Multi-source signal acquisition module, used to acquire pulse signals generated by impeller sensor, pressure signals generated by pressure sensor and temperature signals generated by temperature sensor, and generate multi-source sensor data;

[0031] Optionally, the multi-source signal acquisition module includes:

[0032] The signal sampling unit is used to synchronously acquire pulse signals, pressure signals, and temperature signals to generate a raw signal set.

[0033] An analog-to-digital conversion unit is used to perform analog-to-digital conversion on the original signal set to generate a digital signal set;

[0034] A digital signal filtering unit is used to filter the digital signal set to generate multi-source sensor data.

[0035] Specifically, multi-source signals are first synchronously acquired through a signal sampling unit. This unit integrates signal conditioning circuits connected to the impeller sensor, pressure sensor, and temperature sensor, respectively, and is controlled by a unified clock source using a multi-channel synchronous sample-and-hold circuit. When the clock signal is triggered, the sample-and-hold circuit simultaneously locks the instantaneous values ​​of the analog signals output by the three sensors: the pulse signal output by the impeller sensor, the pressure signal representing the static and dynamic pressure of the water body output by the pressure sensor, and the temperature signal reflecting the water temperature output by the temperature sensor. This synchronous sampling mechanism ensures that all signals acquired at the same time correspond to the same water flow state, avoiding data mismatch problems caused by time deviations, thereby generating a raw signal set containing three time-aligned analog signal waveforms. The sampling frequency setting must satisfy the Nyquist sampling theorem to ensure that the effective frequency components in the signal can be reproduced without distortion, especially the turbulent fluctuation information that may be contained in the pressure signal. Next, the analog-to-digital conversion unit digitizes the raw signal set. This unit employs a high-precision, high-speed analog-to-digital converter (ADC). After the synchronous sample-and-hold latch signal is engaged, it sequentially or in parallel converts the analog voltage value of each channel into a binary digital quantity with a specific width, such as 16 bits or 24 bits. After conversion, a digital signal set consisting of a series of discrete time-point digital sequences is generated, which reproduces the original analog signal waveform in the digital domain. Finally, a digital signal filtering unit filters the digital signal set to eliminate various noises introduced during the acquisition process. Appropriate digital filtering algorithms are used for different signal characteristics. For example, for pressure and temperature signals, which typically exhibit relatively smooth changes superimposed with high-frequency noise, a moving average filtering algorithm is used for smoothing. The calculation method for the moving average filter is as follows:

[0036] ,

[0037] in, This represents the filtered signal value output at the current time n. This represents the raw digital signal values ​​acquired at the current time and the N-1 time points prior, where N is the width of the filtering window, an integer preset based on noise levels and signal response speed requirements. For the pulse signals generated by the impeller sensor, an algorithm with edge detection and digital stabilization is used to filter out glitches and spurious pulses, ensuring accurate pulse counting. After filtering, multi-source sensor data with improved signal-to-noise ratio and cleaner data is obtained and output to the feature fusion processing module.

[0038] For example, consider a flow velocity monitoring device installed in the tailrace of a large hydropower station. This device monitors the flow rate released after power generation. The device's multi-source signal acquisition module operates as follows: First, its internal signal sampling unit synchronously acquires signals. For instance, on a day when the hydropower station units are operating at full capacity and the water flow is turbulent, a synchronous sample-and-hold circuit controlled by a unified clock source simultaneously locks the analog signal outputs of the impeller sensor, pressure sensor, and temperature sensor. The impeller sensor outputs a dense pulse signal due to the impact of the high-speed water flow; the pressure sensor outputs a pressure signal representing 150 kPa due to the deep water level and high-speed dynamic pressure; and the temperature sensor outputs a temperature signal reflecting the current water temperature as 18 degrees Celsius. These three signals, captured at the same instant, constitute the original signal set for that moment. Then, the analog-to-digital converter digitizes the analog signals in this set. For example, the 150 kPa pressure signal is converted into a digital value of 456789 by a 24-bit ADC, and the 18-degree Celsius temperature signal is converted into 23456. This process repeats 1000 times per second, generating a set of digital signals. Finally, the digital signal filtering unit processes this set. Considering the high-frequency electronic noise interference in the tailrace channel, the system employs a moving average filtering algorithm for the pressure and temperature signals. Taking the pressure signal as an example, if the digital values ​​of four consecutive sampling points are [456799, 456779, 456795, 456784], and the window width N is set to 4, then the filtered value is...

[0039] The processed, clean data stream, which is the multi-source sensor data, is transmitted to subsequent modules. This provides the raw data foundation for advanced processing steps such as feature fusion, adaptive calibration, and dynamic compensation, ensuring the stability and reliability of the entire measurement system from the source, improving the system's ability to suppress environmental interference, and enhancing the accuracy of the final flow velocity measurement results.

[0040] S2, Feature fusion processing module, used to fuse the multi-source sensor data to generate fused flow velocity features;

[0041] Optionally, the feature fusion processing module includes:

[0042] A multi-source signal decoupling unit is used to extract filtered pressure signal, filtered temperature signal and filtered pulse signal from the multi-source sensor data, respectively.

[0043] The turbulence feature analysis unit is used to calculate the turbulence intensity of the water flow based on the filtered pressure signal and generate turbulence intensity parameters.

[0044] The fluid property correction unit is used to calculate the water density correction factor based on the filtered temperature signal and generate density correction parameters.

[0045] The pulse signal compensation unit is used to compensate the filtered pulse signal by combining the turbulence intensity parameter and the density correction parameter to generate compensated pulse data;

[0046] The flow velocity feature extraction unit is used to extract features from the compensation pulse data and generate fused flow velocity features.

[0047] Specifically, firstly, the multi-source signal decoupling unit analyzes and separates the multi-source sensor data received from the multi-source signal acquisition module. The mixed data stream is decomposed into three independent digital signal sequences: filtered pressure signal, filtered temperature signal, and filtered pulse signal. Each signal sequence retains its original timestamp information to ensure time synchronization in subsequent processing. Secondly, the turbulence characteristic analysis unit uses the filtered pressure signal to quantify the turbulence characteristics of the water flow. Turbulence in the water flow causes severe fluctuations in the torque acting on the impeller, resulting in a nonlinear relationship between the impeller speed and the average flow velocity, which is one of the main sources of error in traditional flow meters. This unit calculates the pulsation intensity by statistically analyzing the filtered pressure signal sequence within a time window. Specifically, the standard deviation of the pressure signal within this time window is calculated to characterize the magnitude of the pressure pulsation, and it is compared with the average pressure to generate a dimensionless turbulence intensity parameter. The calculation formula is as follows:

[0048] ,

[0049] in, These are the generated turbulence intensity parameters. It is obtained by calculating the standard deviation of the filtered pressure signal sequence within the time window, representing the pressure fluctuation range. This is the arithmetic mean of the filtered pressure signals within the time window, representing the average water pressure. This turbulence intensity parameter quantitatively describes the degree of turbulence in the water flow. Furthermore, the fluid property correction unit corrects the physical properties of the water based on the filtered temperature signal. Water density is a key physical parameter affecting the hydrodynamic response of the impeller, and density changes with temperature. This unit calculates the actual water density at the current temperature using a built-in water temperature-density relationship model, typically based on a polynomial function fitted to standard experimental data or a lookup table method, and compares it with a preset reference density to generate a density correction parameter. The calculation formula is as follows:

[0050] ,

[0051] in, These are the generated density correction parameters. The actual water density is calculated based on the filtered temperature signal. This is a reference water density value under standard conditions, such as the water density at a specific temperature. This density correction parameter reflects the degree of deviation in water density caused by temperature changes. Then, the pulse signal compensation unit performs signal fusion and compensation operations. This unit comprehensively utilizes the turbulence intensity parameter and the density correction parameter to dynamically compensate the filtered pulse signal. The specific compensation process is represented by a compensation function that takes the frequency of the filtered pulse signal as input and adjusts it in conjunction with the turbulence intensity parameter and the density correction parameter to generate compensated pulse data. The compensation model is as follows:

[0052] ,

[0053] in, It is the compensated pulse frequency, i.e., the compensated pulse data. It is the frequency of the original filtered pulse signal. For turbulence intensity parameters, For density correction parameters, The turbulence compensation coefficient characterizes the sensitivity of impeller speed to turbulence intensity. It is calibrated experimentally. The density compensation coefficient, characterizing the impeller rotation speed's sensitivity to changes in water density, is calibrated experimentally. Finally, the velocity feature extraction unit processes the compensated pulse data to extract stable and representative velocity features. Within a measurement cycle, the compensated pulse frequency may still fluctuate. This unit performs statistical calculations on the compensated pulse data sequence during this period to extract its core features, generating the final fused velocity features, which are then output to the adaptive calibration module.

[0054] For example, taking tailrace monitoring at a hydropower station as an example, multi-source sensor data generated by the multi-source signal acquisition module is sent to the feature fusion processing module. First, the multi-source signal decoupling unit separates the mixed data stream into independent filtered pressure signals, filtered temperature signals, and filtered pulse signals. Next, the turbulence feature analysis unit begins operation. Due to the unit's full-load operation, the tailrace turbulence is intense, and the pressure signal fluctuates significantly within a certain time window. This unit calculates the average value of the pressure signal within this window. The corresponding standard deviation is 150 kPa. This corresponds to 3 kPa, representing the amplitude of pressure fluctuations. Turbulence intensity parameters are calculated using the formula. Meanwhile, the fluid property correction unit calculates the water density correction factor based on the filtered temperature signal. The current water temperature is 18 degrees Celsius; the water density at this temperature is determined using the built-in water temperature-density relationship model. The value is 998.60 kg per cubic meter. A reference density is set. The density is 999.97 kg / m³. The calculated density correction parameter is... Then, the pulse signal compensation unit performs crucial compensation. It receives a filtered pulse signal acquired simultaneously with the aforementioned signal, the frequency of which... The frequency is 250 Hz. This is combined with the calculated turbulence intensity parameters. and density correction parameters And using a preset turbulence compensation coefficient and density compensation coefficient The compensation pulse data is calculated using a compensation model. The calculation process is as follows:

[0055] Hertz. Finally, the velocity feature extraction unit processes this compensated pulse frequency sequence to extract the fused velocity features for subsequent use. This multi-source information fusion processing method can overcome the inherent defects of traditional impeller-type current meters under complex flow conditions and environmental changes. The generated fused velocity features can reflect the average motion state of the water body, thus laying a data foundation for obtaining velocity measurement results and enhancing the adaptability of the measurement system to complex working conditions and the robustness of the measurement.

[0056] Optionally, the feature extraction of the compensation pulse data includes:

[0057] Calculate the average value of the compensation pulse data to generate an average pulse value;

[0058] Calculate the variance of the compensated pulse data to generate the pulse variance;

[0059] The average pulse value and the pulse variance are combined to generate a fused flow velocity feature.

[0060] Specifically, the flow velocity feature extraction unit receives a sequence of compensated pulse data within a fixed measurement time window from the pulse signal compensation unit. This sequence consists of a series of discrete instantaneous compensated pulse frequency values. First, the average value of this compensated pulse data sequence is calculated to generate an average pulse value. The average pulse value reflects the central trend of the impeller rotation speed within the measurement time window and is a core indicator characterizing the average velocity of the water flow. Its calculation formula is as follows:

[0061] ,

[0062] in, The average pulse value generated, is the i-th instantaneous frequency sample value in the compensated pulse data sequence, and N is the total number of samples collected within this measurement time window. This average pulse value provides a first-order approximation of the flow velocity magnitude. Next, the variance of this compensated pulse data sequence is calculated to generate the pulse variance. Variance measures the dispersion of the compensated pulse frequency values ​​around their average value; it reflects the fluctuations in impeller speed caused by residual pulsations or instabilities in the water flow, even after compensation. Its calculation formula is:

[0063] ,

[0064] in, For the generated pulse variance, It is the instantaneous frequency value in the compensated pulse data sequence. The average pulse value is the calculated value, and N is the total number of samples. This pulse variance, as a second-order statistic of the velocity characteristic, contains important information about flow field stability. Finally, the calculated average pulse value and pulse variance are structurally combined to generate a fused velocity feature. This combination is not a simple arithmetic operation, but rather it combines these two parameters with different physical meanings to form a multi-dimensional feature vector. A two-dimensional vector is constructed. This vector represents the final output fused flow velocity feature. This feature vector will be passed to the subsequent adaptive calibration module for processing.

[0065] For example, consider the compensated pulse data generated during tailrace monitoring at a hydropower station. Within a 1-second measurement period, the flow velocity feature extraction unit receives a series of compensated instantaneous pulse frequency values; this compensated pulse data sequence is... The unit is Hertz. First,

[0066] This unit calculates the average value of the sequence to generate the average pulse value. According to the formula... The total number of samples N is 5.

[0067] achievable Hertz. This value represents the central trend of the compensated impeller speed within that second. Secondly, the variance of the unit calculation sequence is used to generate the pulse variance, reflecting the residual fluctuations in impeller speed under tailrace turbulence. According to the formula... ,

[0068] calculate ,

[0069] ,

[0070] ,

[0071] ,

[0072] .therefore,

[0073] Hertz². Finally, the unit structurally combines the average impulse value and impulse variance into a two-dimensional vector. This vector represents the final fused velocity feature and is transmitted to the adaptive calibration module. This method enriches the connotation of the velocity feature, enabling subsequent calibration and analysis to be based on more comprehensive flow field information, rather than relying solely on an average rotational speed value. It allows the system to distinguish flow fields with the same average velocity but different turbulence characteristics, providing a foundation for achieving higher accuracy and stronger robustness in velocity measurement.

[0074] S3, Adaptive calibration module, used to adaptively adjust the fused flow velocity features based on historical flow velocity dataset to generate calibrated flow velocity values;

[0075] Optionally, the adaptive calibration module includes:

[0076] An online learning calibration unit is used to calculate adaptive calibration parameters based on the historical flow velocity dataset and the fused flow velocity features;

[0077] The flow rate calculation unit is used to adjust the fused flow rate characteristics by applying the adaptive calibration parameters to generate a calibrated flow rate value.

[0078] Specifically, the online learning calibration unit first receives the fused flow velocity features of the current measurement cycle output by the feature fusion processing module. This feature is a feature vector containing the average pulse value and pulse variance. Simultaneously, this unit accesses and analyzes the historical flow velocity dataset stored in the system. The historical flow velocity dataset contains a large number of validated final flow velocity values ​​associated with specific operating conditions. The core task of this unit is to learn and adjust the parameters of an initial or previous calibration model based on this historical experience base, generating a set of adaptive calibration parameters best suited to the current state. This learning process iteratively optimizes the preset mapping function by analyzing the relationship between the current fused flow velocity features and the flow velocity values ​​corresponding to similar features in the historical flow velocity dataset, enabling the function to map the fused flow velocity features to the actual flow velocity. This process is equivalent to allowing the instrument to continuously learn and correct itself during use, adapting to long-term drift in sensor performance or slow changes in the environment. Secondly, the flow velocity value calculation unit applies the adaptive calibration parameters generated by the online learning calibration unit to mathematically transform the current fused flow velocity features, thereby calculating the calibrated flow velocity value. This process is the specific execution step of the calibration model. Assuming the fused flow velocity characteristics include the average pulse value and impulse variance Given a vector and a set of adaptive calibration parameters, the calibration flow rate value is calculated using a multivariate linear model:

[0079] ,

[0080] in, This is the final generated calibration flow rate value. and These are the average pulse value and pulse variance in the input fused flow velocity features, respectively. , and These are a set of adaptive calibration parameters calculated in real time by the online learning calibration unit. They represent the weighting coefficients and basic offsets of each feature component, and have corresponding physical dimensions to ensure that the equation holds true. This solution process will transform the feature quantities, which have undergone multi-source information compensation and carry the characteristics of flow field fluctuations, into a preliminary flow velocity estimate, i.e., the calibration flow velocity value, through a dynamically optimized model, and output it to the error dynamic compensation module.

[0081] For example, consider a monitoring device that has been operating continuously for a year in the tailrace of a hydropower station. Due to long-term erosion by sediment-laden water, the impeller surface has experienced slight wear, resulting in a decrease in its response efficiency compared to when it was manufactured. In this situation, the adaptive calibration module plays a crucial role. First, the online learning calibration unit receives the fused flow velocity characteristics generated during the current measurement cycle. Simultaneously, the unit retrieves and analyzes a historical flow rate dataset stored in system memory, containing tens of thousands of verified flow rate records from the past year. Using an online learning algorithm, the unit analyzes the relationship between current characteristics and historical data, discovering that for an average pulse value of 274.58 Hz, historical flow rates were generally higher than predicted by the initial calibration model, a clear indication of impeller wear. Therefore, the unit calculates a new set of adaptive calibration parameters, such as... (meters per second) / Hertz = -0.25 (m / s) / Hertz², meters per second—the dimensions of these parameters ensure the physical meaning of the calculations. Subsequently,

[0082] The velocity calculation unit applies this new set of parameters to adjust the fused velocity characteristics, generating calibrated velocity values. Based on the model... ,calculate

[0083] The calibrated flow rate is measured in meters per second. This calibrated flow rate value is then output to the next module. Through the adaptive calibration module, this system overcomes the limitations of traditional flowmeters that use fixed, offline calibration curves. This module utilizes the instrument's historical measurement data to build an online learning mechanism, enabling it to adjust and optimize its core flow rate calculation model.

[0084] Optionally, the calculation of adaptive calibration parameters includes:

[0085] A feature deviation is generated by calculating the difference between the historical flow velocity dataset and the fused flow velocity features;

[0086] The initial calibration coefficients are adjusted proportionally based on the aforementioned characteristic deviation to generate adaptive calibration parameters.

[0087] Specifically, firstly, a feature deviation is generated by correlating the historical velocity dataset with the fused velocity features. It's important to clarify that the historical velocity dataset contains the final velocity value measured in velocity units, while the fused velocity features are feature vectors containing average pulse values ​​and pulse variance. Since their dimensions are different, they cannot be directly interpolated. Therefore, the calculation process here is as follows: First, using the calibration coefficients from before the current moment (i.e., the initial calibration coefficients), a predictive solution is performed on the currently acquired fused velocity features to obtain a predicted velocity value. Then, one or more verified final velocity values ​​that are close to the current operating conditions or are the closest in time are extracted from the historical velocity dataset as a reference benchmark. The feature deviation is the difference between this reference benchmark velocity value and the predicted velocity value. Its calculation formula is as follows:

[0088] ,

[0089] in, It is the generated characteristic deviation. These are reference flow rate values ​​obtained from historical flow rate datasets. and These are the two components of the current fused flow velocity characteristics. , , These are the initial calibration coefficients used before this iteration update; the part within parentheses represents the predicted flow rate value. This feature deviation is essentially the prediction error of the current calibration model under the current operating conditions. Next, the initial calibration coefficients are adjusted proportionally based on the calculated feature deviation to generate a new set of optimized adaptive calibration parameters. This adjustment follows the principles of adaptive algorithms such as gradient descent or minimum mean square error, meaning the direction of adjustment is related to the direction of the error, and the magnitude of the adjustment is proportional to the magnitude of the error and the magnitude of the input features. The adjustment process uses the following update rules to calculate the new adaptive calibration parameters:

[0090] ,

[0091] ,

[0092] ,

[0093] in, , , These are the newly generated adaptive calibration parameters, which will be used to calculate the flow rate value for the current cycle; , , These are the initial calibration coefficients before this round of adjustments; It is the calculated characteristic deviation; , It is a set of learning rates with specific dimensions. Through this step, the coefficients of the calibration model are fine-tuned according to the actual prediction error.

[0094] For example, consider a hydropower station monitoring system that has been operating for a year. Before this update, the system's initial calibration coefficient was... , , The currently acquired fusion flow rate features are: .first,

[0095] The system uses initial coefficients to make predictions and obtains the predicted flow velocity value.

[0096] The velocity is measured in meters per second. Then, from the historical velocity dataset, the system extracts a reference velocity value that is highly similar to the current operating conditions and has been verified as having high reliability. This value was calibrated by professionals using higher-precision equipment during the last inspection. meters per second. Calculate the characteristic deviation.

[0097] meters per second. This positive bias indicates that the current model's prediction is underestimated. Finally, the coefficients are adjusted proportionally based on this bias. Let the learning rate be... , , Its dimensions match the formula. Adaptive calibration parameters are generated based on the update rules: ,

[0098] ,

[0099] This method can effectively track and compensate for slowly varying system errors introduced by factors such as sensor performance drift, impeller wear, and the influence of deposits, enabling the measurement system to continuously improve itself over time, thereby maintaining high accuracy and reliability of flow rate measurement over a long period without human intervention.

[0100] S4. Error dynamic compensation module, used to perform dynamic error compensation on the calibration flow rate value based on the multi-source sensor data, and generate the final flow rate value;

[0101] Optionally, the error dynamic compensation module includes:

[0102] The working condition compensation unit is used to perform weighted summation and polynomial transformation on the filtered temperature signal and the filtered pressure signal to generate an error compensation value.

[0103] The dynamic correction unit is used to arithmetically add the error compensation value and the calibrated flow rate value to generate a compensated flow rate value.

[0104] The flow velocity filtering unit is used to perform moving average filtering on the compensated flow velocity value to generate the final flow velocity value.

[0105] Specifically, the operating condition compensation unit is activated first. This unit is specifically designed to quantify the dynamic impact of current temperature and pressure on the measurement results. It receives filtered temperature and pressure signals provided by the multi-source signal acquisition module. The core of this unit is a pre-established polynomial transformation model calibrated based on extensive experimental data. This model describes the complex nonlinear relationship between temperature and pressure changes and the residual error in flow velocity measurement. The so-called weighted summation is reflected in this polynomial model, where the coefficients of different variable terms represent their weights. The unit substitutes the real-time filtered temperature and pressure signals as independent variables into the model to calculate an error compensation value with flow velocity dimensions. A typical second-order polynomial transformation model can be expressed as:

[0106] ,

[0107] in, This is the generated error compensation value, and its physical unit is flow velocity. It is the filtered temperature signal value of the input. It is the filtered input pressure signal value. , , , , , These are a set of constant coefficients with appropriate dimensions obtained through offline calibration to ensure the consistency of the entire expression in terms of physical meaning and units. They reflect the linear and nonlinear contributions of temperature and pressure to the error. Next, the dynamic correction unit applies the error compensation value to the calibration flow rate value. This unit receives the calibration flow rate value output by the adaptive calibration module and arithmetically adds it to the error compensation value calculated by the operating condition compensation unit. The calculation formula is:

[0108] ,

[0109] in, This is the generated compensated flow velocity value. It is the calibration flow rate value input from the adaptive calibration module. This is the error compensation value calculated by the operating condition compensation unit. Finally, the flow velocity filtering unit smooths the dynamically corrected result. The compensated flow velocity value may experience slight, high-frequency fluctuations due to the introduction of the compensation algorithm or residual noise in the sensor signal. To provide a stable and reliable final reading, this unit uses a moving average filtering algorithm to process the time series of the compensated flow velocity value, filtering out these short-term fluctuations and generating the final flow velocity value.

[0110] For example, taking tailrace monitoring at a hydropower station as an example, on a summer afternoon, due to the release of floodwater from the upstream reservoir, the tailrace water level rises and the water temperature changes rapidly. The error dynamic compensation module comes into play at this time. Assume the calibration flow rate output by the adaptive calibration module is 2.964 m / s. The operating condition compensation unit receives a real-time filtered temperature signal showing a sudden rise in water temperature to T=26 degrees Celsius, while the filtered pressure signal shows an increase in water pressure to p=165 kPa. This unit uses a preset polynomial transformation model to calculate the error compensation value. Let the model coefficients be... , , , , , Substituting into the calculation, we get...

[0111] m / s. This value indicates that the current temperature and pressure conditions will introduce an additional error of approximately 0.0115 m / s. The dynamic correction unit then arithmetically adds this value to the calibrated flow rate value to generate the compensated flow rate value. The velocity is measured in meters per second (m / s). Finally, to provide stable readings, the velocity filtering unit performs a moving average of the compensated velocity value sequence over a short period, outputting a smooth final velocity value, for example, 2.976 m / s. This dynamic compensation mechanism suppresses instantaneous measurement errors introduced by rapid changes in water viscosity and density with temperature and pressure, enabling the measurement system to respond faster and be more robust to changes in environmental conditions. Ultimately, this module ensures that the output final velocity value achieves higher accuracy and stability under various complex and changing actual measurement conditions.

[0112] S5. Dataset update module, used to append the final flow rate value to the historical flow rate dataset to generate an updated historical flow rate dataset for the next adjustment.

[0113] Optionally, the dataset update module includes:

[0114] The validity verification unit is used to verify the validity of the final flow rate value and generate valid flow rate data.

[0115] A time-series associated storage unit is used to associate the effective flow velocity data with the corresponding collection timestamp and store it in the historical flow velocity dataset to generate a historical flow velocity dataset with time-series tags.

[0116] The timeliness filtering unit is used to perform timeliness filtering on the historical flow rate dataset with time sequence markers to generate an updated historical flow rate dataset.

[0117] Specifically, such as Figure 2 As shown, firstly, the validity verification unit rigorously reviews and filters the final flow rate value output by the error dynamic compensation module. This unit compares the received final flow rate value with the statistical characteristics in the historical flow rate dataset to determine if it falls within a reasonable range. Only data that passes verification is considered valid flow rate data and allowed to proceed to the next step. Abnormal data that fails verification is discarded and does not participate in dataset updates. Secondly, the time-series correlation storage unit is responsible for normalizing and storing the verified valid flow rate data. This unit appends a precise timestamp to each valid flow rate data point, derived from the system clock and strictly corresponding to the time when the multi-source signal acquisition module acquired the data. Then, it appends this time-stamped valid flow rate data as a new record to the system's historical flow rate dataset. In this way, the historical flow rate dataset is constructed into a time-stamped historical flow rate dataset. This dataset not only records the magnitude of the flow rate but also the time of its occurrence, forming a historical trajectory of flow rate evolution over time. Finally, the time-series filtering unit maintains and manages the entire time-stamped historical flow rate dataset. Over time, a large amount of data accumulates in the dataset. However, outdated data may not accurately reflect the current performance status of the sensor or the long-term trends of the measurement environment. To ensure the timeliness and representativeness of the adaptive learning samples, this unit periodically or when the dataset reaches a certain size, filters the dataset. For example, a maximum retention window can be set, such as retaining data from the most recent week or month, or a fixed data volume limit can be set, employing a first-in-first-out (FIFO) strategy to remove the oldest data. After this timeliness filtering, an updated historical flow velocity dataset is obtained that contains the latest information while eliminating outdated interference. This updated dataset will serve as the basis for the next online learning iteration of the adaptive calibration module.

[0118] For example, taking a hydropower station monitoring device after completing a measurement, its dataset update module will be activated. The system generates a final flow velocity value of 2.976 m / s, with a collection timestamp of "2025-02-01 15:00:00". First, the validity verification unit reviews the data. By comparing it with the statistical characteristics of historical data, the value is determined to be within a reasonable range and is therefore confirmed as valid flow velocity data. Subsequently, the time-series association storage unit pairs the valid flow velocity data with its timestamp to form a record ("2025-02-01 15:00:00", 2.976), and stores it in the device's historical flow velocity dataset, thus creating a time-series-tagged historical flow velocity dataset. Finally, the time-series filtering unit maintains the entire dataset. For example, the system's strategy is to retain only data from the most recent 90 days to accommodate seasonal hydrological changes. The unit scanned the dataset and found a record with a timestamp of "2024-09-30 12:00:00", which exceeded the 90-day expiration period, so it was deleted from the dataset. Through this series of operations, the system obtained an updated historical flow rate dataset that contained both the latest valid data and outdated information, providing high-quality training samples for the next round of adaptive learning. Through the implementation of the dataset update module, this system established an intelligent data management and feedback mechanism. It not only achieved automatic accumulation of historical data, but more importantly, ensured the quality of data used for adaptive learning through validity verification and timeliness filtering.

[0119] Optionally, the validity verification of the final flow rate value includes:

[0120] Calculate the deviation between the final flow velocity value and the statistical feature value of the historical flow velocity dataset, and generate the historical deviation amount;

[0121] The historical deviation is compared with a preset confidence interval threshold. When the historical deviation is within the range of the confidence interval threshold, it is determined to be valid flow velocity data.

[0122] Specifically, the first step is to calculate the deviation of the current final flow velocity value to be verified relative to the center of historical data distribution, generating a historical deviation. The first step in this process is to extract the core statistical features from the existing time-stamped historical flow velocity dataset, namely the arithmetic mean and standard deviation of all flow velocity values ​​within the dataset. Then, the current final flow velocity value received from the error dynamic compensation module is compared with the calculated historical average, and normalized using the historical standard deviation. The historical deviation calculated in this way is a dimensionless standardized measure that objectively reflects the degree to which the current measurement deviates from the central trend of the historical data. Its calculation formula is as follows:

[0123] ,

[0124] in, For the generated historical deviation, This is the final flow rate value to be verified. It is a historical statistical characteristic value obtained by calculating the arithmetic mean of all data in the historical flow velocity dataset. This is another historical statistical characteristic value obtained by calculating the standard deviation of all data in the historical flow velocity dataset. It represents the dispersion of the historical data. If the historical dataset is empty or the sample size is too small to calculate a stable standard deviation, an initial, preset variance value can be used, or the data can be directly determined as valid. Next, the calculated historical deviation is compared with a preset confidence interval threshold to make a final validity determination. The confidence interval threshold is a positive number preset according to statistical principles. For example, setting the threshold to 3 corresponds to the range of approximately 99.7% of the data in a normal distribution. During the comparison, the absolute value of the historical deviation is compared with this threshold. When the absolute value of the historical deviation is less than or equal to the preset confidence interval threshold, it indicates that the current final flow velocity value is within the normal fluctuation range of the historical data. The system therefore determines it as valid flow velocity data and allows it to enter the subsequent time-series correlation storage unit. Conversely, if its absolute value exceeds the threshold, the data is considered an outlier or abnormal value caused by sudden interference or system failure, and will be determined as invalid and discarded directly. Figure 3 As shown in the figure, this diagram visually illustrates the core principle of determining the validity of data by comparing the current flow rate value with the statistical distribution of historical data to determine whether it falls within a preset normal range.

[0125] For example, consider a water quality monitoring section of an urban river. The monitoring equipment reports a final flow velocity value, which needs to be validated. Suppose that after a heavy rain, the river flow surges, but then a large piece of floating garbage briefly gets caught on the impeller, causing the final flow velocity value of a single measurement to change. The velocity is abnormally low, at 0.15 m / s. The validity verification unit is initiated. First, it calculates the statistical characteristics of the historical flow velocity dataset. Based on data from the most recent week, the historical average flow velocity is... The speed is 1.20 m / s, and the standard deviation is 1.20 m / s. The value is 0.25 m / s, reflecting the normal state of high and unstable river flow velocity after a rainstorm. Then, the unit calculates the historical deviation. Next, the unit will use the absolute value of the historical deviation. The data is compared with a preset confidence interval threshold, typically set to 3.0. Since 4.2 is greater than 3.0, this historical deviation exceeds the confidence interval range. Therefore, the system determines that the final flow velocity value of 0.15 m / s is an outlier, most likely caused by a sudden event such as garbage entanglement, and discards it as invalid flow velocity data, not adding it to the historical flow velocity dataset. Compared to simple fixed threshold judgment, this method uses a dynamically changing judgment criterion, adaptively adjusting as the historical flow velocity dataset is updated, thus reflecting the normal data fluctuation range under current operating conditions. This mechanism prevents outlier data from contaminating the historical sample library used for learning, ensuring the purity and reliability of the data foundation upon which the adaptive calibration module relies, thereby ensuring the stability and convergence of the entire system's online learning process.

[0126] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides a method for measuring flow velocity using an impeller-type hydraulic current meter, the method comprising:

[0127] The pulse signal generated by the impeller sensor, the pressure signal generated by the pressure sensor, and the temperature signal generated by the temperature sensor are acquired to generate multi-source sensor data.

[0128] The multi-source sensor data is fused to generate fused flow velocity features;

[0129] The fused flow velocity features are adaptively adjusted based on historical flow velocity datasets to generate calibrated flow velocity values.

[0130] Based on the multi-source sensor data, dynamic error compensation is performed on the calibrated flow rate value to generate the final flow rate value;

[0131] The final flow rate value is appended to the historical flow rate dataset to generate an updated historical flow rate dataset for the next adjustment.

[0132] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0133] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A flow velocity measurement system for a turbine-type hydraulic current meter, characterized in that, The system includes: The multi-source signal acquisition module is used to acquire pulse signals generated by the impeller sensor, pressure signals generated by the pressure sensor, and temperature signals generated by the temperature sensor, and generate multi-source sensor data. The feature fusion processing module is used to fuse the multi-source sensor data to generate fused flow velocity features; An adaptive calibration module is used to adaptively adjust the fused flow velocity features based on a historical flow velocity dataset to generate a calibrated flow velocity value. The error dynamic compensation module is used to perform dynamic error compensation on the calibrated flow rate value based on the multi-source sensor data to generate the final flow rate value. The dataset update module is used to append the final flow rate value to the historical flow rate dataset to generate an updated historical flow rate dataset for the next adjustment.

2. The flow velocity measurement system for a turbine-type hydraulic flow meter according to claim 1, characterized in that, The multi-source signal acquisition module includes: The signal sampling unit is used to synchronously acquire pulse signals, pressure signals, and temperature signals to generate a raw signal set. An analog-to-digital conversion unit is used to perform analog-to-digital conversion on the original signal set to generate a digital signal set; A digital signal filtering unit is used to filter the digital signal set to generate multi-source sensor data.

3. A flow velocity measurement system for a turbine-type hydraulic flow meter according to claim 2, characterized in that, The feature fusion processing module includes: A multi-source signal decoupling unit is used to extract filtered pressure signal, filtered temperature signal and filtered pulse signal from the multi-source sensor data, respectively. The turbulence feature analysis unit is used to calculate the turbulence intensity of the water flow based on the filtered pressure signal and generate turbulence intensity parameters. The fluid property correction unit is used to calculate the water density correction factor based on the filtered temperature signal and generate density correction parameters. The pulse signal compensation unit is used to compensate the filtered pulse signal by combining the turbulence intensity parameter and the density correction parameter to generate compensated pulse data; The flow velocity feature extraction unit is used to extract features from the compensation pulse data and generate fused flow velocity features.

4. A flow velocity measurement system for a turbine-type hydraulic current meter according to claim 3, characterized in that, The feature extraction of the compensation pulse data includes: Calculate the average value of the compensation pulse data to generate an average pulse value; Calculate the variance of the compensated pulse data to generate the pulse variance; The average pulse value and the pulse variance are combined to generate a fused flow velocity feature.

5. A flow velocity measurement system for a turbine-type hydraulic current meter according to claim 3, characterized in that, The adaptive calibration module includes: An online learning calibration unit is used to calculate adaptive calibration parameters based on the historical flow velocity dataset and the fused flow velocity features; The flow rate calculation unit is used to adjust the fused flow rate characteristics by applying the adaptive calibration parameters to generate a calibrated flow rate value.

6. A flow velocity measurement system for a turbine-type hydraulic flow meter according to claim 5, characterized in that, The calculation of adaptive calibration parameters includes: A feature deviation is generated by calculating the difference between the historical flow velocity dataset and the fused flow velocity features; The initial calibration coefficients are adjusted proportionally based on the aforementioned characteristic deviation to generate adaptive calibration parameters.

7. A flow velocity measurement system for a turbine-type hydraulic current meter according to claim 3, characterized in that, The error dynamic compensation module includes: The working condition compensation unit is used to perform weighted summation and polynomial transformation on the filtered temperature signal and the filtered pressure signal to generate an error compensation value. The dynamic correction unit is used to arithmetically add the error compensation value and the calibrated flow rate value to generate a compensated flow rate value. The flow velocity filtering unit is used to perform moving average filtering on the compensated flow velocity value to generate the final flow velocity value.

8. A flow velocity measurement system for a turbine-type hydraulic flow meter according to claim 7, characterized in that, The dataset update module includes: The validity verification unit is used to verify the validity of the final flow rate value and generate valid flow rate data. A time-series associated storage unit is used to associate the effective flow velocity data with the corresponding collection timestamp and store it in the historical flow velocity dataset to generate a historical flow velocity dataset with time-series tags. The timeliness filtering unit is used to perform timeliness filtering on the historical flow rate dataset with time sequence markers to generate an updated historical flow rate dataset.

9. A flow velocity measurement system for a turbine-type hydraulic flow meter according to claim 8, characterized in that, The validity verification of the final flow rate value includes: Calculate the deviation between the final flow velocity value and the statistical feature value of the historical flow velocity dataset, and generate the historical deviation amount; The historical deviation is compared with a preset confidence interval threshold. When the historical deviation is within the range of the confidence interval threshold, it is determined to be valid flow velocity data.

10. A method for measuring flow velocity using a turbine-type hydraulic current meter, characterized in that, The method includes: The pulse signal generated by the impeller sensor, the pressure signal generated by the pressure sensor, and the temperature signal generated by the temperature sensor are acquired to generate multi-source sensor data. The multi-source sensor data is fused to generate fused flow velocity features; The fused flow velocity features are adaptively adjusted based on historical flow velocity datasets to generate calibrated flow velocity values. Based on the multi-source sensor data, dynamic error compensation is performed on the calibrated flow rate value to generate the final flow rate value; The final flow rate value is appended to the historical flow rate dataset to generate an updated historical flow rate dataset for the next adjustment.

Images