A flow meter data self-adaptive calibration system

Abstract

The application discloses a flowmeter data self-adaptive calibration system, and belongs to the technical field of flowmeter calibration, which comprises four modules of flow signal acquisition, working condition parameter monitoring, main control processing and intelligent calibration operation, and the main control processing unit is in bidirectional communication connection with the remaining modules. The acquisition module processes flowmeter original signals to generate standardized data and extract effective signals; the monitoring module collects working condition parameters in a verification field in real time, completes space registration and characteristic fitting modeling; the main control unit analyzes data, schedules algorithms, trains iterations and verifies and optimizes a calibration model; and the calibration operation module quantifies flow error characteristics to realize dynamic self-adaptive correction of measurement data. The modules are cooperated to realize online real-time calibration of flowmeters without manual intervention, production is not interrupted, working condition changes and instrument drift trends can be matched, measurement errors can be corrected in time, the problems of poor timeliness and interference with production in traditional calibration are solved, and long-term high-precision measurement of flowmeters is ensured.

Description

Technical Field

[0001] This invention relates to the field of flow measurement instrument calibration technology, specifically to a flow meter data adaptive calibration system. Background Technology

[0002] Flow meters are core metering equipment in industrial production control, energy trade settlement, environmental pollution monitoring, and municipal water affairs operation and maintenance. Their measurement accuracy directly determines the rationality of production scheduling, the fairness of cost accounting, and the compliance of data reporting. They are key hardware for ensuring closed-loop industrial processes and refined energy management.

[0003] During long-term continuous service, the complex and ever-changing operating environment in industrial settings continuously interferes with measurement accuracy: real-time fluctuations in medium temperature and pipeline pressure, and dynamic changes in fluid viscosity and velocity, can break the measurement benchmarks set at the factory calibration of the flowmeter. Furthermore, aging of internal sensing elements, zero-point drift, and mechanical wear and tear further exacerbate the accumulation of measurement deviations, leading to distortion of the original flow data and exceeding measurement error limits. Currently, the mainstream flowmeter calibration method in the industry still relies on the traditional approach of manually disassembling the instrument and sending it to a professional metrology institution for offline calibration. This method not only fails to synchronize with the real-time changes in on-site operating conditions and the dynamic drift trend of the instrument, but also cannot provide real-time online correction of measurement errors, making it difficult to maintain a long-term stable and high-precision measurement state. This is currently a major technical challenge in flow meter calibration. Summary of the Invention

[0004] To address the problems of existing technologies, this invention provides a flow meter data adaptive calibration system, comprising: The system includes a flow signal acquisition module, an operating condition parameter monitoring module, a main control processing unit, and an intelligent calibration calculation module. The main control processing unit establishes bidirectional communication connections with the flow signal acquisition module, the operating condition parameter monitoring module, and the intelligent calibration calculation module. The flow signal acquisition module performs multi-stage processing on the raw measurement signal output by the flow meter to generate standardized flow data, and simultaneously integrates multiple types of raw signals from the flow meter to construct a raw flow sensing dataset and extracts effective sensing signals from this dataset. The operating condition parameter monitoring module performs real-time acquisition and validity verification of multi-dimensional operating condition parameters at the flow measurement site, and simultaneously completes spatial registration and parameter feature fitting modeling of operating condition parameters. The main control processing unit realizes the fusion analysis of traffic data and operating parameters and algorithm scheduling. It fuses standardized traffic data with spatially registered operating parameters to generate a spatiotemporally fused traffic and operating condition joint dataset. The fused dataset is divided into an algorithm training set and an algorithm verification set according to a preset ratio. Based on the above division of the fused dataset, the training iteration and verification optimization of the calibration algorithm model are completed. The intelligent calibration calculation module relies on the partitioning results of the fused dataset to achieve dynamic adaptive correction of flow measurement data through quantitative analysis of flow error characteristics. The modules work together to complete the real-time calibration of the flow meter without human intervention.

[0005] Furthermore, the intelligent calibration calculation module incorporates an error tracing unit, a dynamic compensation unit, and a correction execution unit; The error tracing unit compares the standardized flow data with the flow reference value adapted to the operating conditions in real time, identifies the cause of the error and classifies the error into two categories: operating condition fluctuation error and instrument drift error. At the same time, it uses the flow error feature area calibration method and the flow error feature contour analysis algorithm to extract and quantify the flow error features and generate the flow error feature quantification value. The dynamic compensation unit constructs an error dynamic compensation model that matches the on-site working conditions based on the identified error causes, the actual changes in on-site operating parameters, and the quantified value of flow error characteristics. The correction execution unit performs point-by-point optimization on the standardized flow data based on the real-time correction coefficients output by the error dynamic compensation model, thereby achieving differentiated calibration of operating condition fluctuation errors and instrument drift errors.

[0006] Furthermore, the operating condition parameter monitoring module adopts a multi-sensor synchronous acquisition architecture, which synchronously acquires multi-dimensional operating condition parameters such as medium temperature, pipeline pressure, and fluid characteristics at the flow measurement site through sensor acquisition units with different functions. The collected multi-source operating condition parameters are sequentially subjected to time series smoothing and outlier intelligent removal to generate a continuous and effective time series dataset of operating condition parameters. A reference domain for fitting the features of operating parameters is constructed based on the time series dataset of operating parameters. Spatial registration of operating parameters is completed by centering the reference domain, and a spatially registered dataset of operating parameters is generated and synchronously transmitted to the main control processing unit. The operating condition parameter monitoring module has a built-in medium adaptation unit, which can dynamically adjust the acquisition frequency and abnormal value judgment threshold of the operating condition parameters according to the type of the measured medium, and adapt to the operating condition change characteristics of different measured media such as liquids, gases and steam.

[0007] Furthermore, the main control processing unit is equipped with a data fusion engine and a self-learning algorithm scheduling module; The data fusion engine is equipped with a spatiotemporal feature association algorithm. It first performs feature layer association matching on the received standardized traffic data and spatially registered operating condition parameter dataset and extracts the association feature values ​​between traffic and operating conditions. Then, it completes the data layer fusion and splicing based on the association feature values, while removing invalid data that is mismatched in the spatiotemporal dimension, and generating the spatiotemporally fused traffic and operating condition joint dataset. The self-learning algorithm scheduling module performs flow and operating condition feature annotation processing on the partitioned algorithm training set and generates a feature annotation dataset. The algorithm training set and the feature annotation dataset are then used to perform full-dimensional training on the initial error compensation algorithm model.

[0008] Furthermore, the flow signal acquisition module integrates a signal adaptation unit, a signal conditioning unit, and a digital-to-analog conversion unit; The signal adapter unit is compatible with the input of various raw measurement signals, including pulse, analog, and digital signals, from different types of flow meters, and integrates these raw measurement signals to construct a set of raw flow sensing data. The signal conditioning unit sequentially performs filtering, amplification, and anti-interference processing on the original measurement signal, and simultaneously performs targeted extraction of effective sensing signals on the original flow sensing data set, completing the separation and noise reduction of flow sensing features, and generating noise-reduced flow sensing feature data. The digital-to-analog conversion unit converts the analog signal processed by the signal conditioning unit into standardized digital flow data, adds a timestamp to the standardized digital flow data, and simultaneously associates and binds the flow sensing feature noise reduction data with the standardized digital flow data using timestamps. The flow signal acquisition module has a built-in signal adaptation and self-adjustment unit, which can automatically adjust the signal adaptation parameters and sensing feature extraction strategy according to the type of flow meter connected, and adapt to the signal output characteristics of different types of flow meters such as electromagnetic, vortex, and differential pressure.

[0009] Furthermore, the system is also equipped with a data interaction module; The data interaction module receives the calibrated flow data, raw flow data, operating parameters and flow error characteristic analysis data output by the main control processing unit in real time. The local display unit displays the above data in real time in the form of numerical values, trend graphs and feature fitting graphs. At the same time, the multi-protocol communication unit performs encryption processing on various types of data and uploads them to the remote monitoring platform. The data interaction module stores the complete set of raw flow sensor data and calibration algorithm model-related data, providing underlying data support for the iterative upgrade of the calibration algorithm model.

[0010] Furthermore, the dynamic compensation unit constructs a real-time response compensation sub-model for operating condition fluctuation error and a trend fitting compensation sub-model for instrument drift error based on the differentiated characteristics of error types. The real-time response compensation sub-model makes rapid coefficient adjustments for instantaneous changes in operating parameters, and the trend fitting compensation sub-model makes trend-based compensation corrections based on the long-term variation law of instrument drift. The dynamic compensation unit integrates the two sub-models into a unified error dynamic compensation model through dynamic weight allocation, adapting to the changing characteristics of different error types.

[0011] Furthermore, the self-learning algorithm scheduling module uses the partitioned algorithm verification set to perform multi-scenario verification and parameter tuning on the trained initial error compensation algorithm model, thereby obtaining the optimal error compensation algorithm model that is adapted to the field. The multi-scenario verification includes verification scenarios with different measurement media, different operating condition parameter fluctuation ranges, and different flow meter running times. Parameter optimization takes minimizing the flow measurement error in each scenario as the core objective. The self-learning algorithm scheduling module continuously iterates and optimizes the parameter weights of the optimal error compensation algorithm model based on the real-time updated spatiotemporal fusion flow and operating condition joint dataset, and dynamically adjusts the operation frequency and calibration strategy of the intelligent calibration operation module according to the real-time changes in the field operating conditions.

[0012] Furthermore, the intelligent calibration calculation module has a closed-loop iterative optimization function for calibration effect. After completing the first round of standardized flow data correction, it continuously compares the calibrated flow data with the flow reference value that is adapted to the real-time operating conditions to generate multi-dimensional calibration effect feedback data. The calibration effect feedback data is subjected to secondary extraction of effective sensor signals and re-analysis of flow error feature quantification values. The calibration effect feedback data after secondary processing is then sent back to the error tracing unit and the dynamic compensation unit. The error tracing unit dynamically adjusts the error identification criteria based on feedback data, and the dynamic compensation unit optimizes the parameter settings of the error dynamic compensation model based on feedback data, continuously iterating and optimizing the correction effect of flow data to ensure that the measurement error of the flow meter is always within the preset threshold range.

[0013] The beneficial effects of this invention are: The various modules work together to complete the online real-time calibration of the flow meter without interrupting the industrial production process. It synchronously matches the real-time changes in the on-site operating conditions and the dynamic drift trend of the instrument, and can immediately correct measurement errors online. This effectively avoids the problems of poor timeliness and interference with production caused by traditional calibration, and can maintain the flow meter's long-term stable high-precision measurement state. By classifying the causes of errors into two categories, namely operating condition fluctuation errors and instrument drift errors, it constructs a dedicated compensation sub-model for different error types and dynamically integrates them to achieve differentiated and accurate calibration. At the same time, the intelligent calibration calculation module has a closed-loop iterative optimization function for calibration effects. Through continuous comparison of the calibrated data with the reference value, secondary processing of feedback data, and dynamic adjustment of model parameters, it ensures that the flow meter measurement error is always within the preset threshold range. This fundamentally solves the problem of measurement deviation accumulation caused by complex industrial on-site operating conditions and instrument hardware attenuation, and avoids distortion of the original flow data and excessive measurement errors. Attached Figure Description

[0014] Figure 1A schematic flowchart of the flowmeter data adaptive calibration system provided by the present invention; Figure 2 This is a schematic flowchart of the intelligent calibration calculation module provided by the present invention; Figure 3 A flowchart illustrating the working condition parameter monitoring module provided by this invention. Detailed Implementation

[0015] 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, and 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.

[0016] Please see Figure 1-3 This invention provides a flow meter data adaptive calibration system, which consists of a flow signal acquisition module, an operating condition parameter monitoring module, a main control processing unit, and an intelligent calibration calculation module. These four modules enable the system to perform flow calibration. The main control processing unit serves as the core control and data interaction node of the system. It establishes bidirectional communication connections with the flow signal acquisition module, the operating condition parameter monitoring module, and the intelligent calibration calculation module, enabling bidirectional command issuance, real-time bidirectional data transmission, and bidirectional feedback of operating status between each module and the main control processing unit.

[0017] The flow signal acquisition module performs multi-stage standardization processing, including filtering, conditioning, and conversion, on the raw measurement signals output by various flow meters to generate standardized flow data with unified format and range. At the same time, it integrates multiple types of raw signals, including pulse, analog, and digital signals output by the flow meters to construct a raw flow sensor dataset. It then accurately extracts noise-free and interference-free effective sensor signals from this dataset. The process of extracting effective sensor signals includes the following steps: First, the signal conditioning unit performs three types of processing on the original measurement signal: analog / digital dual filtering, weak signal amplification, and electromagnetic interference suppression. This directly filters out high-frequency noise, power frequency interference, and random clutter from the original signal, eliminating external interference from pipelines and the field environment, and transforming the impure signal into a clean signal. Second, the signal conditioning unit performs targeted extraction of effective sensor signals after preprocessing. Based on the type of flow meter connected, it automatically matches the corresponding flow sensor feature template. From the preprocessed signal, it accurately separates the core sensor features that represent only the physical quantity of flow, eliminating all interference components that are not flow features. This completes the separation and deep noise reduction of flow sensor features, generating denoised flow sensor feature data. Third, the digital-to-analog conversion unit adds a timestamp to the effective signal and binds it to the standardized flow data timestamp, ensuring that the extracted effective signal is spatiotemporally synchronized with the flow data, without distortion or misalignment.

[0018] The operating condition parameter monitoring module continuously collects multi-dimensional core operating condition parameters such as medium temperature, pipeline pressure, and fluid characteristics at the flow measurement site in real time. At the same time, it performs multi-dimensional validity verification on the collected raw operating condition parameters, eliminates invalid and distorted data, and simultaneously completes spatial coordinate registration and multi-dimensional parameter feature fitting modeling of the operating condition parameters to form an operating condition parameter model that fits the actual site conditions.

[0019] The main control processing unit receives and performs in-depth fusion analysis of traffic data and operating parameters. At the same time, it completes the scheduling, start-up, shutdown and management of all calibration algorithms in the system. Based on the scientific division of fused data, it completes the full-process training iteration and multi-scenario verification and optimization of the calibration algorithm model to ensure that the model is adapted to the field operating conditions.

[0020] The intelligent calibration calculation module relies on the comprehensive data analysis results after the fusion of flow data and operating parameters. Through the accurate identification and quantitative analysis of flow error characteristics, it realizes dynamic, adaptive and corrective flow measurement data. Each module cooperates and operates in conjunction according to the preset logic, and finally completes the online real-time calibration of the flow meter without human operation or intervention.

[0021] In some embodiments, the intelligent calibration calculation module has built-in an error tracing unit, a dynamic compensation unit, and a correction execution unit that are interconnected and execute functions sequentially. The three units form a complete calibration calculation process.

[0022] The error tracing unit compares the standardized flow data output by the flow signal acquisition module with the flow reference value that matches the real-time field conditions in real time, and calculates the instantaneous flow error and error time series. It then uses the flow error characteristic region calibration method to determine the distribution characteristics of the error in the dimension of the operating condition parameters and to determine the correlation strength between the error and the operating condition. Next, it uses the flow error characteristic profile analysis algorithm to analyze the error time series change profile and identify the abrupt / gradual and cumulative / non-cumulative morphological characteristics of the error. Based on the joint discrimination of the operating condition correlation characteristics and the error time series profile characteristics, the algorithm automatically identifies the specific causes of the flow measurement error and classifies the error into two categories: operating condition fluctuation error and instrument drift error.

[0023] The dynamic compensation unit, based on the error cause type identified by the error tracing unit, the real-time changes in on-site operating parameters, and the quantitative value of flow error characteristics, constructs a real-time response compensation sub-model for operating condition fluctuation error using a real-time mapping fitting algorithm between operating conditions and errors, and constructs a trend fitting compensation sub-model for instrument drift error using an instrument drift time series fitting algorithm. Then, the two sub-models are weighted and fused using a dynamic weight adaptive allocation algorithm to construct a dynamic error compensation model that is highly matched to the real-time operating conditions of the flow measurement site and is dynamically updated.

[0024] In some embodiments, the operating condition parameter monitoring module adopts a multi-sensor synchronous acquisition architecture with multiple sensing units at the same location and in the same sequence. Through sensing acquisition units with different functions such as temperature sensing, pressure sensing, and fluid characteristic sensing, it synchronously and accurately acquires multi-dimensional core operating condition parameters such as medium temperature, pipeline pressure, fluid viscosity / flow velocity, and fluid characteristics at the flow measurement site, ensuring the spatiotemporal consistency of parameter acquisition.

[0025] The collected multi-source, multi-dimensional operating condition parameters are sequentially subjected to time series smoothing to eliminate random jitter during the data acquisition process. Then, an intelligent algorithm is used to intelligently remove outliers, identify and remove outlier, distorted, and invalid parameter data, and generate a continuous, stable, and effective time series dataset of operating condition parameters.

[0026] Based on the time-series dataset of operating parameters, a feature fitting benchmark domain covering the entire operating condition range is constructed. By accurately locating the center of the fitting benchmark domain, spatial coordinate registration of all operating parameters is completed, unifying the spatial benchmark of the parameters and generating a spatially registered standardized operating condition parameter dataset. This dataset is then transmitted in real-time and synchronously to the main control processing unit for further processing. The operating parameter monitoring module incorporates a media adaptation unit. Through a multi-feature joint discrimination algorithm for media types, it matches and compares the real-time collected operating parameter features with a pre-stored standard feature benchmark library for liquids, gases, and steam. This automatically identifies the specific type of flow measurement medium and dynamically adjusts the sampling frequency and outlier threshold based on the operating condition variation characteristics of different measurement media (liquid, gas, steam). For example, the sampling frequency is increased for frequently fluctuating gas media and decreased for stable liquid media, ensuring that parameter acquisition is adapted to the actual operating conditions of different media.

[0027] In some embodiments, the main control processing unit is equipped with a data fusion engine that performs data fusion and a self-learning algorithm scheduling module that performs algorithm model management. The two modules work together to achieve data processing and algorithm optimization.

[0028] The data fusion engine has a built-in spatiotemporal feature association algorithm specifically for traffic and operating condition data. For the received standardized traffic data and spatially registered operating condition parameter dataset, it first performs deep association matching at the feature layer to mine and extract the correlation feature values ​​between traffic data and operating condition parameters. Then, based on the extracted correlation feature values, it completes seamless fusion and splicing of the data layer. At the same time, it strictly removes invalid data with spatiotemporal dimension mismatch due to asynchronous time and inconsistent spatial locations. Finally, it generates a spatiotemporally fused traffic and operating condition joint dataset with unified spatiotemporal features and data association.

[0029] The self-learning algorithm scheduling module scientifically divides the generated traffic and operating condition joint dataset into an algorithm training set for model training and an algorithm validation set for model validation according to a customizable preset ratio. The algorithm training set performs classification and labeling processing on traffic and operating condition features to generate a feature-labeled dataset with clear feature tags. Using the algorithm training set and the feature-labeled dataset as the training basis, the initial error compensation algorithm model is trained in all dimensions, covering all parameters and all operational dimensions, to ensure the training effect of the model.

[0030] In some embodiments, the flow signal acquisition module integrates a signal adaptation unit, a signal conditioning unit, and a digital-to-analog conversion unit that operate sequentially according to the signal processing flow.

[0031] The signal adapter unit has wide signal compatibility, and can be compatible with the input of various raw measurement signals such as pulse, 4-20mA analog, and RS485 digital signals output by different types of flow meters, including electromagnetic, vortex, and differential pressure flow meters. At the same time, it classifies and integrates the various types of raw measurement signals to construct a complete and traceable set of raw flow sensing data.

[0032] The signal conditioning unit sequentially performs digital and analog dual filtering, precise amplification of weak signals, and electromagnetic interference suppression on various raw measurement signals. At the same time, it performs targeted extraction of effective sensing signals on the raw flow sensing data set, removes noise and interference signals, completes effective separation and deep noise reduction of flow sensing features, and finally generates noise-reduced flow sensing feature data without interference.

[0033] The analog-to-digital conversion unit accurately converts the analog signal processed by the signal conditioning unit into standardized digital flow data with a unified format and range. It adds a timestamp to each set of standardized digital flow data and simultaneously associates and binds the flow sensor feature noise reduction data with the standardized digital flow data based on the timestamp, ensuring the spatiotemporal consistency of the two types of data. The flow signal acquisition module has a built-in signal adaptation and self-adjustment unit. Through a flow meter type signal feature matching and recognition algorithm, it matches and compares the real-time acquired raw measurement signal features with a pre-stored standard signal feature reference library for electromagnetic, vortex, and differential pressure flow meters. It automatically identifies the specific type of flow meter connected and automatically adjusts the signal adaptation parameters and sensor feature extraction strategy based on the identification results to adapt to the signal output characteristics of different types of flow meters, including electromagnetic, vortex, and differential pressure flow meters.

[0034] In some embodiments, the system is further equipped with a data interaction module for data display, transmission, and storage, serving as an interaction between the system and operators and remote platforms.

[0035] The data interaction module receives, in real time, all data related to the calibration flow data, raw flow data, real-time on-site operating parameters, and flow error characteristic analysis output by the main control processing unit through two-way communication.

[0036] This module uses its built-in local display unit to visualize the received data in real-time, presenting real-time numerical values, dynamic trend graphs, and multi-dimensional feature fitting maps, allowing on-site operators to view the data intuitively. Simultaneously, through its built-in multi-protocol communication unit, it is compatible with mainstream industrial communication protocols such as Modbus and TCP / IP. All displayed and transmitted data undergoes dual encryption—data encryption and transmission encryption—before being securely uploaded to a remote monitoring platform, enabling real-time remote monitoring.

[0037] In addition, the data interaction module is equipped with a large-capacity storage unit, which can store the full set of original sensor data and training, optimization and operation data of the calibration algorithm model generated during system operation for a long time, without missing any core data, providing a comprehensive data foundation for the subsequent offline iteration and online upgrade of the calibration algorithm model.

[0038] In some embodiments, the dynamic compensation unit constructs targeted compensation sub-models based on the differentiated change characteristics of the two error types: operating condition fluctuation error and instrument drift error.

[0039] To address the characteristics of instantaneous and sudden changes in operating condition fluctuation errors, a real-time response compensation sub-model for operating condition fluctuation errors is constructed. This model can perform millisecond-level rapid coefficient adjustments for instantaneous changes and small fluctuations in operating parameters, thereby achieving real-time and rapid compensation for operating condition fluctuation errors.

[0040] To address the characteristics of instrument drift error, which changes slowly and exhibits a long-term trend, a trend fitting compensation sub-model for instrument drift error is constructed. This model is based on historical drift data from long-term instrument operation, and uses an algorithm to fit the drift change trend curve. Based on the trend curve, long-term and gradual compensation correction is performed to achieve long-term and accurate compensation for instrument drift error.

[0041] The dynamic compensation unit uses a real-time weight allocation algorithm to dynamically adjust the weight coefficients of the two sub-models based on the actual proportion of the two types of errors in the field conditions. For example, when the field conditions fluctuate drastically, the weight of the real-time response compensation sub-model is increased; when the instrument runs for a long time and drifts significantly, the weight of the trend fitting compensation sub-model is increased. The two sub-models are organically integrated into a whole dynamic error compensation model, so that the integrated model has both real-time and long-term effectiveness and can adapt to the changing characteristics of different error types.

[0042] In some embodiments, the self-learning algorithm scheduling module utilizes a predefined algorithm verification set to conduct full-coverage multi-scenario verification and fine-tuning of the initial error compensation algorithm model after full-dimensional training. Through multi-scenario verification and optimization, the optimal error compensation algorithm model that is highly adapted to the actual working conditions and actual operating status of the flow meter at the flow measurement site is finally obtained.

[0043] The multi-scenario verification comprehensively covers various application scenarios in industrial sites, including scenarios with different measurement media such as liquids, gases, and steam; scenarios with different operating parameter fluctuation ranges of ±5%, ±10%, and ±20% of rated parameters; and scenarios with different operating lengths of flow meters, such as new instruments, flow meters that have been running for 1 year, 3 years, and 5 years.

[0044] The core objective of model parameter tuning is to minimize both the absolute and relative errors of traffic measurement in each validation scenario. The model parameters are iteratively optimized using a dedicated gradient descent algorithm.

[0045] The self-learning algorithm scheduling module continuously and online iteratively optimizes the parameter weights of the optimal error compensation algorithm model based on a joint dataset of real-time updated traffic flow and operating conditions, ensuring that the model always adapts to dynamically changing on-site conditions. Simultaneously, this module can dynamically adjust the computation frequency of the intelligent calibration module and the system's calibration strategy according to real-time changes in on-site conditions. For example, it reduces the computation frequency to save system resources when operating conditions are stable, and increases the computation frequency and switches to a real-time calibration strategy when operating conditions fluctuate drastically, ensuring a balance between calibration accuracy and system resource utilization.

[0046] In some embodiments, the intelligent calibration calculation module has a closed-loop iterative optimization function for the calibration effect throughout the entire process, forming a calibration, feedback, optimization, and recalibration system.

[0047] After completing the first correction of the standardized flow data, the module does not complete the calibration in one go, but continuously compares the calibrated flow data with the flow reference value of the real-time matching working condition in a dynamic and uninterrupted manner. The comparison time interval can be customized, and the module generates calibration effect feedback data that includes multiple dimensions such as error value, error change rate, and calibration accuracy.

[0048] The generated calibration effect feedback data is subjected to a second extraction of effective sensor signals to remove noise interference in the feedback data. At the same time, the quantization value of the flow error characteristics is re-analyzed and updated to ensure the accuracy of the feedback data.

[0049] The accurate calibration effect feedback data after secondary processing is transmitted back to the error tracing unit and dynamic compensation unit of the intelligent calibration calculation module in real time and synchronously, serving as the core basis for adjusting and optimizing the parameters of the two units.

[0050] The error tracing unit dynamically and adaptively adjusts the judgment criteria and thresholds for error identification based on the feedback data, thereby improving the accuracy of error identification. The dynamic compensation unit refines and optimizes the parameter settings and weight allocation of the dynamic error compensation model based on the feedback data.

[0051] Through the above iterative operations, the correction effect of the flow data is continuously optimized to ensure that the measurement error of the flow meter remains stable within the preset accuracy threshold range during long-term operation, thus achieving continuous self-optimization of the calibration effect.

[0052] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, substitutions, or improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A flow meter data adaptive calibration system, characterized in that, include: The system includes a flow signal acquisition module, an operating condition parameter monitoring module, a main control processing unit, and an intelligent calibration calculation module. The main control processing unit is bidirectionally connected to the flow signal acquisition module, the operating condition parameter monitoring module, and the intelligent calibration calculation module. The flow signal acquisition module processes the raw measurement signal output by the flow meter to generate standardized flow data, and simultaneously integrates multiple types of raw signals from the flow meter to construct a raw flow sensing dataset. Effective sensing signals are then extracted from this dataset. The operating condition parameter monitoring module performs real-time acquisition and validity verification of multi-dimensional operating condition parameters at the flow measurement site, and simultaneously completes spatial registration and parameter feature fitting modeling of operating condition parameters. The main control processing unit realizes the fusion analysis of traffic data and operating parameters and algorithm scheduling. It fuses standardized traffic data with spatially registered operating parameters to generate a spatiotemporally fused traffic and operating condition joint dataset. The fused dataset is divided into an algorithm training set and an algorithm verification set according to a preset ratio. Based on the above division of the fused dataset, the training iteration and verification optimization of the calibration algorithm model are completed. The intelligent calibration calculation module relies on the partitioning results of the fused dataset to achieve dynamic adaptive correction of flow measurement data through quantitative analysis of flow error characteristics. The modules work together to complete the real-time calibration of the flow meter without human intervention.

2. The flow meter data adaptive calibration system according to claim 1, characterized in that, The intelligent calibration calculation module has a built-in error tracing unit, a dynamic compensation unit, and a correction execution unit. The error tracing unit compares the standardized flow data with the flow reference value adapted to the operating conditions in real time, identifies the cause of the error and classifies the error into two categories: operating condition fluctuation error and instrument drift error. At the same time, it uses the flow error feature area calibration method and the flow error feature contour analysis algorithm to extract and quantify the flow error features and generate the flow error feature quantification value. The dynamic compensation unit constructs an error dynamic compensation model that matches the on-site working conditions based on the identified error causes, the actual changes in on-site operating parameters, and the quantified value of flow error characteristics. The correction execution unit performs point-by-point optimization on the standardized flow data based on the real-time correction coefficients output by the error dynamic compensation model, thereby achieving differentiated calibration of operating condition fluctuation errors and instrument drift errors.

3. The flow meter data adaptive calibration system according to claim 1, characterized in that, The operating condition parameter monitoring module adopts a multi-sensor synchronous acquisition architecture, which synchronously acquires multi-dimensional operating condition parameters such as medium temperature, pipeline pressure, and fluid characteristics at the flow measurement site through sensor acquisition units with different functions. The collected multi-source operating condition parameters are sequentially subjected to time series smoothing and outlier intelligent removal to generate a continuous and effective time series dataset of operating condition parameters. A reference domain for fitting the features of operating parameters is constructed based on the time series dataset of operating parameters. Spatial registration of operating parameters is completed by centering the reference domain, and a spatially registered dataset of operating parameters is generated and synchronously transmitted to the main control processing unit. The operating condition parameter monitoring module has a built-in medium adaptation unit, which can dynamically adjust the acquisition frequency and abnormal value judgment threshold of the operating condition parameters according to the type of the measured medium, and adapt to the operating condition change characteristics of different measured media such as liquids, gases and steam.

4. The flow meter data adaptive calibration system according to claim 1, characterized in that, The main control processing unit is equipped with a data fusion engine and a self-learning algorithm scheduling module; The data fusion engine is equipped with a spatiotemporal feature association algorithm. It first performs feature layer association matching on the received standardized traffic data and spatially registered operating condition parameter dataset and extracts the association feature values ​​between traffic and operating conditions. Then, it completes the data layer fusion and splicing based on the association feature values, while removing invalid data that is mismatched in the spatiotemporal dimension, and generating the spatiotemporally fused traffic and operating condition joint dataset. The self-learning algorithm scheduling module performs flow and operating condition feature annotation processing on the partitioned algorithm training set and generates a feature annotation dataset. The algorithm training set and the feature annotation dataset are then used to perform full-dimensional training on the initial error compensation algorithm model.

5. The flow meter data adaptive calibration system according to claim 1, characterized in that, The flow signal acquisition module integrates a signal adaptation unit, a signal conditioning unit, and a digital-to-analog conversion unit; The signal adapter unit is compatible with the input of various raw measurement signals, including pulse, analog, and digital signals, from different types of flow meters, and integrates these raw measurement signals to construct a set of raw flow sensing data. The signal conditioning unit sequentially performs filtering, amplification, and anti-interference processing on the original measurement signal, and simultaneously performs targeted extraction of effective sensing signals on the original flow sensing data set, completing the separation and noise reduction of flow sensing features, and generating noise-reduced flow sensing feature data. The digital-to-analog conversion unit converts the analog signal processed by the signal conditioning unit into standardized digital flow data, adds a timestamp to the standardized digital flow data, and simultaneously associates and binds the flow sensing feature noise reduction data with the standardized digital flow data using timestamps. The flow signal acquisition module has a built-in signal adaptation and self-adjustment unit, which can automatically adjust the signal adaptation parameters and sensing feature extraction strategy according to the type of flow meter connected, and adapt to the signal output characteristics of different types of flow meters such as electromagnetic, vortex, and differential pressure.

6. The flow meter data adaptive calibration system according to claim 1, characterized in that, The system is also equipped with a data interaction module; The data interaction module receives the calibrated flow data, raw flow data, operating parameters and flow error characteristic analysis data output by the main control processing unit in real time. The local display unit displays the above data in real time in the form of numerical values, trend graphs and feature fitting graphs. At the same time, the multi-protocol communication unit performs encryption processing on various types of data and uploads them to the remote monitoring platform. The data interaction module stores the complete set of raw flow sensor data and calibration algorithm model-related data, providing underlying data support for the iterative upgrade of the calibration algorithm model.

7. The flow meter data adaptive calibration system according to claim 2, characterized in that, The dynamic compensation unit constructs a real-time response compensation sub-model for operating condition fluctuation error and a trend fitting compensation sub-model for instrument drift error based on the differentiated characteristics of error types. The real-time response compensation sub-model makes rapid coefficient adjustments for instantaneous changes in operating parameters, and the trend fitting compensation sub-model makes trend-based compensation corrections based on the long-term variation law of instrument drift. The dynamic compensation unit integrates the two sub-models into a unified error dynamic compensation model through dynamic weight allocation, adapting to the changing characteristics of different error types.

8. The flow meter data adaptive calibration system according to claim 4, characterized in that, The self-learning algorithm scheduling module uses the partitioned algorithm validation set to perform multi-scenario validation and parameter tuning on the trained initial error compensation algorithm model, so as to obtain the optimal error compensation algorithm model that is adapted to the field. The multi-scenario verification includes verification scenarios with different measurement media, different operating condition parameter fluctuation ranges, and different flow meter running times. Parameter optimization takes minimizing the flow measurement error under each scenario as the core objective. The self-learning algorithm scheduling module continuously iteratively optimizes the parameter weights of the optimal error compensation algorithm model based on the real-time updated spatiotemporal fusion traffic and operating condition joint dataset, and dynamically adjusts the operation frequency and calibration strategy of the intelligent calibration operation module according to the real-time changes in the on-site operating conditions.

9. The flow meter data adaptive calibration system according to claim 2, characterized in that, The intelligent calibration calculation module has a closed-loop iterative optimization function for calibration effect. After completing the first round of standardized flow data correction, it continuously compares the calibrated flow data with the flow reference value that is adapted to the real-time operating conditions to generate multi-dimensional calibration effect feedback data. The calibration effect feedback data is subjected to secondary extraction of effective sensor signals and re-analysis of flow error feature quantification values. The calibration effect feedback data after secondary processing is then sent back to the error tracing unit and the dynamic compensation unit. The error tracing unit dynamically adjusts the error identification criteria based on feedback data, and the dynamic compensation unit optimizes the parameter settings of the error dynamic compensation model based on feedback data, continuously iterating and optimizing the correction effect of flow data to ensure that the measurement error of the flow meter is always within the preset threshold range.

Images