Dynamic compensation method for low temperature differential pressure liquid level meter

Abstract

The application discloses a dynamic compensation method for a low-temperature differential pressure liquid level meter, and has the characteristics that the original signal of a low-temperature medium measuring container is acquired, the original signal including an original differential pressure, a liquid-phase temperature, a gas-phase temperature, a measuring pipeline system temperature and a gas-phase pressure; after the original signal is processed, a key feature vector is extracted, the key feature vector including a temperature difference type first-level derivative feature obtained from the liquid-phase temperature, the gas-phase temperature and the measuring pipeline system temperature, and a time change rate type second-level derivative feature of each temperature and the first-level derivative feature; the processed original signal and the key feature vector are input into a pre-trained multi-physical field coupling analysis model, and a dynamic compensation factor is output; the original differential pressure is corrected by using the dynamic compensation factor, and the liquid level value of the low-temperature medium in the measuring container is calculated in combination with the liquid-phase medium density and the gas-phase medium density. The application realizes dynamic and comprehensive compensation of the multi-physical field coupling error in a low-temperature environment, and significantly improves the precision, dynamic response and working condition adaptability of the low-temperature medium liquid level measurement.

Description

Technical Field

[0001] This application relates to the field of liquid level measurement technology, and more specifically, to a dynamic compensation method for a cryogenic differential pressure liquid level gauge. Background Technology

[0002] Differential pressure level gauges are commonly used instruments in industrial processes to measure the liquid level in storage tanks and containers. They calculate the liquid level by measuring the pressure difference between the bottom of the container and a reference pressure in the gas phase, combined with the density of the medium. This technology is relatively mature in applications under normal temperature or stable conditions. However, in the storage, transportation, and handling of cryogenic media such as liquefied natural gas, liquid nitrogen, and liquid oxygen, the measurement data from conventional differential pressure level gauges deviates significantly from the actual results, making it difficult to meet the requirements for high-precision metering and fine control.

[0003] The closest compensation scheme in the field is the static density compensation method based on medium temperature. This scheme adds a temperature sensor of the immersed medium to the differential pressure transmitter, obtains the density value by looking up the pre-stored medium temperature-density curve, and calculates the liquid level by combining it with the original differential pressure. Some improved schemes only add simple zero-point temperature drift compensation of the instrument.

[0004] However, in actual low-temperature industrial environments, this static density compensation scheme has significant technical flaws, as follows: 1.1 Ignoring multiphysics coupling effects leads to a one-sided compensation model. Existing compensation schemes only attribute measurement errors under low-temperature conditions to density changes caused by changes in medium temperature, failing to recognize that measurement errors in low-temperature environments are the result of the combined effects of multiple physical field factors. Their compensation logic is designed solely around the single correlation between medium temperature and density.

[0005] 1.2 The static compensation method lacks dynamic adaptability. Existing compensation schemes rely on pre-set, fixed temperature-density curves for table lookup calculations. Their entire compensation logic is based on the ideal assumption that the medium temperature is uniformly distributed and the measurement system and the surrounding environment are in thermal equilibrium, which is out of touch with the actual working conditions in low-temperature industrial sites.

[0006] In summary, existing compensation schemes for cryogenic differential pressure level gauges, due to their design of compensation logic around a single physical field and the use of a completely static compensation method, cannot adapt to the complex error formation patterns and dynamic operating condition changes in cryogenic environments. They are unable to solve the problem of multi-factor coupled measurement errors under cryogenic conditions and cannot meet the practical application requirements for the measurement accuracy and stability of differential pressure level gauges in the storage, transportation, and handling of cryogenic media such as liquefied natural gas and liquid nitrogen. Summary of the Invention

[0007] The main objective of this application is to provide a dynamic compensation method for a cryogenic differential pressure level gauge to solve the problem of insufficient accuracy in level measurement for cryogenic containers in related technologies.

[0008] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description or may be learned by practice of this application.

[0009] According to a first aspect of this application, a dynamic compensation method for a cryogenic differential pressure level gauge is provided, comprising: Acquire the raw signals from the cryogenic medium measurement container, including the raw differential pressure, liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and gas phase pressure; After processing the original signal, key feature vectors are extracted. The key feature vectors include first-level derived features of temperature difference obtained from liquid phase temperature, gas phase temperature, and temperature of the measuring pipeline system, as well as second-level derived features of time change rate of each temperature and the first-level derived features. The processed original signal and key feature vectors are input into a pre-trained multiphysics coupling analysis model, which outputs a dynamic compensation factor. The original differential pressure is corrected using a dynamic compensation factor, and the liquid level of the cryogenic medium in the measuring container is calculated by combining the density of the liquid phase medium and the density of the gas phase medium.

[0010] The dynamic compensation method for cryogenic differential pressure level gauges in this application integrates multi-dimensional raw signals including original differential pressure, liquid / gas / pipeline temperatures, and gas phase pressure, overcoming the limitation of existing technologies that only collect single-medium temperature data. Simultaneously, by extracting first-level derived features of temperature difference and second-level derived features of time change rate, the static state and dynamic trend of multi-physics fields in cryogenic measurements are transformed into model-recognizable feature vectors. Then, a dynamic compensation factor is generated using a pre-trained multi-physics coupling analysis model to correct the original differential pressure. Finally, the liquid level value is calculated by combining the liquid and gas phase media densities. This application forms a complete multi-physics dynamic compensation logic from data acquisition and feature extraction to compensation calculation, thoroughly solving the core problems of existing technologies that only compensate for medium density changes and whose static models cannot adapt to cryogenic dynamic conditions. It achieves dynamic and comprehensive compensation for multi-physics coupling errors in cryogenic environments, significantly improving the accuracy, dynamic response, and adaptability of cryogenic medium level measurements.

[0011] In one exemplary embodiment of this application, the original signal of the cryogenic medium measurement container is obtained by synchronously acquiring the original differential pressure, liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and gas phase pressure.

[0012] By using the above settings, the time difference error in the acquisition process of each physical quantity signal is eliminated through synchronous acquisition, ensuring the spatiotemporal consistency of multi-physics data. This makes the basic data for subsequent feature extraction and coupled model calculation more accurate, avoids feature distortion and compensation factor calculation deviation caused by asynchronous data acquisition, improves the reliability of the input data of the entire compensation model, and lays the data foundation for subsequent accurate compensation.

[0013] In an exemplary embodiment of this application, the first-level derived features of the temperature difference category include gas-liquid temperature difference, pipeline-liquid phase temperature difference, and pipeline-gas phase temperature difference. The gas-liquid temperature difference is the difference between the gas phase temperature and the liquid phase temperature. The pipeline-liquid phase temperature difference is the difference between the temperature of the measuring pipeline system and the liquid phase temperature. The pipeline-gas phase temperature difference is the difference between the temperature of the measuring pipeline system and the gas phase temperature.

[0014] Through the above settings, three core physical states in the cryogenic measurement scenario are accurately quantified: the vertical temperature gradient between gas and liquid inside the container, the thermal imbalance between the measurement system and the measured medium, and the heat exchange between the measurement system and the gas phase environment. This directly and concretely characterizes the core error sources of multi-physics coupling in cryogenic liquid level measurement, enabling the multi-physics coupling analysis model to specifically identify and capture different types of error characteristics. This provides clear and effective characteristic basis for the accurate generation of dynamic compensation factors, and improves the model's ability to identify and correct errors.

[0015] In one exemplary embodiment of this application, the second-level derived feature of the time change rate class is the rate of change of the liquid phase temperature, the gas phase temperature, the temperature of the measuring pipeline system, and the gas-liquid temperature difference, the pipeline-liquid phase temperature difference, and the pipeline-gas phase temperature difference over time.

[0016] Through the above settings, the dynamic thermal imbalance process caused by scenarios such as tank filling / discharging and sudden environmental changes under low-temperature conditions can be quantitatively captured. The dynamic change trend of temperature and temperature difference can be accurately characterized, enabling the multi-physics coupling analysis model to respond to the dynamic changes of the low-temperature field in real time. This completely solves the problem that the existing static compensation model cannot adapt to dynamic conditions, effectively reduces the dynamic error and hysteresis of the measured values ​​under varying conditions, and improves the stability and real-time performance of liquid level measurement under dynamic conditions.

[0017] In one exemplary embodiment of this application, the multiphysics coupling analysis model is a machine learning model or a gray box model that integrates thermodynamic mechanisms, wherein the machine learning model includes at least one of a neural network model and a gradient boosting tree model.

[0018] Through the above settings, the gray box model, which integrates thermodynamic mechanisms, takes into account both the physical laws of low-temperature measurement and the advantages of data fitting, ensuring that the compensation logic conforms to actual thermodynamic and mechanical principles. Machine learning models such as neural networks and gradient boosting trees can accurately fit the complex nonlinear mapping relationship between multi-physics field variables and composite measurement errors, adapting to the complex error laws of multi-physics field coupling in low-temperature environments. Different model types can adapt to the computing resources, data volume, and real-time requirements of different industrial sites, significantly improving the scenario adaptability of this compensation method. At the same time, all types of models can ensure the accuracy of dynamic compensation factor generation, meeting the compensation needs of different low-temperature operating conditions.

[0019] In one exemplary embodiment of this application, a dynamic compensation factor is used to correct the composite measurement error caused by changes in gas-liquid density difference, temperature drift of the sensor and measuring pipeline, gas phase thermal stratification effect, and dynamic thermal hysteresis.

[0020] Through the above settings, comprehensive coverage of the core composite error sources in cryogenic liquid level measurement is achieved, completely solving the problem that existing technologies can only compensate for single errors caused by density difference due to changes in medium temperature. Various errors caused by multi-physical field coupling are specifically corrected, realizing comprehensive and synchronous correction of composite measurement errors. This effectively eliminates residual measurement errors under operating conditions such as sensor cold start, changes in ambient temperature gradient, and system pressure fluctuations, further improving the accuracy of liquid level measurement.

[0021] In one exemplary embodiment of this application, the density of the liquid phase medium and the density of the gas phase medium are obtained in the following manner: The density of the liquid medium is obtained by querying a physical property database based on the liquid phase temperature and the gas phase pressure. The density of the gaseous medium is obtained by querying a physical property database based on the gaseous temperature and gaseous pressure.

[0022] By taking into full account the combined effects of temperature and pressure on the density of cryogenic media, the problem of existing technologies that only look up the density from the medium temperature and ignore the influence of pressure on the density is solved, which greatly improves the accuracy of obtaining the density of liquid and gaseous media. At the same time, obtaining the density of gaseous media separately makes up for the deficiency of existing technologies that only consider the density of liquid phase, so that the selection of medium density is more in line with the actual working conditions of cryogenic containers, laying a key parameter foundation for the subsequent accurate calculation of liquid level.

[0023] In one exemplary embodiment of this application, the original differential pressure is corrected using the dynamic compensation factor, and the liquid level value of the cryogenic medium in the measuring container is calculated by combining the liquid phase medium density and the gas phase medium density, specifically: The original differential pressure is corrected by the dynamic compensation factor to obtain the compensated differential pressure ΔP_comp; The liquid level H is calculated according to the formula H=ΔP_comp / [(ρ_l-ρ_v)×g], where ρ_l is the density of the liquid phase medium, ρ_v is the density of the gas phase medium, and g is the gravitational acceleration.

[0024] The above settings clarify the liquid level calculation logic based on the difference in pressure after compensation combined with the gas-liquid density difference. Compared with the existing technology that only uses the liquid phase density to calculate the liquid level, this fully considers the actual influence of the gas phase medium on the pressure difference inside the container, making the physical model of liquid level calculation more consistent with the pressure transmission law of cryogenic containers. It corrects the systematic error caused by neglecting the gas phase density in the traditional calculation formula, realizes the accurate calculation of liquid level value, and makes the calculation results more consistent with the actual liquid level height inside the cryogenic container.

[0025] In one exemplary embodiment of this application, after calculating the liquid level value, the method further includes self-learning and optimization of the multiphysics coupling analysis model: The calculated liquid level value is compared with the reading of an external high-precision standard liquid level gauge to obtain a real-time error sequence; The parameters of the multiphysics coupling analysis model are fine-tuned and updated using the real-time error sequence and its corresponding multiphysics state data as training samples.

[0026] By incorporating feedback from an external high-precision standard through the above setup, a closed-loop model optimization mechanism of compensation-comparison-optimization is constructed. This solves the problems of existing technologies where compensation parameters remain unchanged and cannot adapt to slow time-varying factors such as long-term sensor aging, media characteristic drift, and pipeline characteristic changes. It enables continuous autonomous optimization of the multi-physics coupling analysis model, allowing the compensation model to track changes in operating conditions at low temperatures in real time, effectively offsetting the accuracy decay caused by slow time-varying factors, and ensuring the long-term accuracy and reliability of liquid level measurement.

[0027] In one exemplary embodiment of this application, the parameters of the multiphysics coupling analysis model are fine-tuned and updated using the real-time error sequence and its corresponding multiphysics state data as training samples, including: The real-time error sequence and its corresponding multiphysics state data are used as training samples. The training samples are used to drive an online learning algorithm, and at least one of incremental learning and sliding window training techniques are used to fine-tune and update the parameters of the multiphysics coupling analysis model.

[0028] By adopting the above settings, the problems of large computational load, poor real-time performance, and easy loss of recent operating conditions that exist in traditional full training are avoided. The model parameters are only lightly fine-tuned using new training samples, which greatly improves the real-time performance and computational efficiency of model optimization and adapts to the real-time compensation needs of industrial sites. At the same time, sliding window training can focus on the recent operating conditions of the low-temperature site, and incremental learning realizes the gradual optimization of the model, so that the model always fits the changing patterns of the actual measurement conditions, further ensuring the long-term stability and accuracy of low-temperature liquid level measurement.

[0029] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0030] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 This is a flowchart illustrating the dynamic compensation method according to the embodiments of this application; Figure 2 This is a schematic diagram of the dynamic compensation system according to the embodiments of this application; Figure 3 This is a schematic diagram of the connection between the cryogenic medium measurement container and the sensing module according to an embodiment of this application; The system comprises: 1. Sensing module; 100. Temperature acquisition unit; 110. First differential pressure acquisition unit; 120. Pressure acquisition unit; 130. Standard differential pressure acquisition unit; 2. Data acquisition and processing module; 3. Model processing module; 4. Liquid level calculation module; 5. Low-temperature medium measurement container; 6. Liquid phase temperature sensor; 7. Gas phase temperature sensor; 8. Gas phase pressure sensor; 9. First measurement pipeline; 10. Measurement pipeline system temperature sensor; 11. First differential pressure level gauge; 12. Standard differential pressure acquisition unit; 13. Second measurement pipeline. Detailed Implementation

[0031] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and therefore their detailed descriptions will be omitted. Furthermore, the drawings are merely illustrative of this application and are not necessarily drawn to scale.

[0032] Although relative terms such as "upper" and "lower" are used in this specification to describe the relative relationship of one component of an icon to another, these terms are used only for convenience, such as according to the orientation of the examples in the accompanying drawings. It is understood that if the device of the icon is flipped so that it is upside down, the component described as "upper" will become the component described as "lower." When a structure is "upper" of another structure, it may mean that the structure is integrally formed on the other structure, or that the structure is "directly" mounted on the other structure, or that the structure is "indirectly" mounted on the other structure through another structure.

[0033] The terms “a,” “one,” “the,” and “at least one” are used to indicate the existence of one or more elements / components / etc.; the terms “including” and “having” are used to indicate an open-ended inclusion and to mean that there may be other elements / components / etc. in addition to the listed elements / components / etc.; the terms “first” and “second” are used only as markers and are not a limitation on the number of objects.

[0034] Furthermore, the terms "set up," "equipped with," "connected," and "fixed" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0035] In addition, the term "multiple" should mean two or more.

[0036] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0037] To address the relevant technical issues, the following detailed description of the dynamic compensation method for the cryogenic differential pressure level gauge, based on the technical solution of this application, is provided. The compensation method of this embodiment is applicable to the storage, transportation, and processing of cryogenic media such as liquefied natural gas (LNG), liquid nitrogen, liquid oxygen, and cryogenic chemical raw materials. It achieves multi-physical field dynamic compensation for differential pressure level gauges in cryogenic medium measurement containers, effectively improving the accuracy, dynamic response, and adaptability of level measurement under cryogenic conditions. This solves the technical problem of existing static density compensation schemes having a single compensation factor and lacking dynamic adaptability.

[0038] The dynamic compensation method for cryogenic differential pressure level gauges disclosed in this application primarily involves acquiring multi-dimensional physical quantity signals under cryogenic operating conditions, extracting feature vectors that characterize the coupling state and dynamic changes of multiple physics fields, and generating dynamic compensation factors by combining these with a pre-trained multi-physics field coupling analysis model. This process accurately corrects the original differential pressure signal before calculating the level value. The method can be further enhanced with model self-learning and optimization steps to achieve continuous iteration of the compensation model and ensure long-term measurement accuracy. In the following embodiments, the measuring container is a conventional cryogenic storage device such as a storage tank or sealed container for storing and transporting cryogenic media. The multi-physics field coupling analysis model is a model pre-trained to convergence based on multi-physics field sample data from cryogenic level measurements, possessing the ability to identify composite measurement errors under cryogenic conditions and generate compensation factors.

[0039] Specifically, this embodiment provides a dynamic compensation method for a cryogenic differential pressure level gauge, comprising: Acquire the raw signals from the cryogenic medium measurement container, including the raw differential pressure, liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and gas phase pressure; The core of this step is to acquire the multi-dimensional raw physical quantity signals required for measuring the liquid level in the cryogenic medium measurement container. The raw signals include raw differential pressure, liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and gas phase pressure. To ensure the accuracy of subsequent feature extraction and model calculation, all the above raw signals are acquired synchronously to eliminate the feature distortion problem caused by the time difference of different signal acquisition. The time accuracy of synchronous acquisition can be set according to the dynamic requirements of cryogenic conditions. Under normal cryogenic storage and transportation conditions, the acquisition time synchronization error can be controlled within 10ms to meet the requirements.

[0040] To obtain the corresponding physical quantity signals, appropriate sensors can be placed on the cryogenic medium container. For example, the original differential pressure is collected by a differential pressure level gauge installed on the measuring container, and the liquid phase temperature is collected by a temperature sensor immersed in the liquid phase region at the bottom of the measuring container, directly reflecting the body temperature of the measured cryogenic medium. The gas phase temperature is collected by a temperature sensor installed in the gas phase space at the top of the measuring container, reflecting the gas phase ambient temperature inside the container. Temperature sensors attached to the pressure core of the differential pressure transmitter, the main pressure pipeline, or the valve assembly are used to collect data, characterizing the thermal state of the entire pressure transmission path. The pressure in the gas phase space at the top of the measuring container, i.e., the gas phase reference pressure, is collected by a gas phase pressure transmitter.

[0041] The temperature sensors mentioned above can be high-precision temperature detection elements such as Pt100 and temperature-sensing diodes that are suitable for low-temperature conditions. The differential pressure level gauge and gas phase pressure transmitter are industrial-grade detection instruments that are suitable for low-temperature and high-pressure conditions to ensure the accuracy and stability of the original signal acquisition.

[0042] The acquired raw signals may contain invalid signals due to electromagnetic interference in the industrial environment and instrument noise. Therefore, the raw differential pressure, liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and gas phase pressure acquired synchronously are preprocessed. The preprocessing operation includes one or more of the following: filtering, noise reduction, and normalization of the raw signals. Among them, filtering can be done by means of median filtering, Kalman filtering, etc. to eliminate impulse interference, noise reduction can be done by wavelet noise reduction to reduce the inherent noise of the instrument, and normalization can map the physical quantity signals to the same numerical range to eliminate the influence of dimensional differences on subsequent model calculations.

[0043] Key feature vectors are extracted from the preprocessed raw signal. These key feature vectors include first-level derived features based on temperature difference and second-level derived features based on time change rate. These two types of derived features characterize the coupling state of multiple physical fields under low-temperature conditions from both static and dynamic perspectives, providing accurate feature basis for the generation of compensation factors. The specific extraction method is as follows: 1. Extract first-level derived features of temperature difference. These characteristics are calculated from the liquid phase temperature (T_liquid), gas phase temperature (T_vapor), and the temperature of the measuring pipeline system (T_pipe). Specifically, they include three types of characteristics: gas-liquid temperature difference, pipeline-liquid phase temperature difference, and pipeline-gas phase temperature difference. The calculation formulas for each characteristic are as follows: The gas-liquid temperature difference ΔT_vl=T_vapor-T_liquid characterizes the intensity of the vertical temperature gradient inside the measuring container and reflects the core feature of the change in gas-liquid density difference. The pipe-liquid temperature difference ΔT_pl = T_pipe - T_liquid quantifies the thermal imbalance between the measurement system and the measured low-temperature medium, and is the core characterization feature of the sensor and pipe temperature drift. The pipe-gas phase temperature difference ΔT_pv = T_pipe - T_vapor reflects the heat exchange between the measurement system and the gas phase environment inside the container, and is a key characteristic of the gas phase thermal stratification effect.

[0044] 2. Extracting second-level derived features based on the rate of change over time. These characteristics include the rates of change of liquid phase temperature, gas phase temperature, measuring pipeline system temperature, and the aforementioned gas-liquid temperature difference, pipeline-liquid phase temperature difference, and pipeline-gas phase temperature difference over time. They characterize the dynamic trends of each physical quantity and temperature difference feature, accurately capturing the thermal imbalance process under dynamic operating conditions such as tank filling, discharge, and sudden environmental changes. The calculation method for the time-varying rates of change of each physical quantity and temperature difference feature is as follows: Select two consecutive synchronous acquisition times t1 and t2, calculate the numerical difference of the same feature at the two times, and then the ratio of this difference to the time interval Δt (t2-t1) is the time change rate of the feature during that time interval. The formula is: k=(X2-X1) / Δt, where X is the value of the corresponding physical quantity or temperature difference feature, and k is the time change rate.

[0045] The pre-processed raw signals (raw differential pressure, liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and gas phase pressure) and all extracted key feature vectors are input into a pre-trained multiphysics coupling analysis model. The model then outputs a dynamic compensation factor in real time based on the input feature data.

[0046] The multiphysics coupling analysis model can be either a machine learning model or a gray box model that integrates thermodynamic mechanisms. The machine learning model includes at least one of a neural network model and a gradient boosting tree model. The neural network model can be a BP neural network, an LSTM neural network, etc., which is suitable for the analysis and calculation of dynamic time-series features. The gradient boosting tree model can be an XGBoost, LightGBM, etc., which has efficient nonlinear feature fitting capabilities. The gray box model that integrates thermodynamic mechanisms combines the physical mechanism formulas of low-temperature heat transfer and fluid mechanics with a data-driven fitting model, taking into account both the rationality of physical laws and the adaptability to field data.

[0047] The aforementioned multiphysics coupling analysis model, through pre-training, has learned the complex nonlinear mapping relationship between various physical quantities, derived characteristics, and composite errors of the measurement system under low-temperature conditions. Therefore, the output dynamic compensation factor is a multidimensional vector that can be used to correct the composite measurement error introduced by the changes in gas-liquid density difference, temperature drift of sensors and measurement pipelines, gas phase thermal stratification effect, and dynamic thermal hysteresis, providing a core basis for the accurate correction of the original differential pressure.

[0048] Then, based on the temperature and pressure parameters in the original signal, the accurate liquid and gas phase media densities are obtained. A dynamic compensation factor is then used to correct the original differential pressure. Finally, the actual liquid level of the cryogenic medium inside the measuring container is calculated using physical formulas based on the liquid and gas phase media densities. The specific implementation process is as follows: 1. Obtaining the density of the liquid and gaseous media: Based on the physical properties of low-temperature media, the density of the medium is affected by the interaction of temperature and pressure. Therefore, this embodiment obtains accurate density values ​​by querying a physical property database. Specifically: Based on the pre-treated liquid phase temperature and gas phase pressure, a matching query is performed in the pre-stored low-temperature medium property database to obtain the liquid phase medium density ρ_l under the corresponding working condition; Based on the pre-processed gas phase temperature and pressure, a matching query is performed in the same property database to obtain the gas phase medium density ρ_v under the corresponding operating conditions. The property database is a pre-stored three-dimensional correspondence database of temperature-pressure-density for the target cryogenic medium, constructed from basic experimental data or industry standard property data of cryogenic media, and can be switched according to the type of cryogenic medium in actual application.

[0049] 2. Correcting the original differential pressure and calculating the liquid level: First, the dynamic compensation factor output by the multiphysics coupling analysis model is used to nonlinearly correct the preprocessed original differential pressure, resulting in the compensated differential pressure ΔP_comp. Then, combined with the liquid phase medium density ρ_l and gas phase medium density ρ_v obtained above, the cryogenic medium liquid level H in the measuring container is obtained according to the physical formula for liquid level calculation. The specific calculation formula is: H=ΔP_comp / [(ρ_l-ρ_v)×g], where g is the acceleration due to gravity, which can be taken as the conventional industrial calculation value of 9.8N / kg. This formula fully considers the actual influence of gas phase medium density on the pressure difference in the container. Compared with the traditional calculation method that only considers liquid phase density, it is more in line with the pressure transmission law of cryogenic containers and effectively eliminates the systematic error caused by ignoring gas phase density.

[0050] The acquisition, processing, and calculation of raw signals can be centralized within a single field instrument (transmitter), or a distributed architecture combining edge sensing and centralized computing can be employed. Specifically, the intelligent sensing unit on the field side is only responsible for acquiring the raw signal and uploading it via a digital communication interface (such as IO-Link or WirelessHART). The centralized processing unit, located in the control room, edge gateway, or cloud platform, is responsible for running the multiphysics coupling analysis model, performing complex dynamic compensation calculations, and then distributing the compensated high-precision liquid level value H to the required systems. This embodiment reduces the design complexity and power consumption requirements of the field instrument, facilitating centralized management of multiple measuring points and unified model upgrades.

[0051] In one embodiment, based on the long-term accuracy requirements of cryogenic liquid level measurement, after calculating the liquid level value, a self-learning and optimization stage of the multiphysics coupling analysis model can be added. By constructing a closed-loop optimization mechanism, continuous fine-tuning and updating of model parameters can be achieved, allowing the model to adapt to the influence of slow time-varying factors such as sensor aging, media characteristic drift, and pipeline characteristic changes, thus ensuring the long-term accuracy and stability of liquid level measurement. The specific implementation steps are as follows: 1. Obtain the real-time error sequence By using the standard comparison interface configured on the measuring container, the liquid level value H calculated by this method is continuously or periodically compared with the liquid level standard value H_std measured by an external high-precision standard liquid level gauge. The measurement error e(t) = H - H_std at each comparison time is calculated. The continuous measurement errors are integrated in time sequence to obtain the real-time error sequence.

[0052] 2. Construct model training samples The aforementioned real-time error sequence is used as the label data for model optimization, and the multiphysics state data corresponding to the error sequence is used as the input data for model optimization to jointly construct the model training sample; the multiphysics state data includes the preprocessed original signal at the corresponding time and the extracted key feature vector.

[0053] 3. Fine-tuning and updating model parameters The constructed training samples are input into an online learning algorithm, and at least one of the techniques of incremental learning and sliding window training is used to fine-tune and update the parameters of the multiphysics coupling analysis model: Incremental learning: New training samples are gradually input into the model, and only local parameters of the model are iteratively optimized to achieve incremental updates of the model, avoiding the large amount of computation required for full training and improving optimization efficiency; Sliding window training: Set a fixed-length time window and select only the most recent training samples within the window to train the model, so that the model always fits the recent operating characteristics of the measurement container and adapts to the dynamic changes in operating conditions at low-temperature sites.

[0054] Through the above self-learning and optimization steps, the compensation model of this application can achieve the effect of "calibration upon installation and increasing accuracy with use", which greatly reduces the maintenance cost of manual on-site calibration, effectively offsets the measurement accuracy decay caused by slow time-varying factors, and further improves the long-term reliability of cryogenic liquid level measurement.

[0055] The dynamic compensation method for low-temperature differential pressure level gauges implemented in this embodiment overcomes the technical limitations of existing static density compensation schemes. By simultaneously acquiring multi-dimensional signals of liquid, gas, pipeline temperature, differential pressure, and gas phase pressure, it achieves comprehensive perception of multi-physical field information under low-temperature operating conditions. By extracting static features such as temperature difference and dynamic features such as time change rate, it accurately characterizes the coupling state and dynamic change trend of multi-physical fields. Through the dynamic compensation factor generated by the multi-physical field coupling analysis model, it achieves comprehensive and synchronous correction of composite measurement errors. Combining the density acquisition method based on the cross-influence of temperature and pressure with the liquid level calculation formula based on the gas-liquid density difference, it further improves the accuracy of liquid level calculation.

[0056] Meanwhile, the optional model self-learning and optimization mechanism establishes a closed-loop model iteration mechanism, ensuring long-term measurement accuracy. This method can achieve real-time and accurate liquid level measurement under complex cryogenic conditions such as sensor cold start, sudden changes in ambient temperature, system pressure fluctuations, and tank filling / discharging. It effectively reduces residual measurement errors, improves the stability and dynamic responsiveness of measured values, and fully meets the practical needs of high-precision metering and fine control in the storage, transportation, and handling of cryogenic media.

[0057] According to the second aspect of this application, such as Figure 2 and Figure 3 As shown, a dynamic compensation system for a cryogenic differential pressure level gauge is provided, mainly including a sensing module 1, a data acquisition and processing module 2, a model processing module 3, and a level calculation module 4. The modules cooperate with each other to complete the entire process of raw signal acquisition, feature extraction, compensation factor output, and level calculation. The specific connection relationship is as follows: the signal output terminal of the sensing module 1 is connected to the signal input terminal of the data acquisition and processing module 2, the output terminal of the data acquisition and processing module 2 is connected to the input terminal of the model processing module 3, and the output terminal of the model processing module 3 is connected to the input terminal of the level calculation module 4, realizing the step-by-step transmission and processing of signals.

[0058] Among them, the sensing module 1 serves as the signal acquisition source of the system, used to acquire various raw physical signals related to the cryogenic medium measurement container 5. It includes a temperature acquisition unit 100, a first differential pressure acquisition unit 110, and a pressure acquisition unit 120. Optionally, it also includes a standard differential pressure acquisition unit 130. Each unit has a clear division of labor and works together to ensure the comprehensiveness and accuracy of the raw signal acquisition.

[0059] Specifically, the temperature acquisition unit 100 is used to monitor the liquid phase temperature, gas phase temperature, and measurement pipeline system temperature of the cryogenic medium measurement container 5. It further includes three sensors, each adapted to the acquisition requirements of different temperature parameters: (1) Liquid phase temperature sensor 6: It is located in the liquid phase region of the low-temperature medium measuring container 5, preferably close to the middle of the container and far away from the container wall, so as to avoid the interference of the container wall temperature on the measurement results. It is used to accurately monitor the liquid phase temperature of the low-temperature medium and provide basic data for subsequent density calculation and gas-liquid temperature difference extraction. (2) Gas phase temperature sensor 7: It is located in the gas phase region of the low temperature medium measuring container 5, preferably near the top of the container and adjacent to the pressure acquisition unit 120, to monitor the temperature of the gas phase space inside the container, synchronously capture gas phase temperature changes, and assist in correcting gas phase density errors. (3) Temperature sensor 10 for measuring pipeline system: It is installed on the measuring pipeline connected to the low temperature medium measuring container 5. Preferably, there are two sensors, located on the first measuring pipeline 9 and the second measuring pipeline 13 respectively. They can comprehensively monitor the temperature distribution of the two measuring pipelines. The average value of the temperature values ​​monitored by the two measuring pipeline system temperature sensors 10 is used as the temperature value of the measuring pipeline system. This avoids the problem of inaccurate pipeline temperature drift compensation caused by single-point temperature measurement. It is used to monitor the real-time temperature of the measuring pipeline system and provide data support for the extraction of pipeline-liquid phase temperature difference and pipeline-gas phase temperature difference. The pressure acquisition unit 120 is used to monitor the gas phase pressure of the cryogenic medium measurement container 5. It includes a gas phase pressure sensor 8, which is located in the gas phase region of the cryogenic medium measurement container 5 and is arranged adjacent to the gas phase temperature sensor 7. This facilitates the synchronous acquisition of gas phase temperature and pressure signals, reduces signal transmission delay, and provides pressure parameters for subsequent queries of liquid phase medium density and gas phase medium density. The first differential pressure acquisition unit 110 includes a first differential pressure level gauge 11, which is connected to the cryogenic medium measurement container 5 and is used to monitor the original differential pressure of the cryogenic medium measurement container 5. The original differential pressure is the pressure difference generated by the liquid phase medium in the container, which is the core basic signal for liquid level calculation. Its measurement accuracy directly affects the final liquid level calculation result. Optionally, the standard differential pressure acquisition unit 130 is used to monitor the standard differential pressure of the cryogenic medium measurement container 5 as a high-precision comparison benchmark for error comparison in subsequent optimization modules, thereby improving the system's self-learning optimization effect. Specifically, the standard differential pressure acquisition unit 130 and the first differential pressure acquisition unit 110 are connected in parallel, and then connected to the liquid phase region and gas phase region of the cryogenic medium measurement container 5 respectively through the first measurement pipeline 9 and the second measurement pipeline 13. The first measurement pipeline 9 connects the liquid phase region of the container to the liquid phase interface of the two acquisition units, and the second measurement pipeline 13 connects the gas phase region of the container to the gas phase interface of the two acquisition units. This parallel connection method ensures that the two differential pressure acquisition units operate independently and that the acquired differential pressure signals correspond to the same measurement scenario, avoiding signal deviations caused by different connection methods.

[0060] The data acquisition and processing module 2 is connected to the acquisition units of the sensing module 1 and is used to receive various raw signals transmitted by the sensing module 1 (including liquid phase temperature, gas phase temperature, measurement pipeline system temperature, raw differential pressure, gas phase pressure, and optionally standard differential pressure), and to preprocess these raw signals and extract key feature vectors to provide standardized and structured input data for the subsequent model processing module 3.

[0061] Extraction of key feature vectors is the core function of this module, which includes first-level derived features of temperature difference and second-level derived features of time change rate. The specific extraction method is as follows: (1) First-level derived features of temperature difference: calculated based on the liquid phase temperature, gas phase temperature, and measurement pipeline system temperature obtained by the temperature acquisition unit 100, specifically including three types: gas-liquid temperature difference, pipeline-liquid phase temperature difference, and pipeline-gas phase temperature difference. ① Gas-liquid temperature difference: This is the difference between the gas phase temperature and the liquid phase temperature. It is used to reflect the temperature gradient between the gas and liquid phases in the container and to help correct measurement errors caused by changes in gas-liquid density difference. ② Pipeline-Liquid Phase Temperature Difference: This is the difference between the temperature of the measuring pipeline system and the temperature of the liquid phase. It is used to quantify the temperature difference between the measuring pipeline and the liquid medium and to correct the influence of pipeline temperature drift on differential pressure measurement. ③ Pipeline-Gas Phase Temperature Difference: This is the difference between the temperature of the measured pipeline system and the temperature of the gas phase. It is used to quantify the temperature difference between the measured pipeline and the gas phase medium, and to further supplement the basis for compensating for pipeline temperature drift.

[0062] (2) Second-level derived features of time change rate: Based on the above liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and the extracted gas-liquid temperature difference, pipeline-liquid phase temperature difference, and pipeline-gas phase temperature difference, the change rate of each over time is calculated to capture the dynamic change trend of each physical parameter, correct the measurement error caused by dynamic thermal hysteresis, and ensure the compensation accuracy of the system in the scenario of dynamic parameter change.

[0063] The preprocessed original signal and the extracted key feature vector are used together as input data for model processing module 3, and are synchronously transmitted to model processing module 3 by data acquisition and processing module 2.

[0064] The model processing module 3 is used to store the pre-trained multiphysics coupling analysis model and receive the processed original signal and key feature vector output by the data acquisition and processing module 2. After inputting them into the multiphysics coupling analysis model, it outputs a dynamic compensation factor. This dynamic compensation factor is used to correct the composite measurement error introduced by the changes in gas-liquid density difference, temperature drift of the sensor and measurement pipeline, gas phase thermal stratification effect and dynamic thermal hysteresis.

[0065] Specifically, the multiphysics coupling analysis model is a machine learning model or a gray box model that integrates thermodynamic mechanisms. The machine learning model includes at least one of a neural network model and a gradient boosting tree model. Optionally, a fusion model of a neural network model and a gradient boosting tree model can be used to balance the model's fitting accuracy and generalization ability.

[0066] The pre-training process of the multiphysics coupling analysis model is as follows: collect original signal samples, key feature vector samples, and corresponding actual liquid level standard values ​​and composite measurement error samples under different low-temperature conditions (different media, different temperatures, different pressures, and different liquid levels). Use the original signal samples and key feature vector samples as inputs and the corresponding dynamic compensation factor (calculated based on the actual liquid level standard value and measurement error) as outputs. Train, verify, and optimize the model until the compensation factor error output by the model meets the preset accuracy requirements. After training, store the model in the model processing module 3 for actual runtime.

[0067] During the actual operation of the system, the model processing module 3 receives the input data transmitted by the data acquisition and processing module 2 in real time, calls the pre-trained multiphysics coupling analysis model, calculates the dynamic compensation factor that is adapted to the current working condition through the model, and transmits the dynamic compensation factor to the liquid level calculation module 4 in real time.

[0068] The liquid level calculation module 4 is configured to use the dynamic compensation factor output by the model processing module 3 to correct the original differential pressure, and calculate the liquid level value of the low temperature medium in the measuring container by combining the liquid phase medium density and the gas phase medium density, thus completing the final liquid level measurement.

[0069] The liquid phase medium density and gas phase medium density are obtained as follows: based on the liquid phase temperature and gas phase temperature obtained by the temperature acquisition unit 100 and the gas phase pressure obtained by the pressure acquisition unit 120, a preset physical property database is queried to obtain the liquid phase medium density and gas phase medium density under the corresponding temperature and pressure conditions, respectively; the physical property database pre-stores the density parameters of different cryogenic media (such as liquid nitrogen, liquid oxygen, liquefied natural gas, etc.) under different temperature and pressure conditions, which can be adapted and called according to the actual measured medium type to ensure the accuracy of the density parameters.

[0070] The specific process of liquid level calculation consists of two steps: The first step is differential pressure correction: the original differential pressure acquired by the first differential pressure acquisition unit 110 is corrected by the dynamic compensation factor to eliminate the influence of composite measurement error and obtain the compensated differential pressure ΔP_comp. The second step is liquid level calculation: The liquid level value is calculated according to the preset formula, which is H=ΔP_comp / [(ρ_l-ρ_v)×g], where H is the liquid level value in the cryogenic medium measuring container 5, ρ_l is the density of the liquid phase medium, ρ_v is the density of the gas phase medium, and g is the acceleration due to gravity (taken as 9.8m / s²). Through this formula, the corrected differential pressure signal can be directly converted into liquid level height, realizing accurate liquid level calculation.

[0071] To further improve the long-term operating accuracy of the system and address the problem of model accuracy decline caused by factors such as changes in operating conditions and sensor aging, this system also includes an optimization module. This optimization module establishes signal connections with the liquid level calculation module 4, the model processing module 3, and the sensing module 1 (standard differential pressure acquisition unit 130), and is configured to perform self-learning and optimization of the multi-physics coupling analysis model to achieve dynamic updates of model parameters.

[0072] The specific working process of the optimization module is as follows: (1) Error comparison: The liquid level value calculated by the liquid level calculation module 4 is compared with the reading of the standard differential pressure acquisition unit 130 in the sensing module 1 (as a high-precision standard liquid level reference value), and the difference between the two is calculated to obtain the real-time error sequence; (2) Sample collection: Collect the real-time error sequence and the corresponding multi-physics state data (i.e., liquid phase temperature, gas phase temperature, measurement pipeline system temperature, original differential pressure, gas phase pressure, key feature vectors, etc. at the corresponding time) and use them as training samples; (3) Model update: The collected training samples are sent to the model processing module 3. The online learning algorithm is driven by the training samples. At least one of incremental learning and sliding window training is used to fine-tune and update the parameters of the multiphysics coupling analysis model. Incremental learning is used to gradually absorb the information of new samples to avoid the model forgetting the historical training results. Sliding window training is used to select recent training samples for model fine-tuning to ensure that the model can adapt to the latest working conditions, thereby maintaining the long-term measurement accuracy of the system.

[0073] Other embodiments of this application will readily conceive of by those skilled in the art upon consideration of the specification and practice of the embodiments thereof. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not claimed in this application. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this application are indicated by the appended claims.

Claims

1. A dynamic compensation method for a low-temperature differential pressure level gauge, characterized in that, include: Acquire the raw signals from the cryogenic medium measurement container, including the raw differential pressure, liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and gas phase pressure; After processing the original signal, key feature vectors are extracted. The key feature vectors include first-level derived features of temperature difference obtained from the liquid phase temperature, gas phase temperature, and temperature of the measuring pipeline system, as well as second-level derived features of the time change rate of each temperature and the first-level derived feature. The processed original signal and the key feature vector are input into a pre-trained multiphysics coupling analysis model, which outputs a dynamic compensation factor. The original differential pressure is corrected using the dynamic compensation factor, and the liquid level of the cryogenic medium in the measuring container is calculated by combining the liquid phase medium density and the gas phase medium density.

2. The dynamic compensation method according to claim 1, characterized in that, The acquisition of the original signal from the cryogenic medium measurement container involves synchronously acquiring the original differential pressure, liquid phase temperature, gas phase temperature, measurement pipeline system temperature, and gas phase pressure.

3. The dynamic compensation method according to claim 1, characterized in that, The first-level derived features of the temperature difference category include gas-liquid temperature difference, pipeline-liquid phase temperature difference, and pipeline-gas phase temperature difference. The gas-liquid temperature difference is the difference between the gas phase temperature and the liquid phase temperature. The pipeline-liquid phase temperature difference is the difference between the temperature of the measuring pipeline system and the liquid phase temperature. The pipeline-gas phase temperature difference is the difference between the temperature of the measuring pipeline system and the gas phase temperature.

4. The dynamic compensation method according to claim 3, characterized in that, The second-level derived features of the time-change rate category are the rates of change of the liquid phase temperature, the gas phase temperature, the temperature of the measuring pipeline system, the gas-liquid temperature difference, the pipeline-liquid phase temperature difference, and the pipeline-gas phase temperature difference over time.

5. The dynamic compensation method according to claim 1, characterized in that, The multiphysics coupling analysis model is a machine learning model or a gray box model that integrates thermodynamic mechanisms. The machine learning model includes at least one of a neural network model and a gradient boosting tree model.

6. The dynamic compensation method according to claim 1, characterized in that, The dynamic compensation factor is used to correct the combined measurement error introduced by changes in gas-liquid density difference, temperature drift of sensors and measuring pipelines, gas phase thermal stratification effect, and dynamic thermal hysteresis.

7. The dynamic compensation method according to claim 1, characterized in that, The density of the liquid phase medium and the density of the gas phase medium are obtained in the following manner: The density of the liquid medium is obtained by querying a physical property database based on the liquid phase temperature and the gas phase pressure. The density of the gaseous medium is obtained by querying a physical property database based on the gaseous temperature and gaseous pressure.

8. The dynamic compensation method according to claim 7, characterized in that, The process involves correcting the original differential pressure using the dynamic compensation factor and calculating the liquid level of the cryogenic medium inside the measuring container by combining the liquid phase density and the gas phase density. Specifically: The original differential pressure is corrected by the dynamic compensation factor to obtain the compensated differential pressure ΔP_comp; The liquid level H is calculated according to the formula H=ΔP_comp / [(ρ_l-ρ_v)×g], where ρ_l is the density of the liquid phase medium, ρ_v is the density of the gas phase medium, and g is the gravitational acceleration.

9. The dynamic compensation method according to claim 1, characterized in that, After calculating the liquid level value, the process also includes self-learning and optimization of the multiphysics coupling analysis model: The calculated liquid level value is compared with the reading of an external high-precision standard liquid level gauge to obtain a real-time error sequence; The parameters of the multiphysics coupling analysis model are fine-tuned and updated using the real-time error sequence and its corresponding multiphysics state data as training samples.

10. The dynamic compensation method according to claim 9, characterized in that, The parameters of the multiphysics coupling analysis model are fine-tuned and updated using the real-time error sequence and its corresponding multiphysics state data as training samples, including: The real-time error sequence and its corresponding multiphysics state data are used as training samples. The training samples are used to drive an online learning algorithm, and at least one of incremental learning and sliding window training techniques are used to fine-tune and update the parameters of the multiphysics coupling analysis model.

Images