Adsorption mass transfer front surface positioning method of molecular sieve drying tower for producing hydrogen by electrolyzing water

Abstract

The invention relates to the technical field of water electrolysis hydrogen production post-treatment, and discloses a water electrolysis hydrogen production molecular sieve drying tower adsorption mass transfer leading surface positioning method, which comprises the following steps: synchronously collecting bed axial temperature distribution vector, inlet gas flow and moisture content; calculating theoretical total adsorption heat release power serving as global physical constraint according to the inlet working condition; on the basis of the actually measured temperature field and preset model parameters, adsorption heat source distribution is preliminarily estimated through an inversion algorithm, an integral result is compared with theoretical power, the model parameters are automatically adjusted until deviation convergence, and then the corrected parameters are utilized to calculate a real adsorption heat source intensity field; and finally, analyzing the heat source field, and calculating a spatial second derivative of the heat source field to accurately identify the mass transfer leading edge position and the mass transfer area length. According to the method, the self-adaptive model with double constraints is constructed, so that real-time accurate positioning of the mass transfer leading edge surface and quantitative evaluation of the health degree of the adsorbent are realized, and the predictability, reliability and economical efficiency of operation are improved.

Description

Technical Field

[0001] This invention relates to the field of post-treatment technology for hydrogen production from water electrolysis, specifically a method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production from water electrolysis. Background Technology

[0002] In the hydrogen production process via water electrolysis, the deep purification of hydrogen is a crucial step in ensuring that the final product meets or exceeds fuel cell-grade standards and other industrial application standards. Molecular sieve drying towers play a core role in removing trace amounts of moisture from hydrogen. The stable operation of the drying tower and the effective utilization of its adsorption capacity directly depend on the accurate determination of the adsorption state within the bed, particularly the location of the mass transfer zone (MTZ).

[0003] Currently, the control strategies for molecular sieve drying towers in industrial settings mainly rely on preset fixed-time programs, with some systems supplemented by online dew point analyzers at the drying tower outlet as a monitoring method. However, both methods exhibit limitations when facing fluctuating operating conditions. Fixed-time control, primarily based on design conditions, struggles to dynamically adapt to changes in gas volume and humidity caused by adjustments in the hydrogen production load of the upstream electrolyzer, easily leading to deviations in switching timing. Furthermore, outlet dew point detection is essentially a passive, post-event feedback mechanism. When the instrument indicates an excessive dew point, it often means the adsorption mass transfer front has already penetrated the bed and affected the outlet gas quality. This lag in condition assessment prevents operators from knowing the specific degree of penetration within the bed and the true location of the adsorption saturation interface. This results in either premature switching for safety reasons, wasting the effective capacity of the molecular sieve, or delayed assessment leading to water vapor leakage, threatening the safety of downstream equipment.

[0004] Furthermore, existing technologies lack visualization methods for the adsorption process, meaning they cannot monitor the specific morphology, location, and migration rate of the adsorption mass transfer front within the bed in real time. This forces the regeneration operation of the drying tower to heavily rely on manual experience or excessive design margins. This data-unsupported decision-making model directly impacts the system's energy consumption and safety: overly conservative regeneration strategies lead to unnecessary and frequent heating and purging, wasting electricity and regeneration gas; and if subtle changes occur in operating conditions, inaccurate judgments based on experience may result in untimely regeneration, posing risks to product gas quality. Simultaneously, this black-box operation also restricts the full lifecycle management of molecular sieve adsorbents. Current lifespan predictions are often based simply on cumulative operating time or processed gas volume, failing to quantify and analyze performance degradation caused by dynamic adsorption load impacts or localized bed pulverization and aging during actual operation. This extensive management approach makes it difficult to identify the true health of the adsorbent, often leading to premature disposal, increased costs, or unexpected failure before reaching the expected lifespan, causing production interruptions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis. This method aims to solve the problems in existing technologies where the large heat capacity of the adsorption bed causes temperature monitoring signals to lag significantly behind the actual mass transfer process, making it impossible to accurately define the adsorption mass transfer front position. This can easily lead to resource waste caused by switching before the adsorbent is saturated or product gas quality exceeding standards due to adsorbent breakthrough. Furthermore, it lacks online assessment methods for the health of the adsorbent throughout its entire lifecycle.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis, comprising the following steps:

[0007] Step S100: Synchronous acquisition of multidimensional thermal-hydraulic data: Synchronously acquire the axial temperature distribution vector of the bed, radial temperature difference data, instantaneous inlet gas mass flow rate, and instantaneous inlet gas moisture content of the adsorption drying tower body.

[0008] Step S200: Calculation of macroscopic theoretical adsorption enthalpy change: Based on the instantaneous inlet gas mass flow rate, the instantaneous inlet gas moisture content, and the preset differential adsorption heat value of water molecules on the molecular sieve material, the theoretical total adsorption heat release power is calculated.

[0009] Step S300, Adaptive Correction and Inversion Solution of Model Parameters: Based on the axial temperature distribution vector of the bed and the preset model parameters, the distribution of adsorption heat sources is initially estimated through the inversion algorithm. The spatial integral result of the adsorption heat source distribution is compared with the theoretical total adsorption heat release power. If there is a deviation, the preset model parameters are automatically adjusted until the deviation converges within the preset threshold. The corrected model parameters are used to perform deconvolution operation to solve the real adsorption heat source intensity field.

[0010] Step S400, Mass transfer front feature reconstruction: Perform spatial distribution morphology analysis on the actual adsorption heat source intensity field, calculate the spatial second derivative of the actual adsorption heat source intensity field, identify and locate the activation front and deactivation trailing edge of the adsorption mass transfer band, and calculate the length of the mass transfer zone.

[0011] Step S500, State Assessment and Closed-Loop Control: Based on the position of the activation front and the length of the mass transfer zone, perform state assessment and closed-loop control.

[0012] This invention does not directly rely on concentration changes, but rather utilizes the exothermic effect that inevitably accompanies the adsorption process. By analyzing the temperature distribution characteristics within the bed, it reverse-engineers the intensity and location distribution of the adsorption heat source, thereby accurately reconstructing the dynamic characteristics of the adsorption mass transfer frontier. Its innovative principle lies in: first, establishing a global energy constraint, i.e., calculating the theoretical total adsorption exothermic power using inlet operating conditions; then, using the measured temperature field of the bed, solving for the microscopic distribution of the adsorption heat source through an inversion algorithm; finally, comparing and correcting the results of the macroscopic theoretical calculation and the microscopic inversion solution, unifying them through iteration, thereby obtaining a high-precision true adsorption heat source intensity field, and locating the mass transfer zone accordingly.

[0013] Preferably, in the data acquisition step, an axial sensor group arranged at equal intervals along the axial coordinates of the gas flow direction is used to acquire the axial temperature distribution vector of the bed, and a radial differential sensor group arranged in pairs at the center of the bed and near the wall on a preset axial section is used to acquire radial temperature difference data.

[0014] Preferably, in the theoretical calculation step, the theoretical total adsorption heat release power is derived based on the principle of macroscopic mass balance. Specifically, the product of the instantaneous inlet gas moisture content and the instantaneous inlet gas mass flow rate is used as the moisture mass flow rate, which is then multiplied by the differential adsorption heat value to calculate the total theoretical adsorption heat release power. This total theoretical adsorption heat release power is used as the global physical constraint benchmark for subsequent model inversion.

[0015] Preferably, in the model calibration step, the preset model parameter is the effective thermal conductivity of the bed in the unsteady-state thermal conduction inversion model; the specific operation of automatically adjusting the preset model parameter is as follows: when the relative residual between the spatial integration result and the theoretical total adsorption heat release power exceeds the preset threshold, the effective thermal conductivity of the bed is iteratively corrected according to the relative residual.

[0016] Preferably, in the inversion solution step, performing deconvolution operation using the modified model parameters specifically includes: constructing a convolution equation describing the relationship between the adsorption bed temperature response and the heat source intensity based on the modified model parameters, discretizing it into matrix form, and solving the matrix form using a regularization method (such as Tikhonov regularization) to calculate the true adsorption heat source intensity field that eliminates the bed heat capacity hysteresis effect.

[0017] Preferably, in the feature reconstruction step, the discrete real adsorption heat source intensity field is numerically differentiated using the central difference method to solve for the spatial second derivative; the specific basis for identifying and locating the adsorption mass transfer zone boundary is: the first zero-crossing point where the spatial second derivative changes from positive to negative is identified as the activation front position, and the first zero-crossing point where the spatial second derivative changes from negative to positive is identified as the deactivation trailing edge position.

[0018] Preferably, the feature reconstruction step further includes calculating the moving speed of the adsorption mass transfer band: searching for the peak position of the true adsorption heat source intensity field between the activation front position and the deactivation rear position, and obtaining the moving speed by calculating the displacement change rate of the peak position at adjacent sampling times.

[0019] Preferably, the state assessment and closed-loop control steps include real-time penetration risk warning: comparing the identified activation front position with the preset outlet safety boundary in real time; if the activation front position reaches or exceeds the preset outlet safety boundary, immediately issuing an adsorption tower switching or regeneration command to the external process control system. More preferably, the axial position of the preset outlet safety boundary is set between 80% and 95% of the total height of the main bed of the adsorption drying tower.

[0020] Preferably, the state assessment and closed-loop control steps further include adsorbent health assessment: recording the length of the mass transfer zone at the end of each operating cycle and comparing it with the trend data of the mass transfer zone length in historical operating cycles; if the comparison results show that the length of the mass transfer zone shows an increasing trend in multiple consecutive cycles and exceeds the preset health threshold, then it is determined that the molecular sieve adsorbent performance has deteriorated and a maintenance warning is generated.

[0021] This invention provides a method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis. It offers the following advantages:

[0022] 1. This invention, by arranging a temperature monitoring array in the bed and combining it with model inversion, can detect the specific location, shape and movement speed of the adsorption mass transfer front within the bed in real time and dynamically. Operators can clearly observe the entire process of the mass transfer front moving towards the outlet and receive an early warning before it reaches the safety boundary, thereby achieving precise switching and effectively avoiding the waste of molecular sieve capacity and the risk of water vapor leakage caused by judgment lag.

[0023] 2. This invention, by enabling real-time and precise detection of the mass transfer frontier's position, eliminates reliance on rough experience in determining regeneration timing. Regeneration can be delayed when the mass transfer front is not yet close to the outlet, maximizing the effective adsorption capacity of the molecular sieve. Conversely, regeneration can be initiated promptly when its movement speed increases due to changes in operating conditions, ensuring product gas quality. This effectively avoids energy waste caused by premature regeneration and quality risks associated with delayed regeneration, thus improving the system's operational economy.

[0024] 3. This invention, by continuously monitoring the key parameter of the length (shape) of the mass transfer zone and comparing it with historical data, can quantitatively assess the true health status of the molecular sieve due to dynamic changes in adsorption load or localized deterioration. This changes the past extensive lifespan prediction based solely on operating time or cumulative throughput, allowing the replacement and maintenance of the molecular sieve to be based on its actual performance status, thereby avoiding unnecessary premature replacement and preventing production interruptions due to unexpected failures. Attached Figure Description

[0025] Figure 1 This is a system framework diagram of the present invention;

[0026] Figure 2 This is a schematic diagram of the method flow of the present invention.

[0027] The components include: 100, Adsorption Drying Tower Main Body; 200, Multidimensional Temperature Monitoring Module; 210, Axial Sensor Group; 220, Radial Differential Sensor Group; 300, Inlet Operating Condition Monitoring Module; 310, Mass Flow Meter; 320, Online Trace Water Analyzer; 400, Computing and Control Center; 410, Data Acquisition Unit; 420, Model Inversion and Correction Unit; 430, Mass Transfer Front Analysis Unit; 440, Status Assessment and Early Warning Unit; and 500, Human-Machine Interface. Detailed Implementation

[0028] The technical solutions in 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.

[0029] See attached document Figure 1 The present invention provides an adsorption mass transfer front positioning system for a molecular sieve drying tower for hydrogen production by water electrolysis. The system includes: an adsorption drying tower body 100, a multi-dimensional temperature monitoring module 200, an inlet operating condition monitoring module 300, a computing and control center 400, and a human-machine interface 500.

[0030] The main body 100 of the adsorption drying tower is filled with molecular sieve adsorbent for deep drying of raw hydrogen gas carrying trace amounts of moisture. The main body 100 of the adsorption drying tower is provided with a raw gas inlet and a dry gas outlet to form a gas flow path.

[0031] The multi-dimensional temperature monitoring module 200 is used to acquire real-time spatiotemporal temperature field data of the bed layer inside the adsorption drying tower body 100. This module specifically includes multiple high-precision temperature sensors installed at specific locations inside the adsorption bed or on its walls, forming a monitoring array. This array is divided into:

[0032] An axial sensor group 210, arranged at equal intervals along the axial coordinate (z-axis) of the gas flow direction, is used to collect the axial temperature distribution vector of the bed; a radial differential sensor group 220, arranged in pairs at the center of the bed and near the wall on one or more preset key axial sections, is used to obtain radial temperature gradient information at the same height.

[0033] The inlet operating condition monitoring module 300 is used to collect the physical parameters of the raw material hydrogen entering the adsorption drying tower body 100 in real time. The inlet operating condition monitoring module 300 mainly includes:

[0034] The mass flow meter 310, installed on the raw gas inlet pipeline, is used to measure the instantaneous inlet gas mass flow rate;

[0035] An online trace water analyzer 320, installed downstream of the mass flow meter 310 and upstream of the adsorption drying tower body 100, is used to measure the water content or dew point in the instantaneous inlet gas.

[0036] A computing and control center 400 communicates with the multi-dimensional temperature monitoring module 200 and the inlet condition monitoring module 300, and is connected to the process control system (DCS / PLC) of the adsorption drying tower. The computing and control center 400 is programmed to execute the positioning method of this invention. The center internally includes:

[0037] The data acquisition unit 410 is used to synchronously receive and preprocess raw measurement data from the multi-dimensional temperature monitoring module 200 and the inlet condition monitoring module 300;

[0038] The model inversion and correction unit 420 stores a pre-established unsteady-state heat conduction inversion model and is configured to perform parameter adaptive correction based on global enthalpy and mass conservation as well as deconvolution operation of adsorbed heat sources.

[0039] The mass transfer front analysis unit 430 is used to analyze the adsorption heat source distribution data output by the model inversion and correction unit 420 and calculate characteristic parameters such as the position, length and movement velocity of the adsorption mass transfer front (MTZ).

[0040] The status assessment and early warning unit 440 is used to assess the working status of the adsorption tower and the health of the adsorbent based on the output results of the mass transfer front analysis unit 430, generate early warning signals and control commands (such as switching regeneration commands), and send the commands to the external process control system.

[0041] The human-machine interface 500 is connected to the computing and control center 400 and is used to display early warning information and system operating status to operators in real time.

[0042] See attached document Figure 2 This invention provides a method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis. This method is implemented using the aforementioned system and includes the following steps:

[0043] Step S100: Synchronous Acquisition of Multidimensional Thermal-Hydraulic Data: At each preset sampling time, the calculation and control center 400 synchronously acquires real-time operating status data of the adsorption drying tower body 100 through its data acquisition unit 410. This includes acquiring the bed axial temperature distribution vector and radial temperature difference data from the multidimensional temperature monitoring module 200, and acquiring the instantaneous inlet gas mass flow rate and instantaneous inlet gas moisture content from the inlet operating condition monitoring module 300. The acquired data, after preliminary processing, is used as input for subsequent calculations.

[0044] Step S200: Calculation of Macroscopic Theoretical Adsorption Enthalpy Change: The model inversion and correction unit 420 of the calculation and control center 400 calculates the theoretical total adsorption heat release power based on mass balance at the current moment, based on the inlet gas mass flow rate and moisture content collected in step S100, and combined with the preset differential adsorption heat value of water molecules on the molecular sieve material. This power value serves as a global physical constraint benchmark for verification and correction in subsequent steps.

[0045] Step S300: Adaptive Correction and Inversion Solution of Model Parameters: This step is a nested iterative optimization process executed by the model inversion and correction unit 420. This unit first uses historical model parameters and measured temperature fields to preliminarily estimate the distribution of adsorption heat sources inside the bed through an inversion algorithm. Then, it performs spatial integration on this heat source distribution to obtain the total microscopic inverted heat power, and compares it with the macroscopic theoretical total adsorption heat release power calculated in step S200. If there is a discrepancy, the optimization algorithm automatically adjusts key parameters in the thermodynamic model (such as the effective thermal conductivity of the bed) until the discrepancy converges within a preset threshold. Finally, the corrected model parameters are used to perform the final deconvolution operation to calculate the true adsorption heat source intensity field after eliminating thermal hysteresis.

[0046] Step S400, Mass Transfer Front (MTZ) Feature Reconstruction: The mass transfer front analysis unit 430 of the calculation and control center 400 performs spatial distribution morphology analysis on the actual adsorption heat source intensity field output in step S300. By calculating the spatial second derivative of the heat source intensity curve, the activation front position and deactivation trailing edge position of the adsorption mass transfer zone (MTZ) are identified and located, and then key features such as the length and movement velocity of the mass transfer zone at the current moment are calculated.

[0047] Step S500, Status Assessment and Closed-Loop Control: The status assessment and early warning unit 440 of the calculation and control center 400 performs process assessment and control based on the MTZ characteristic parameters output in step S400. On the one hand, it determines in real time whether the activation front position of the MTZ has reached the preset outlet safety boundary. If it has, it immediately sends an adsorption tower switching command to the external process control system and displays an early warning on the human-machine interface 500. On the other hand, it compares the calculated mass transfer zone length with historical data to dynamically assess the health status of the molecular sieve adsorbent and generates a maintenance early warning when performance significantly deteriorates.

[0048] The above steps S100 to S500 constitute a complete real-time cycle, which is continuously executed during the molecular sieve adsorption and drying process to achieve precise and dynamic tracking of the adsorption mass transfer frontier. The technical details of each of the above steps will be explained in more detail below.

[0049] In the initial stage of the adsorption mass transfer front positioning method in the molecular sieve drying tower for hydrogen production by water electrolysis, i.e., step S100, the calculation and control center 400, through its data acquisition unit 410, sets a preset time step. The thermal and hydraulic parameters of the main body 100 of the adsorption drying tower are collected periodically to provide accurate initial and boundary conditions for subsequent model calculations.

[0050] Specifically, in this step, the acquisition and preprocessing of axial temperature field data are performed as follows:

[0051] The computing and control center 400 sends a data request command to the axial sensor group 210 installed inside the adsorption drying tower body 100. This sensor group consists of... equidistant It consists of high-precision temperature sensors (such as PT100 platinum resistance thermometers or K-type thermocouples), whose axial coordinates along the gas flow direction are as follows: The analog voltage or resistance signals collected by these sensors are converted into digital temperature at each sampling moment by a signal conditioner. The data acquisition unit 410 will receive the data. The discrete temperature values ​​are combined into a one-dimensional column vector, namely the axial temperature distribution vector. :

[0052] ;

[0053] in, Indicates at time Located in the axial position The raw temperature value measured by the sensor.

[0054] To eliminate the impact of sensor noise and on-site electromagnetic interference on measurement accuracy, the data acquisition unit 410 processes the acquired raw temperature vector. Digital filtering is performed. In this embodiment, a moving average filtering algorithm is used. For each element in the vector... its filtered value It is calculated by the following formula:

[0055] ;

[0056] in, The width of the moving average window is a preset integer. This indicates that the time traveled backwards from the current moment... The elements are accumulated; for the specific selection and implementation of the digital filtering algorithm, those skilled in the art can choose methods such as Kalman filtering and wavelet transform according to the actual working conditions. These are all well-known technologies in the field and will not be elaborated here. The processed temperature vector It will be stored and transmitted to the model inversion and correction unit 420.

[0057] Simultaneously with the acquisition of axial temperature field data, the acquisition of radial thermal effect and inlet operating condition data is also completed in step S100. The data acquisition unit 410 simultaneously acquires data from the radial differential sensor group 220. This sensor group is installed on a preset cross-section (e.g., near the expected penetration zone) where the radial thermal effect of the bed is significant, and each cross-section includes a central measuring point. and a near-wall measuring point The collected radial temperature difference data It is used to characterize the radial heat loss effect of the bed and serves as an auxiliary verification basis in the subsequent adaptive correction step of model parameters.

[0058] Simultaneously, the data acquisition unit 410 acquires data from the inlet condition monitoring module 300 installed on the inlet pipe of the adsorption drying tower body 100. Specifically:

[0059] Read the instantaneous inlet gas mass flow rate from mass flow meter 310. ;

[0060] The water content of the instantaneous inlet gas is read from the online trace water analyzer 320, and this water content can be expressed as a mass fraction. Or volume concentration.

[0061] Thus, step S100 completes the synchronous acquisition of key thermo-hydraulic parameters inside and outside the adsorption bed. These processed, synchronous datasets include the axial temperature distribution vector. Radial temperature difference and inlet operating parameters and Together, they form the basis for subsequent steps of model calculation and state assessment.

[0062] After data acquisition is completed in step S100, the calculation and control center 400 proceeds to step S200, which is the calculation of the macroscopic theoretical adsorption enthalpy change. The core of this step lies in calculating the total adsorption heat that the entire adsorption bed should theoretically release per unit time, based on the inlet hydrogen flow rate and moisture load, from the perspective of macroscopic mass and energy conservation. This theoretical value will serve as a key macroscopic constraint for evaluating the rationality of subsequent microscopic model inversion results.

[0063] The calculation model for theoretical adsorption heat power is executed by the model inversion and correction unit 420 of the calculation and control center 400. Specifically, the model inversion and correction unit 420 first calls the instantaneous inlet gas mass flow rate collected and stored in step S100. and inlet water content .

[0064] Inlet water content This characterizes the mass of water carried per unit mass of dry hydrogen gas, with dimensions of kg(H2O) / kg(H2). Therefore, it represents the mass flow rate of water entering the adsorption drying tower body 100 per unit time. It can be calculated using the formula for water input rate:

[0065] ;

[0066] Assuming that under ideal operating conditions, the water entering the adsorption tower can be completely adsorbed by the molecular sieve, then This is equal to the mass of water adsorbed by the molecular sieve per unit time. When each unit mass of water is adsorbed by the molecular sieve, a specific amount of heat is released, known as the differential heat of adsorption. Differential heat of adsorption These are characteristic parameters of the molecular sieve material itself, and their values ​​are related to the amount of water adsorbed and the temperature. They are usually determined experimentally or provided by the molecular sieve supplier. In this embodiment, The nominal differential adsorption heat value at 25℃ is used, and the unit is J / kg. .

[0067] Based on the above, the model inversion and correction unit 420 uses the theoretical adsorption heat power calculation formula to obtain the theoretical total adsorption heat power based on macroscopic mass balance at the current moment. The calculation formula is as follows:

[0068] ;

[0069] This represents the total theoretical thermal power generated under the current inlet conditions, assuming all moisture is adsorbed and all adsorption heat is released within the bed. This value will be passed to the subsequent adaptive correction step of the model parameters to verify and constrain the intensity of the adsorption heat source obtained through temperature field inversion at the microscale.

[0070] After completing step S200, the theoretical total adsorption heat release power is obtained. Subsequently, the model inversion and correction unit 420 of the computing and control center 400 executes the core task of step S300. The first step in adaptive correction and inversion solution of model parameters is to construct a mathematical model that can accurately describe the coupling effect of heat conduction and internal heat sources within the adsorption bed, i.e., an unsteady-state heat conduction inversion model containing internal heat sources. This model will serve as a mathematical tool for inferring the distribution of adsorption heat sources throughout the bed from limited temperature measurement data.

[0071] In the model construction process, the molecular sieve bed within the main body 100 of the adsorption drying tower was first reasonably simplified. Considering that the gas flow direction is axial and the radial temperature gradient is relatively small (except for the near-wall region), and in order to reduce model complexity and meet real-time requirements, the bed was treated as a one-dimensional unsteady heat conduction problem, with the axial coordinates denoted as... The tower is [height] During adsorption, water molecules are adsorbed onto the surface of the molecular sieve, releasing heat. This heat acts as an internal heat source, driving temperature changes in the bed. Therefore, describing the bed temperature field... The governing equations, which vary with time and space, are one-dimensional unsteady-state heat conduction equations containing internal heat sources, and can be expressed as the governing equations for heat conduction in the adsorption bed:

[0072] ;

[0073] in: This refers to the bed bulk density, in units of... ; The effective specific heat capacity of the bed is expressed in units of 1000 kJ / m². ; The effective thermal conductivity of the bed is expressed in units of 1000 kJ / m². It comprehensively considers multiple heat transfer mechanisms, including solid particle heat conduction, interparticle contact heat conduction, gas heat conduction within pores, and thermal radiation. The adsorption heat source intensity per unit volume of bed is the heat released during the adsorption process per unit time and per unit volume of bed, expressed in units of... ; This represents the temperature field distribution function of the bed. It is a time variable; The sign for the partial derivative with respect to time; The sign represents the partial derivative with respect to space.

[0074] For the above partial differential equations, initial and boundary conditions also need to be specified. The initial condition is the bed temperature distribution before adsorption begins, which can be expressed as:

[0075] ;

[0076] in The initial axial temperature distribution of the bed can generally be considered as a uniform temperature, i.e., the ambient temperature or the initial feed temperature.

[0077] Boundary conditions are typically adiabatic or convective heat transfer boundary conditions. For both ends of the tower, i.e., from 0 on the z-axis to the top H, in the simplified case where axial heat loss is neglected, adiabatic boundary conditions can be used:

[0078] ;

[0079] Radial heat loss is difficult to represent directly in a one-dimensional model; its impact is usually conveyed through the effective thermal conductivity. Make appropriate corrections or take them into consideration in subsequent parameter adaptive correction steps.

[0080] Next, the above governing equations need to be discretized to facilitate computer solution. The model inversion and correction unit 420 uses the finite difference method for spatial and temporal discretization. The bed is arranged along the axial direction... The direction is evenly divided into There are 1 grid, and the grid node coordinates are... ,in … Grid step size The time direction is based on the sampling time interval. Keep it in mind and keep it in mind at all times. ,in …

[0081] Record nodes At any moment The temperature is The heat source intensity is Using a central difference scheme for the spatial second derivative term and a forward difference scheme for the time first derivative term, after discretization, the following system of algebraic equations can be obtained using the discretized heat conduction equation:

[0082] ;

[0083] in … Boundary nodes and The temperature is handled according to the aforementioned adiabatic boundary conditions; Represents a node At any moment Temperature; Represents a node At any moment Temperature; Represents a node At any moment Temperature; This represents the grid step size.

[0084] In the inversion problem, it is known that there are several discrete measurement points (i.e., the arrangement positions of the axial sensor group 210). ... Temperature measurement at ) The goal is to solve for the unknown heat source intensity. The above discretized equations can be expressed in matrix form, i.e.:

[0085] ;

[0086] in and They are time points The temperature vector and the heat source intensity vector, This is a temperature coefficient matrix. This represents the heat source coefficient matrix. By establishing this linear relationship between the measured temperature value and the unknown heat source intensity, the continuous partial differential equation problem can be transformed into a discrete linear algebraic equation system problem. Methods for solving linear equation systems include the least squares method and regularization methods.

[0087] Through the above steps, the model inversion and correction unit 420 successfully constructed an adsorption heat source inversion model framework based on a one-dimensional unsteady-state heat conduction equation. This model uses the effective thermal conductivity of the bed... Using physical parameters as input and temperature field as the link, the unknown adsorption heat source intensity is... With the measurable axial temperature distribution vector Connect them.

[0088] After constructing the framework of the unsteady-state heat conduction inversion model containing an internal heat source, the key to whether the model can function accurately lies in its physical parameters, especially the effective thermal conductivity of the bed. Is the value of this parameter accurate? As mentioned before, this parameter is difficult to preset precisely. To solve this problem, this invention proposes a parameter adaptive correction mechanism based on global enthalpy and mass conservation. This mechanism is executed by the model inversion and correction unit 420 of the calculation and control center 400. Its basic idea is: at any time, the total microscopic inversion heat power within the entire bed layer is obtained through temperature field inversion calculation. Physically, it should be equal to the total theoretical adsorption and heat release power calculated based on the inlet operating conditions. If the two are not equal, it indicates that the physical parameters used in the inversion model (mainly...) There is a deviation. Therefore, these parameters can be adjusted iteratively to make... Approaching This enables adaptive online correction of model parameters. The specific steps for implementing this adaptive correction mechanism are as follows:

[0089] At each sampling time The model inversion and correction unit 420 first uses an initial or previous corrected effective thermal conductivity of the bed. (where superscript) Represents the number of iterations, initially. This unit, based on the aforementioned unsteady heat conduction model containing an internal heat source, performs an inversion solution on the discretized form of the model. Specifically, it combines the axial temperature distribution vector acquired in step S100. An uncorrected adsorption heat source intensity field was initially derived using a regularization method. .

[0090] Since this inversion is an ill-posed problem, direct solution may lead to results that are extremely sensitive to measurement errors. Therefore, a regularization method is used for the solution. First, an uncorrected adsorption heat source intensity field is initially inverted. Subsequently, by spatially integrating the heat source intensity field along the entire bed axis, the total microscopic inversion heat power for the current iteration step is obtained using the formula for calculating the total microscopic inversion heat power. The calculation formula is as follows:

[0091] ;

[0092] in, The cross-sectional area of ​​the adsorption drying tower body 100; This represents the grid step size.

[0093] The model inversion and correction unit 420 calculates the total microscopic inversion heat power. Compared with the macroscopic theoretical total adsorption heat release power obtained in step S200 based on the theoretical adsorption heat power calculation formula, Compare the two and calculate the relative residuals.

[0094] ;

[0095] Set a preset convergence threshold. (For example, 0.01). If Then it is assumed that the currently used model parameters The parameter accurately reflects the true thermophysical properties of the bed, the iteration process terminates, and the parameter is... As of the present moment The final correction value.

[0096] like This indicates that there is a deviation in the model parameters, which needs to be adjusted. The model inversion and correction unit 420 initiates an optimization algorithm to update the effective thermal conductivity of the bed. A simple and effective iterative update strategy is to use the parameter iteration correction formula for proportional feedback correction, as follows:

[0097] ;

[0098] in, It is a relaxation factor less than 1, used to prevent oscillations during the iteration process and ensure the stability of convergence. The physical meaning of this correction formula is:

[0099] If the total heat power is determined by microscopic inversion Less than the theoretical value This indicates that the heat diffusion effect in the current model may be overestimated (i.e., If the value is too large, the calculated temperature field gradient will be insufficient to support a sufficiently large heat source intensity. In this case, the value should be reduced. Conversely, if Greater than This indicates that the heat diffusion effect in the model may be underestimated (i.e., (Too small), needs to be enlarged .

[0100] After the parameters are updated, the number of iterations Add 1, use the new Repeat the heat source inversion and microscopic total heat power calculation until the residuals meet the convergence requirements.

[0101] The effective thermal conductivity of the bed obtained through this closed-loop iterative correction process is... It is no longer a fixed, experience-dependent constant, but an adaptive parameter that can be dynamically adjusted according to real-time operating conditions. It implicitly compensates for complex factors not directly considered in the one-dimensional model, such as radial heat loss, the impact of gas velocity changes on heat transfer, and uneven bed filling, ensuring the robustness and accuracy of the inversion model under different operating conditions. The online parameter correction mechanism under global enthalpy and mass conservation constraints forms the technical basis for this invention's ability to accurately invert the real adsorption heat source field.

[0102] The accurate effective thermal conductivity of the bed was obtained through an online parameter correction mechanism under the constraints of global enthalpy and mass conservation. Then, the model inversion and correction unit 420 enters the stage of accurate solution of the adsorption heat source field. This is because the adsorption bed itself has a large thermal inertia (i.e., The temperature change measured by the sensor (due to its relatively large value) is the result of the combined effects of adsorption heat generation and bed heat capacity, exhibiting a certain time lag and amplitude attenuation. Therefore, the heat source intensity directly derived from temperature data may not accurately reflect the true adsorption rate change in real time. To address this issue, this invention introduces a deconvolution algorithm to accurately extract and recover the true adsorption heat source intensity field from the temperature response, which includes the heat capacity hysteresis effect. .

[0103] The deconvolution solution of the adsorption heat source field is also performed by the model inversion and correction unit 420 of the calculation and control center 400. Its core idea is to treat the temperature response of the adsorption bed to the adsorption heat source as a linear time-invariant system, where the adsorption heat source intensity... It is the system input, temperature field This is the system output. Given the system output and the system's unit impulse response function (i.e., Green's function), the system input can be obtained through deconvolution. The specific steps are as follows:

[0104] Step 1: Establish the convolution relationship between temperature response and heat source intensity.

[0105] For those already determined The subsequent one-dimensional unsteady heat conduction equation, under given initial and boundary conditions, has a Green's function. Describes the location ,time When a unit pulse heat source is input, at position :time The resulting temperature response. According to the superposition principle of linear systems, at any position within the bed... At any moment temperature This can be expressed as heat source intensity. With Green's function Convolution integral:

[0106] ;

[0107] in, The temperature response caused by the initial temperature distribution and boundary conditions (which can be ignored or treated separately in this problem); This represents the spatiotemporal convolution integral. In practical calculations, this convolution integral equation needs to be discretized. By dividing time and space into a finite grid, the integration operation is transformed into a summation operation. At this point, the Green's function describing the system response... It evolves into a Green's function matrix The intensity of the continuous heat source to be determined Transformed into a discrete heat source intensity vector , and continuous temperature response This corresponds to the temperature observation vector measured by the sensor. Therefore, the above convolution relationship can be expressed in the following discrete matrix form: .

[0108] Step 2: Solve the deconvolution based on Tikhonov regularization.

[0109] Directly solve the inverse problem of the above convolution equation (i.e., deconvolution) to obtain This is unsuitable as it is susceptible to measurement noise. Therefore, the model inversion and correction unit 420 employs the Tikhonov regularization method to stabilize the solution process. Tikhonov regularization balances the fitting error and smoothness of the solution by introducing a regularization term; its objective function is:

[0110] ;

[0111] in, It is a temperature measurement vector acquired and preprocessed by the axial sensor group 210. It is a regularization parameter used to control the weight between the fitting error and the regularization term. Typically, an identity matrix or a first / second order difference matrix is ​​used to constrain the smoothness of the solution. This is achieved by minimizing the above objective function. Stable heat source intensity inversion results can be obtained. .

[0112] For regularization parameters The selection of the optimal cross-validation method can be achieved using the L-curve method or generalized cross-validation (GCV). These methods can automatically determine a relatively optimal cross-validation method. The goal is to suppress noise amplification while preserving as much detailed information about the heat source as possible. For the specific solution of the Tikhonov regularization problem, those skilled in the art can use the conjugate gradient method or the direct matrix factorization method, which are well-known techniques in the field and will not be elaborated upon here.

[0113] Step 3: Obtain the actual adsorption heat source intensity field.

[0114] Through the above Tikhonov regularized deconvolution process, the model inversion and correction unit 420 obtains the solution... This is the adsorption heat source intensity vector after eliminating the bed thermal capacity hysteresis effect. Converting it back to a continuous space description yields the true adsorption heat source intensity field distributed along the bed axis at the current moment. The heat source intensity field directly reflects the rate of adsorption heat generation per unit time and unit volume of the molecular sieve at different axial positions, and thus is related to the adsorption rate at that location.

[0115] By applying the deconvolution algorithm, this method effectively overcomes the "filtering" effect of bed thermal inertia on the temperature signal, making the retrieved temperature signal more accurate and reliable. It can more realistically and timely reflect the dynamic characteristics of the adsorption process, laying a solid foundation for the accurate reconstruction of the mass transfer front (MTZ) in the future.

[0116] In step S300, the true adsorption heat source intensity field is obtained by deconvolution. Subsequently, the mass transfer front analysis unit 430 of the computing and control center 400 begins to execute step S400, namely, the reconstruction of the mass transfer front (MTZ) characteristics. The precise definition of the physical boundary of the MTZ is the primary task of this step, and its goal is to determine the starting position (activation front) and ending position (deactivation trailing edge) of the MTZ along the bed axis at the current moment.

[0117] During adsorption, the MTZ (Mean Transmission Zone) is the region with the largest moisture concentration gradient and the most vigorous adsorption reaction within the bed. Intuitively, this region corresponds to the intensity of the adsorption heat source. The peak region. However, directly defining the boundary of MTZ by the maximum value of the heat source intensity or a certain threshold is susceptible to local fluctuations or noise interference and is not accurate enough.

[0118] By analyzing the intensity field of the adsorption heat source By calculating the second spatial derivative along the axial direction and utilizing its zero-crossing characteristic, the physical boundary of the MTZ can be identified more objectively and accurately. The precise definition of the MTZ physical boundary is performed by the mass transfer front analysis unit 430 of the computation and control center 400, and the specific steps are as follows:

[0119] The mass transfer front analysis unit 430 first obtains the current time value transmitted from the model inversion and correction unit 420. Adsorption heat source intensity field .because It is represented by discrete points (corresponding to the measurement positions of the axial sensor group 210). The mass transfer front analysis unit 430 uses the central difference method to numerically calculate its spatial derivative.

[0120] For discrete heat source intensity vectors Its position First spatial derivative at Approximately:

[0121] ;

[0122] in, This refers to the arrangement spacing of the axial sensor group 210. Representing discrete points exist The location for measuring the time; Representing discrete points exist The measurement location at time; at boundary points (e.g.) and The first derivative of can be obtained using either forward or backward difference schemes.

[0123] Furthermore, in position At this point, the solution is obtained using the formula for calculating the second derivative of the heat source intensity. :

[0124] ;

[0125] The second derivative mentioned above reflects the curvature of the heat source intensity field along the axial direction. At the front and back edges of the MTZ region, the rate of change of the heat source intensity (first derivative) will undergo a transition from increasing to decreasing or from decreasing to increasing. This transition corresponds to the zero-crossing point of the second derivative.

[0126] For the calculated second derivative sequence ( The mass transfer front analysis unit 430 detects the sign change of the heat source. In the MTZ region, the heat source intensity... It typically exhibits a distinct unimodal distribution. Ideally, the concavity of this distribution curve is upward (positive second derivative) before the activation front and after the deactivation rear edge of the MTZ, and downward (negative second derivative) inside the MTZ. Therefore, the first zero-crossing point where the value changes from positive to negative corresponds to the activation front of the MTZ. The first zero-crossing point where the value changes from negative to positive corresponds to the trailing edge of MTZ inactivation. .

[0127] In practice, due to noise or measurement errors, the second derivative may exhibit slight oscillations near zero. The mass transfer front analysis unit 430 performs smoothing preprocessing on the second derivative sequence (e.g., using low-pass filtering or moving average), and then searches for stable sign transition points in the preprocessed sequence. Specifically, from the bed inlet end (usually the end with the smaller axial coordinate, such as...) or Towards the outlet end (the end with the larger axial coordinate, such as...) or Scanning the second derivative sequence:

[0128] Find the point where the second derivative first changes from positive to negative and remains negative for several consecutive points (e.g., 3 points). Denote the axial position corresponding to this point as the starting edge of the MTZ. .

[0129] Continue scanning until the first point where the second derivative changes from negative to positive and remains positive for several consecutive points (e.g., 3 points). Record the axial position corresponding to this point as the deactivation trailing edge of the MTZ. .

[0130] If the second derivatives of the entire bed are all negative, it indicates that the MTZ has not yet fully formed or has covered the entire bed; if the second derivatives are all positive, it indicates that the adsorption process may not have started significantly or has approached complete saturation. For these special cases, the mass transfer front analysis unit 430 will process them according to preset rules (such as setting the activation front as the bed inlet and the deactivation back front as the bed outlet, or vice versa).

[0131] Determined by the above method and Together, these constitute the physical boundary of the mass transfer frontier (MTZ) at the current moment. This method of defining the boundary based on the zero-crossing point of the second derivative can effectively eliminate the interference of local extreme points and better reflect the overall changing trend of the MTZ region, thereby achieving accurate capture of the MTZ physical boundary.

[0132] By precisely defining the MTZ start-up frontier and inactivation after edge Subsequently, the mass transfer front analysis unit 430 further calculates the key characteristic parameters of the MTZ, which are the core basis for evaluating the working state of the adsorption column and predicting the breakthrough time. These mainly include the mass transfer zone length. and MTZ's movement speed .

[0133] Mass transfer region length Defined as the axial distance between the deactivation trailing edge and the activation leading edge of the MTZ. At the current moment... The mass transfer front analysis unit 430 directly uses the formula for calculating the length of the mass transfer region to perform calculations based on the two identified boundary locations:

[0134] ;

[0135] Mass transfer region length The unit is usually meter (m) or millimeter (mm). Its value reflects the size of the area within the bed where effective adsorption is taking place, and is affected by various factors such as feed flow rate, initial adsorbent state, and adsorbate concentration. Smaller... This usually means that the adsorbent utilization rate is high.

[0136] To calculate the moving velocity of the MTZ, it is first necessary to determine the representative positions of the MTZ at different times. This is due to the intensity of the adsorbed heat source inside the MTZ. Typically exhibiting a single-peak distribution, the mass transfer front analysis unit 430 defines the axial position of this peak point as the characteristic peak position of the MTZ. .

[0137] Specifically, within the defined MTZ physical boundary and Within the region, the mass transfer front analysis unit 430 searches for the intensity of the adsorption heat source. The axial coordinates corresponding to the maximum value point:

[0138] ;

[0139] If multiple local maxima exist within the MTZ region, the position of the largest peak is usually selected as the maximum value. .

[0140] The MTZ speed refers to the position of its characteristic peak. The rate of change over time. The mass transfer front analysis unit 430 calculates this rate by tracking the change in the peak position between two adjacent sampling times.

[0141] Let the current time be The previous sampling time was (in (Sampling time interval). At that time, the peak position of MTZ has been determined as Then the current movement speed of MTZ. The movement speed is calculated using the MTZ (Movement Speed) formula, which is:

[0142] ;

[0143] The unit of MTZ moving speed is usually meters per second (m / s) or meters per hour (m / h). A positive speed value indicates that the MTZ moves towards the bed outlet along the gas flow direction.

[0144] To improve the accuracy of velocity calculation and reduce the impact of noise, the mass transfer front analysis unit 430 can also use the sliding window averaging method to average the peak position changes over the most recent sampling periods and then divide by the total time interval to obtain a smoother MTZ moving velocity estimate.

[0145] The mass transfer region length calculated using the above steps and movement speed These parameters together constitute the key parameter set describing the dynamic characteristics of the MTZ. These parameters will be transmitted to the status assessment and early warning unit 440 of the computing and control center 400 for subsequent adsorption column status assessment and breakthrough time prediction.

[0146] After the key characteristic parameters of the MTZ are calculated in step S400, the status assessment and early warning unit 440 of the calculation and control center 400 begins to execute step S500, which involves a comprehensive assessment of the working status of the adsorption drying tower body 100 and the implementation of closed-loop control based on the assessment results. This step not only enables immediate early warning and switching control of breakthrough risk, but also provides the ability to diagnose the long-term health of the molecular sieve.

[0147] The status assessment and early warning unit 440 first performs real-time penetration warning and switching control based on the MTZ position. Its core objective is to prevent insufficiently dried gas (i.e., adsorbate penetration) from flowing out of the adsorption bed. Traditional switching strategies typically rely on fixed time intervals or outlet concentration detection; the former cannot adapt to changes in operating conditions, while the latter is lagging. This invention utilizes real-time tracked MTZ position information to achieve more precise and timely switching control.

[0148] The specific steps to implement this control are as follows:

[0149] A virtual safety boundary is set near the bed outlet of the adsorption drying tower body 100, and its axial position is denoted as . The location of this safety boundary is typically set at the total height of the bed. Between 80% and 95%, for example = This location provides the operator with a buffer zone, ensuring that the system has enough time to perform the switching operation after the switching command is issued, while the MTZ has not yet reached the bed outlet.

[0150] At each sampling time The state assessment and early warning unit 440 obtains the position of the foremost edge of the MTZ calculated by the mass transfer front analysis unit 430, i.e., the activation front. Then, this position is aligned with the preset safety boundary. Compare them.

[0151] like If the condition is not met, the adsorption column is considered to be in a safe operating range, and the status assessment and early warning unit 440 will not take any action and will continue to monitor.

[0152] like This indicates that the leading edge of the MTZ has touched or crossed the safety boundary, and a penetration risk is imminent. The status assessment and early warning unit 440 immediately triggers an instant penetration warning signal. This signal can be displayed on the human-machine interface 500 of the host computer to alert the operator.

[0153] Upon triggering the warning signal, the status assessment and warning unit 440 immediately generates a switching control command. This command is sent to the field PLC (Programmable Logic Controller) or DCS (Distributed Control System) via the communication interface. Upon receiving the command, the PLC or DCS immediately executes the preset valve switching program, switching the gas flow path from the nearly saturated adsorption-drying tower to a standby adsorption-drying tower that has already completed regeneration, thus initiating a new adsorption cycle. This control mechanism based on the MTZ physical location ensures that the switching operation always occurs at the optimal time, avoiding both adsorbent waste due to premature switching and product quality problems caused by delayed switching.

[0154] In addition to real-time control, this system also has the capability to diagnose the long-term performance of molecular sieves. During use, repeated adsorption and regeneration cycles, as well as contamination from impurities in the feed gas, can cause changes in the microporous structure of molecular sieves or the covering of active sites, resulting in a decline in kinetic performance. This decline is directly reflected in the morphological changes of the mass transfer zone (MTZ), specifically the change in the MTZ length. It is a key health indicator.

[0155] The status assessment and early warning unit 440 monitors By analyzing the changing trends over multiple operating cycles, the health status of the molecular sieve can be diagnosed. The specific process is as follows:

[0156] At the end of each complete adsorption cycle (i.e., when a switch occurs), the state assessment and early warning unit 440 calculates and records the length of the mass transfer region during that cycle. A representative value, for example, within this period The average or stable value. This value is stored in the database of the Computing and Control Center 400 and associated with the corresponding run cycle number (or run time).

[0157] As the operating cycle increases, the status assessment and early warning unit 440 will analyze a series of records in the database. Value. A healthy, stable molecular sieve, under similar operating conditions, has a value of [value missing] per cycle. The value should remain relatively stable. If observed A sustained and significant increase in the mass transfer performance of the molecular sieve over multiple consecutive operating cycles indicates a deterioration in its mass transfer capacity. An elongated mass transfer zone signifies increased mass transfer resistance between the gas and the adsorbent, leading to a decreased adsorption rate—a direct manifestation of the decline in the molecular sieve's kinetic performance.

[0158] The status assessment and early warning unit 440 sets a health threshold of one mass transfer zone length. This threshold can be determined based on the initial performance or design specifications of the molecular sieve. When continuously monitored... When the average value exceeds this threshold (e.g., exceeding 150% of the initial value), the status assessment and early warning unit 440 generates a molecular sieve health warning. This warning alerts operators that the performance of the molecular sieve in the current adsorption bed has deteriorated and needs to be replaced or deep regeneration should be arranged. This diagnostic method based on MTZ morphological changes provides a scientific basis for preventive maintenance of the equipment, avoiding sudden production interruptions caused by molecular sieve failure.

Claims

1. A method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis, characterized in that, Includes the following steps: Step S100: Synchronous acquisition of multidimensional thermal-hydraulic data: Synchronously acquire the axial temperature distribution vector, radial temperature difference data, instantaneous inlet gas mass flow rate, and instantaneous inlet gas moisture content of the adsorption drying tower body. Step S200: Calculation of macroscopic theoretical adsorption enthalpy change: Based on the instantaneous inlet gas mass flow rate, the instantaneous inlet gas moisture content, and the preset differential adsorption heat value of water molecules on the molecular sieve material, the theoretical total adsorption heat release power is calculated using the theoretical adsorption heat power calculation formula. Step S300, Adaptive correction and inversion solution of model parameters: Based on the axial temperature distribution vector of the bed and the preset model parameters, the distribution of adsorption heat source is initially estimated by the inversion algorithm. The spatial integral result of the adsorption heat source distribution is compared with the theoretical total adsorption heat release power. If there is a deviation, the preset model parameters are automatically adjusted until the deviation converges within the preset threshold. The corrected model parameters are used to perform deconvolution operation to solve the real adsorption heat source intensity field. Step S400, Reconstruction of Mass Transfer Front Features: The spatial distribution pattern of the real adsorption heat source intensity field is analyzed, and the spatial second derivative of the real adsorption heat source intensity field is solved using the formula for calculating the second derivative of the heat source intensity. The activation front position and the deactivation trailing edge position of the adsorption mass transfer band are identified and located, and the length of the mass transfer zone is calculated using the formula for calculating the length of the mass transfer zone. Step S500, State Assessment and Closed-Loop Control: Based on the position of the activation front and the length of the mass transfer zone, perform state assessment and closed-loop control.

2. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis according to claim 1, characterized in that, In step S100, the axial temperature distribution vector of the bed is collected by an axial sensor group arranged at equal intervals along the axial coordinate of the gas flow direction, and the radial temperature difference data is collected by a radial differential sensor group arranged in pairs at the center of the bed and near the wall on a preset axial section.

3. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis according to claim 1, characterized in that, In step S200, the theoretical total adsorption heat release power is calculated using the theoretical adsorption heat power calculation formula. This is based on the principle of macroscopic mass balance. The product of the instantaneous inlet gas moisture content and the instantaneous inlet gas mass flow rate is used as the moisture mass flow rate, which is then multiplied by the differential adsorption heat value. The theoretical total adsorption heat release power is used as the global physical constraint benchmark for subsequent model inversion.

4. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis according to claim 1, characterized in that, In step S300, the preset model parameter is the effective thermal conductivity of the bed in the unsteady-state thermal conductivity inversion model. The operation of automatically adjusting the preset model parameter is as follows: when the relative residual between the spatial integration result and the theoretical total adsorption heat release power exceeds the preset threshold, the effective thermal conductivity of the bed is iteratively corrected according to the relative residual.

5. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis according to claim 1, characterized in that, In step S300, the deconvolution operation using the modified model parameters specifically includes: constructing a convolution equation describing the relationship between the adsorption bed temperature response and the heat source intensity based on the modified model parameters, discretizing it into a matrix form, and solving the matrix form using the Tikhonov regularization method, thereby calculating the true adsorption heat source intensity field that eliminates the bed heat capacity hysteresis effect.

6. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis according to claim 1, characterized in that, In step S400, the calculation of the spatial second derivative of the actual adsorption heat source intensity field using the formula of the second derivative of the heat source intensity is to perform numerical differentiation on the discrete actual adsorption heat source intensity field using the central difference method; the specific basis for identifying and locating the activation front and deactivation trailing edge of the adsorption mass transfer band is as follows: the first zero-crossing point where the spatial second derivative changes from positive to negative is identified as the activation front, and the first zero-crossing point where the spatial second derivative changes from negative to positive is identified as the deactivation trailing edge.

7. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis according to claim 6, characterized in that, Step S400 further includes the step of calculating the moving speed of the adsorption mass transfer band: searching for the peak position of the true adsorption heat source intensity field between the activation front position and the deactivation rear position, and calculating the displacement change rate of the peak position at adjacent sampling times using the moving speed calculation formula to obtain the moving speed.

8. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis according to claim 1, characterized in that, In step S500, the state assessment and closed-loop control include: Real-time switching control is performed based on the activation front position, and adsorbent health is assessed based on the mass transfer zone length. The system determines in real time whether the activation front position has reached the preset outlet safety boundary. If the activation front position has reached the preset outlet safety boundary, it immediately sends an adsorption tower switching or regeneration command to the external process control system.

9. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production via water electrolysis according to claim 8, characterized in that, The axial position of the preset exit safety boundary is set between 80% and 95% of the total height of the bed.

10. The method for locating the adsorption mass transfer front in a molecular sieve drying tower for hydrogen production by water electrolysis according to claim 8, characterized in that, The process of assessing the health of the adsorbent based on the length of the mass transfer region specifically involves: The calculated mass transfer zone length is compared with historical data to dynamically assess the health status of the molecular sieve adsorbent; and when the assessment results show that the health status is declining, a maintenance warning is generated.

Images