A hydrogen production system and a hydrogen production method for improving hydrogen purity
By acquiring and adjusting hydrogen purification parameters and constructing a hydrogen parameter prediction model, the problem of fixed process parameters being unable to cope with fluctuations in feed gas quality and adsorbent aging was solved, thus achieving precise control and efficient purification of hydrogen purity.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-20
- Publication Date
- 2026-07-14
AI Technical Summary
The existing technology, with its fixed process parameters, is unable to cope with fluctuations in the quality of the raw gas or the aging of the adsorbent, resulting in the inability to effectively improve the purity of hydrogen.
By acquiring hydrogen purification parameters, including raw gas quality parameters, adsorbent performance parameters, and purification process parameters, a PCB board processing correction method and system based on visual calibration is constructed. The hydrogen purification parameters are adjusted in real time using a hydrogen parameter prediction model to achieve precise control of hydrogen purity.
It achieves stability and dynamic control of hydrogen purity, solving the problems of lag in purity prediction and weak resistance to fluctuations in feed gas in the traditional PSA hydrogen production process, and realizing precise control and efficient purification of hydrogen purity.
Smart Images
Figure CN122380302A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of hydrogen purification technology, specifically to a hydrogen production system and method for improving hydrogen purity. Background Technology
[0002] PSA adsorption technology is a highly efficient and energy-saving gas separation technology widely used in hydrogen purification. Its core principle is to utilize the differences in adsorption capacity of adsorbents for different gas components under different pressures, achieving the separation and purification of the target gas by periodically changing the pressure.
[0003] Current technologies for controlling hydrogen purity mainly rely on fixed process parameters and human experience, which makes it difficult to cope with fluctuations in the quality of raw materials or aging of adsorbents. Summary of the Invention
[0004] The technical problem to be solved by the present invention is to overcome the problem that the hydrogen purification process under various conditions is difficult to control with fixed process parameters in the prior art, which makes it impossible to effectively improve the purity of hydrogen. The present invention provides a PCB board processing correction method and system based on visual calibration.
[0005] The technical solution adopted by this invention to solve its technical problem is: a hydrogen production method for improving hydrogen purity, applied to the adsorption stage of a PSA adsorption process, comprising:
[0006] Obtain hydrogen purification parameters, wherein the hydrogen purification parameters include raw gas quality parameters, adsorbent performance parameters, and purification process parameters;
[0007] The dryness of the raw gas is obtained based on the raw gas quality parameters, and the dryness of the raw gas is extracted as the first feature variable.
[0008] Based on the adsorbent performance parameters, the physical parameters of the adsorbent are obtained, and the peak parameters of the adsorbent are obtained based on the physical parameters of the adsorbent. The peak parameters of the adsorbent are then extracted as the second feature variable.
[0009] Based on the purification process parameters, real-time purification status parameters are obtained, and these real-time purification status parameters are extracted as a third feature variable.
[0010] Based on the first feature variable, the second feature variable, and the third feature variable, variable features are obtained, and a hydrogen parameter prediction model is synthesized using the variable features.
[0011] Based on the hydrogen parameter prediction model, a predicted hydrogen purity value is generated;
[0012] The hydrogen purity deviation value is obtained by comparing the predicted hydrogen purity value with the preset hydrogen purity value, and it is then determined whether the hydrogen purity deviation value is within the preset value range.
[0013] If it is in the specified state, the hydrogen purity is determined to meet the requirements;
[0014] If not, the hydrogen purification parameters will be readjusted based on the hydrogen purity deviation value.
[0015] Further, the step of obtaining the dryness of the raw gas based on the raw gas quality parameters and extracting the dryness of the raw gas as the first feature variable includes:
[0016] The raw gas humidity data is obtained based on the raw gas quality parameters, wherein the raw gas humidity data includes water content concentration and water content temperature;
[0017] The dryness of the raw gas is obtained based on the water content concentration and the water content temperature.
[0018] A first correlation coefficient is generated based on the water content concentration, water content temperature, and raw material gas dryness, and a first characteristic variable is generated based on the first correlation coefficient.
[0019] Further, the step of obtaining the physical parameters of the adsorbent based on the performance parameters of the adsorbent, obtaining the peak parameters of the adsorbent based on the physical parameters of the adsorbent, and extracting the peak parameters of the adsorbent as the second feature variable is as follows:
[0020] The physical parameters of the adsorbent are obtained based on the adsorbent performance parameters, wherein the physical parameters of the adsorbent include the adsorbent planar area and the adsorbent thickness;
[0021] The adsorption range is obtained based on the planar area of the adsorbent.
[0022] The adsorption permeability is obtained based on the thickness of the adsorbent.
[0023] The adsorbent saturation is obtained based on the adsorption range and the adsorption permeability.
[0024] The peak parameters of the adsorbent are obtained based on the adsorbent saturation.
[0025] Establish a table of peak value changes over several time periods based on the peak parameters of the adsorbent;
[0026] A second correlation coefficient is generated based on the volatility of the peak change table for the aforementioned several time periods, and a second feature variable is generated based on the second correlation coefficient.
[0027] Further, the step of obtaining real-time purification status parameters based on the purification process parameters and extracting the real-time purification status parameters as a third feature variable includes:
[0028] The purification process parameters include raw material flow rate and process temperature;
[0029] The inlet and outlet flow rates of the raw material gas are obtained based on the raw material flow rate, and the loss flow rate is obtained based on the inlet and outlet flow rates.
[0030] The intake time is obtained based on the intake flow rate.
[0031] The air outlet time is obtained based on the air outlet flow rate.
[0032] The adsorbent capacity parameter is obtained based on the loss flow rate, the inlet time, and the outlet time.
[0033] Obtain the intake pressure at the intake time;
[0034] Obtain the exhaust pressure during the exhaust time;
[0035] The adsorbent pressure difference parameter is obtained based on the inlet pressure and the outlet pressure.
[0036] The minimum and maximum temperatures of the adsorbent are obtained based on the process temperature.
[0037] The temperature range of the adsorbent is obtained based on the lowest and highest temperatures of the adsorbent.
[0038] A third correlation coefficient is generated based on the adsorbent capacity parameter, the adsorbent differential pressure parameter, and the adsorbent temperature range value, and a third characteristic variable is generated based on the third correlation coefficient.
[0039] Further, the steps for obtaining variable features based on the first, second, and third feature variables, and synthesizing a hydrogen parameter prediction model using these variable features, are as follows:
[0040] The set of coupling variables is obtained based on the first feature variable, the second feature variable, and the third feature variable;
[0041] The model training set is obtained according to the generation process of the coupling variable set, and a model training sample library is constructed based on the model training set;
[0042] Obtain multiple sample data from the model training sample library, and derive variable features based on the mean of the multiple sample data;
[0043] A hydrogen parameter prediction model is synthesized based on the generation process of the aforementioned variable characteristics.
[0044] Furthermore, a hydrogen production system for improving hydrogen purity includes...
[0045] The acquisition module is used to acquire various hydrogen purification parameters during the hydrogen purification process, wherein the hydrogen purification parameters include raw gas quality parameters, adsorbent performance parameters, and purification process parameters.
[0046] The raw gas dryness module is used to obtain the raw gas dryness from the raw gas quality parameters and extract the raw gas dryness as a first feature variable.
[0047] The adsorbent peak parameter module is used to obtain the physical parameters of the adsorbent from the adsorbent performance parameters, and to obtain the adsorbent peak parameters based on the physical parameters of the adsorbent, and to extract the adsorbent peak parameters as a second feature variable;
[0048] The real-time purification status parameter module is used to obtain real-time purification status parameters from the purification process parameters and extract the real-time purification status parameters as a third feature variable.
[0049] The hydrogen parameter prediction model module is used to obtain variable features from the first feature variable, the second feature variable, and the third feature variable, and to synthesize the hydrogen parameter prediction model through the variable features.
[0050] The identification module is used to generate a predicted hydrogen purity value from the hydrogen parameter prediction model.
[0051] The judgment module is used to compare the predicted hydrogen purity value with the preset hydrogen purity value to obtain the hydrogen purity deviation value, and to determine whether the hydrogen purity deviation value is within the preset value range.
[0052] Furthermore, a computer device includes a memory and a processor, the memory storing a computer program, characterized in that the processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.
[0053] Furthermore, a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 5.
[0054] The beneficial effects of this invention are that the extracted hydrogen purification parameters are divided into raw material gas quality parameters, adsorbent performance parameters, and purification process parameters. By treating various parameters as adjustable features, a large data model is intelligently created after combining these features. Through multi-parameter coupled modeling and real-time feedback control, the problem of accurately and dynamically controlling the purity of hydrogen in the PSA hydrogen production process is solved, ultimately achieving a purification process with stable purity. Attached Figure Description
[0055] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0056] Figure 1 This is a schematic diagram of the method flow of the present invention.
[0057] Figure 2 This is a schematic diagram of the system structure of the present invention.
[0058] Figure 3 This is a schematic diagram of the internal structure of the computer device described in this application. Detailed Implementation
[0059] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.
[0060] like Figure 1-3 As shown, the present invention provides a method for improving the purity of hydrogen production.
[0061] The technical solution adopted by this invention to solve its technical problem is: a hydrogen production method for improving hydrogen purity, applied to the adsorption stage of a PSA adsorption process, comprising:
[0062] Step S1: Obtain hydrogen purification parameters, wherein the hydrogen purification parameters include raw material gas quality parameters, adsorbent performance parameters, and purification process parameters.
[0063] Step S2: Obtain the dryness of the raw gas based on the raw gas quality parameters, and extract the dryness of the raw gas as the first feature variable;
[0064] Step S3: Obtain the physical parameters of the adsorbent based on the performance parameters of the adsorbent, and obtain the peak parameters of the adsorbent based on the physical parameters of the adsorbent, and extract the peak parameters of the adsorbent as the second feature variable;
[0065] Step S4: Obtain real-time purification status parameters based on the purification process parameters, and extract the real-time purification status parameters as a third feature variable;
[0066] Step S5: Obtain variable features based on the first feature variable, the second feature variable, and the third feature variable, and synthesize a hydrogen parameter prediction model using the variable features;
[0067] Step S6: Generate a predicted hydrogen purity value based on the hydrogen parameter prediction model;
[0068] Step S7: By comparing the predicted hydrogen purity value with the preset hydrogen purity value, the hydrogen purity deviation value is obtained, and it is determined whether the hydrogen purity deviation value is within the preset value range.
[0069] If it is in the specified state, the hydrogen purity is determined to meet the requirements;
[0070] If not, the hydrogen purification parameters are readjusted based on the hydrogen purity deviation value. Specifically, the parameters affecting the obtained hydrogen purity are first identified, including feed gas quality parameters, adsorbent performance parameters, and purification process parameters. A closed-loop logic is used to accurately collect multi-dimensional purification parameters, extract characteristic variables to construct a predictive model, and dynamically adjust process parameters. This solves the technical problems of traditional PSA hydrogen production, such as lag in purity prediction, large adsorbent loss, and weak resistance to feed gas fluctuations. This achieves precise control and efficient purification of hydrogen purity. Step S1 is executed first to comprehensively acquire the core parameters required for the entire hydrogen purification process. These parameters cover three main categories: feed gas quality parameters, adsorbent performance parameters, and purification process parameters. The feed gas quality parameter is the PSA... The purification step enters the purification pre-preparation stage. The dryness of the raw gas directly characterizes the degree of interference of the water vapor content in the raw gas on the adsorption process. The lower the dryness, the higher the water vapor content, and the greater the risk of preemption of the effective sites of the adsorbent. This provides the core judgment basis for subsequent parameter adjustment. The physical parameters of the adsorbent are generally fixed parameters that cannot be modified. These parameters directly affect the bed airflow distribution and the service life of the adsorbent.By combining physical parameters of the adsorbent with adsorption kinetic experimental data, the peak parameters of the adsorbent are derived. Specifically, these are the maximum adsorption capacity of each layer of adsorbent for the corresponding target impurity, the saturation threshold of the effective adsorption site, and the impurity adsorption capacity decay coefficient under the preferential adsorption state of moisture. This accurately reflects the current adsorption capacity limit of the adsorbent, preventing impurity penetration due to adsorbent overload. The purification process parameters are intelligently adjustable and are the core of the entire purification system. They are synchronously acquired through pressure transmitters, temperature sensors, flow meters, and online dew point meters at the inlet and outlet of the PSA adsorption tower. Specifically, these parameters include the real-time pressure difference between the inlet and outlet of the adsorption tower, the real-time temperature rise at different axial positions of the bed, the actual feed flow rate of the raw gas, the initial dew point of the product gas, and the operating load of the pre-dehydration unit. These parameters dynamically reflect the operating status of the PSA adsorption stage, promptly capturing fluctuations in operating conditions such as bed blockage, abnormal adsorption heat, and incomplete water vapor removal. The first, second, and third characteristic variables are preprocessed and fused to obtain variable characteristics of a unified dimension, thereby achieving optimal performance. A hydrogen parameter prediction model is constructed using a machine learning-based regression prediction network. This ensures the model accurately maps the coupling relationship between feature variables and hydrogen purity, enabling early prediction of hydrogen purity. The model inputs the first, second, and third feature variables, collected and preprocessed in real-time, into the trained model. The model quickly outputs the predicted hydrogen purity value under the corresponding operating conditions, simultaneously outputting the predicted concentrations of key impurities and the predicted dew point of the product gas. This provides multi-dimensional reference for subsequent purity determination, avoiding misjudgments based solely on a single purity indicator. The difference between the predicted and preset hydrogen purity values is calculated to obtain the hydrogen purity deviation value. The deviation value is then determined to be within a preset range, set according to the actual application scenario. If the deviation value is within the preset range, the hydrogen purity meets the requirements under the current operating conditions, and the existing purification parameters can be maintained. If the deviation value exceeds the preset range, the hydrogen purification parameters are readjusted based on the magnitude and direction of the deviation value to achieve precise purity correction. When the predicted hydrogen purity is lower than the preset value, adjustments are made to the dryness index corresponding to the first characteristic variable. This is achieved by increasing the cooling efficiency of the precooler, increasing the regeneration frequency of the drying tower to enhance the dehydration process, and reducing the water concentration in the raw gas to improve dryness. Simultaneously, the adsorption time per tower is appropriately shortened to prevent water vapor from competing for effective adsorbent sites and causing impurities to penetrate prematurely. If the purity still does not meet the standard after adjustment, the adsorption pressure can be slightly increased to enhance the adsorption capacity of the adsorbent for impurities. When the predicted hydrogen purity is higher than the preset value, the deviation is positive, and it exceeds the range, it indicates that the current process parameters are redundant. In this case, the adsorption pressure can be appropriately reduced or the adsorption time extended.
[0071] Optionally, the step of obtaining the dryness of the raw gas based on the raw gas quality parameters and extracting the dryness of the raw gas as a first feature variable includes:
[0072] S21. Obtain the raw gas humidity data based on the raw gas quality parameters, wherein the raw gas humidity data includes water content concentration and water content temperature;
[0073] S22. Obtain the dryness of the raw gas based on the water content concentration and the water content temperature;
[0074] S23. A first correlation coefficient is generated based on the water content concentration, water content temperature, and feed gas dryness, and a first characteristic variable is generated based on the first correlation coefficient. Specifically, by accurately collecting water content concentration and water content temperature and correcting the calculated dryness, the first characteristic variable is generated by combining the correlation between the three. Compared with using dryness as a single characteristic variable, this method can more comprehensively and accurately depict the overall state of the feed gas humidity conditions, avoiding incomplete characterization of the characteristic variable due to ignoring the dynamic changes and synergistic effects of water content concentration and temperature, which would affect the prediction accuracy of the subsequent hydrogen parameter prediction model. This step, through layer-by-layer derivation and correlation analysis, enables the first characteristic variable to accurately locate water vapor. The core source of interference is the deviation in predicted purity. This can be quickly determined by the first characteristic variable to determine whether the abnormal dryness is caused by excessive water content concentration, temperature fluctuations, or the combined effect of both. This provides a clear direction for targeted adjustments to the pre-dehydration device and optimization of adsorption process parameters, avoiding blind adjustments. The first correlation coefficient can be used to predict the change in dryness under different combinations of water content, temperature, and concentration, allowing for advance adjustment of the cooler and drying tower operating parameters to stabilize the dryness of the feed gas within a reasonable range. This reduces damage to the adsorbent from the source, extends the adsorbent's service life, and reduces the risk of product purity failure due to humidity fluctuations, thereby improving the stability and continuity of the PSA hydrogen production process.
[0075] Optionally, the step of obtaining the physical parameters of the adsorbent based on the performance parameters of the adsorbent, obtaining the peak parameters of the adsorbent based on the physical parameters of the adsorbent, and extracting the peak parameters of the adsorbent as the second feature variable is as follows:
[0076] S31. Obtain the physical parameters of the adsorbent based on the adsorbent performance parameters, wherein the physical parameters of the adsorbent include the adsorbent planar area and the adsorbent thickness;
[0077] S32. Obtain the adsorption range based on the planar area of the adsorbent;
[0078] S33. Obtain the adsorption permeability based on the thickness of the adsorbent;
[0079] S34. Obtain the adsorbent saturation based on the adsorption range and the adsorption permeability;
[0080] S35. Obtain the peak parameters of the adsorbent based on the adsorbent saturation;
[0081] S36. Establish a table of peak value changes over several time periods based on the peak parameters of the adsorbent.
[0082] S37. A second correlation coefficient is generated based on the volatility of the peak change table for the specified time periods, and a second characteristic variable is generated based on the second correlation coefficient. Specifically, the adsorbent planar area is the effective adsorption cross-sectional area of each layer of adsorbent (activated alumina, silica gel, activated carbon, and molecular sieve) packed in the PSA adsorption tower, and the adsorbent thickness is the actual packing height of each layer of adsorbent. Combined with the particle size distribution of the adsorbent, the influence of particle packing gaps on the effective thickness is corrected to ensure that the parameters fit the actual adsorption conditions of the bed, providing an accurate basis for subsequent adsorption range and permeability calculations. Specifically, by combining the adsorbent's planar area with the feed gas flow rate and adsorption pressure, the effective coverage area of a single-stage adsorbent for the feed gas is calculated using the fluid dynamics seepage equation. This quantifies the adsorbent's ability to capture impurities in the feed gas. Darcy's law, combined with adsorbent thickness, is used to characterize the ease with which feed gas and impurity molecules penetrate the adsorbent bed. Greater thickness and denser particle packing result in lower permeability but longer adsorption contact time. Simultaneously, adsorbent porosity correction is incorporated to eliminate interference from ineffective pores on permeability, ensuring that the parameters accurately reflect the mass transfer efficiency of the adsorbent bed. The coupled calculation of range and permeability quantifies the ratio of the amount of impurity molecules adsorbed in the adsorbent bed per unit time to the maximum adsorption capacity of the bed, i.e., the adsorbent saturation. The saturation value ranges from 0 to 1, with a value closer to 1 indicating that the adsorbent is approaching its adsorption limit. Simultaneously, the raw material gas humidity condition, i.e., the first characteristic variable, is correlated to correct for the influence of moisture on the saturation of different adsorbents. For example, preferential adsorption of water vapor accelerates the increase in alumina layer saturation. The peak parameters of the adsorbent are the maximum adsorption capacity threshold and the peak saturation adsorption rate of each layer of adsorbent, specifically derived from the dynamic change curve of adsorbent saturation. When the rate of increase in saturation drops sharply and then levels off, the corresponding adsorption amount is the maximum adsorption capacity threshold, and the adsorption rate at this point is the peak saturation adsorption rate. Simultaneously, combining data on adsorbent lifespan and regeneration cycle, correction values for peak parameters at different usage stages are calibrated to ensure that the parameters are adapted to the actual operating conditions of adsorbent performance degradation. Time periods are divided according to the PSA adsorption cycle, and measured values of peak parameters for each layer of adsorbent are collected within each time period. A multi-dimensional variation table is constructed, synchronously recording operating parameters such as feed gas humidity, adsorption pressure, and bed temperature for the corresponding time period, clearly defining the changes in peak parameters. To avoid isolated analysis of peak data, the correlation with external operating conditions is considered. Specifically, the volatility of peak parameters in adjacent time periods is calculated to quantify the correlation between peak volatility and fluctuations in feed gas humidity, adjustments to adsorption process parameters, and adsorbent performance degradation. This generates a second correlation coefficient, which includes both the time-series fluctuation correlation of the peak parameters themselves and the cross-parameter correlation with external operating conditions. Finally, the adsorbent peak parameters, time-series volatility, and the second correlation coefficient are fused and normalized to generate a second characteristic variable, achieving a comprehensive characterization of the adsorbent's adsorption capacity, dynamic trends, and adaptability to operating conditions.
[0083] Optionally, the step of obtaining real-time purification status parameters based on the purification process parameters and extracting the real-time purification status parameters as a third feature variable includes:
[0084] S41. The purification process parameters include raw material flow rate and process temperature;
[0085] S42. Obtain the inlet flow rate and outlet flow rate of the raw material gas based on the raw material flow rate, and obtain the loss flow rate based on the inlet flow rate and the outlet flow rate;
[0086] S43. Obtain the intake time based on the intake flow rate;
[0087] S44. Obtain the air outlet time based on the air outlet flow rate;
[0088] S45. Obtain the adsorbent capacity parameter based on the loss flow rate, the inlet time, and the outlet time.
[0089] S46. Obtain the intake pressure for the intake time;
[0090] S47. Obtain the exhaust pressure during the exhaust time;
[0091] S48. Obtain the adsorbent pressure difference parameter based on the inlet pressure and the outlet pressure;
[0092] S49. Obtain the minimum and maximum adsorbent temperatures based on the process temperature.
[0093] S410. Obtain the adsorbent range temperature value based on the lowest and highest adsorbent temperatures.
[0094] S411. Generate a third correlation coefficient based on the adsorbent capacity parameter, the adsorbent differential pressure parameter, and the adsorbent range temperature value, and generate a third characteristic variable based on the third correlation coefficient.
[0095] The purification process parameters include raw material flow rate and process temperature. The raw material flow rate is the standard volumetric flow rate of the raw material gas at the PSA adsorption tower inlet, collected in real-time by a high-precision mass flow meter. The collection frequency is synchronized with the adsorption cycle to eliminate interference from instantaneous flow fluctuations. The process temperature covers the raw material gas inlet temperature, collected by distributed temperature sensors to ensure coverage of the entire process from raw material pretreatment to purification output, providing an accurate thermodynamic basis for subsequent parameter calculations. The inlet flow rate is the actual effective flow rate entering the adsorption tower during the adsorption stage, deducting the retention flow correction values from the pre-gas-liquid separator and drying tower. The outlet flow rate is the qualified product gas flow rate at the adsorption tower outlet, synchronously correlated with the adsorbent permeation side flow data to eliminate flow deviations caused by pipeline leaks. The loss flow rate is calculated through difference. By constructing an adsorption capacity kinetic model, the loss flow rate, inlet and outlet times are coupled with the adsorbent saturation, i.e., the second characteristic variable, to obtain the adsorption capacity per unit mass. The effective adsorption capacity of the adsorbent under current operating conditions is determined, and a loss flow correction coefficient is introduced to eliminate the interference of non-adsorption losses on the capacity parameter, accurately characterizing the real-time adsorption capacity of the adsorbent. The inlet pressure of the adsorption tower is simultaneously collected at the inlet time point with an accuracy of ±0.001 MPa, excluding pressure fluctuations during the pressurization phase. The average value after pressure stabilization is taken as the inlet pressure, which directly affects the adsorption balance of the adsorbent on impurities. At the outlet time point, the outlet pressure of the adsorption tower and the permeation side of the adsorbent are collected. The outlet pressure needs to be maintained stable to ensure subsequent transportation requirements. The pressure difference across the adsorbent directly affects the hydrogen permeation rate. The amplitude of outlet pressure fluctuations is recorded simultaneously to provide a basis for calculating the pressure difference parameter. The pressure difference of the adsorption tower bed is calculated by the difference between the inlet and outlet pressures. Combined with corrections based on bed height and adsorbent particle size distribution, the pressure drop per unit length of bed is obtained. This parameter can intuitively reflect whether there are problems such as pulverization, agglomeration, or blockage in the adsorbent bed. When the pressure difference exceeds 0...At 0.05 MPa, bed anomalies can be predicted. Based on data collected by distributed temperature sensors, the extreme temperature values at each monitoring point in the bed during the adsorption stage are extracted. The lowest temperature corresponds to the inlet temperature after the feed gas is cooled, and the highest temperature corresponds to the peak bed temperature caused by adsorption exothermics. The temperature rise during adsorption is typically ≤10℃. Simultaneously, the extreme temperature values of the adsorbent working section are collected to ensure that the temperature parameters cover the entire operating range of the adsorbent. By calculating the difference between the extreme temperature values of the adsorbent, the fluctuation range of the adsorbent's working temperature is obtained. Combined with the temperature range of the adsorbent bed, a weighted average method is used to generate the temperature values of the adsorbent range. The weight allocation is based on the difference in temperature sensitivity between different adsorbents. The effect of adsorbent temperature fluctuation on hydrogen permeation efficiency is discussed. The rate, weighted at 60%, quantifies the comprehensive impact of temperature fluctuations on adsorption and purification efficiency. A multivariate correlation analysis algorithm is employed to quantify the dynamic correlation among the three parameters. The third correlation coefficient includes the negative correlation coefficient between capacity and pressure difference parameters (the rate of decrease in capacity as pressure difference increases), the positive correlation coefficient between temperature range and capacity parameters (the magnitude of adsorption capacity change when temperature increases within a reasonable range), and the comprehensive influence coefficient of the synergistic effect of the three parameters on purification efficiency. Subsequently, the adsorbent capacity parameter, pressure difference parameter, temperature range, and the third correlation coefficient are normalized and fused to generate a third characteristic variable, achieving a comprehensive characterization of the real-time operating status, equipment condition, and purification efficiency of the PSA adsorption stage.
[0096] Optionally, the steps for obtaining variable features based on the first, second, and third feature variables, and synthesizing a hydrogen parameter prediction model using the variable features, are as follows:
[0097] S51. Obtain the set of coupling variables based on the first feature variable, the second feature variable, and the third feature variable;
[0098] S52. Obtain the model training set according to the generation process of the coupling variable set, and construct the model training sample library according to the model training set;
[0099] S53. Obtain multiple sample data from the model training sample library, and obtain variable features based on the mean value of the multiple sample data;
[0100] S54. Based on the generation process of the aforementioned variable characteristics, a hydrogen parameter prediction model is synthesized. Specifically,
[0101] A set of coupled variables is obtained based on the first, second, and third feature variables. Specifically, the three feature variables are first standardized and preprocessed. Then, through a feature fusion algorithm, the core information strongly correlated with hydrogen purity is extracted from the three feature variables to construct the set of coupled variables. This set retains the independent characteristics of feed gas humidity, adsorbent performance, and real-time operating conditions, while assigning dynamic weights to each feature through an attention mechanism. For example, the weight of feed gas dryness is 35%, the weight of adsorbent peak parameters is 30%, and the weight of bed pressure difference and temperature parameters is 35%. This accurately characterizes the synergistic influence of multi-dimensional parameters on hydrogen purity, providing high-quality input for subsequent model training. The model training set is derived from historical operating condition data and real-time acquired data of the PSA hydrogen production process. To facilitate subsequent targeted optimization of model parameters, the sample data in the training set are first grouped according to the working condition type. The mean, variance, and extreme values of each feature dimension of each group are calculated to obtain the statistical features of the working condition group. Then, the group features are corrected by combining the global sample mean to eliminate the bias of single working condition samples. Subsequently, the statistical features are fused with the core features of the coupled variable set to generate the final variable features. This feature is a high-dimensional vector that contains both the instantaneous characteristics of a single sample and the statistical regularity of similar working conditions. It can take into account the model's ability to respond to instantaneous working condition fluctuations and its ability to adapt to typical working conditions, avoiding insufficient model prediction accuracy due to individual sample differences. Multiple iterations of training ensure that the model can accurately map the coupling relationship between variable features and hydrogen purity.
[0102] Furthermore, a hydrogen production system for improving hydrogen purity includes...
[0103] The acquisition module is used to acquire various hydrogen purification parameters during the hydrogen purification process, wherein the hydrogen purification parameters include raw gas quality parameters, adsorbent performance parameters, and purification process parameters.
[0104] The raw gas dryness module is used to obtain the raw gas dryness from the raw gas quality parameters and extract the raw gas dryness as a first feature variable.
[0105] The adsorbent peak parameter module is used to obtain the physical parameters of the adsorbent from the adsorbent performance parameters, and to obtain the adsorbent peak parameters based on the physical parameters of the adsorbent, and to extract the adsorbent peak parameters as a second feature variable;
[0106] The real-time purification status parameter module is used to obtain real-time purification status parameters from the purification process parameters and extract the real-time purification status parameters as a third feature variable.
[0107] The hydrogen parameter prediction model module is used to obtain variable features from the first feature variable, the second feature variable, and the third feature variable, and to synthesize the hydrogen parameter prediction model through the variable features.
[0108] The identification module is used to generate a predicted hydrogen purity value from the hydrogen parameter prediction model.
[0109] The judgment module is used to compare the predicted hydrogen purity value with the preset hydrogen purity value to obtain the hydrogen purity deviation value, and to determine whether the hydrogen purity deviation value is within the preset value range.
[0110] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described control method for automatic inspection of PCB appearance defects.
[0111] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described control method for automatic inspection of PCB appearance defects.
[0112] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0113] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.
Claims
1. A method for improving hydrogen production purity, applied to the adsorption stage of a PSA adsorption process, characterized in that, include: Obtain hydrogen purification parameters, wherein the hydrogen purification parameters include raw gas quality parameters, adsorbent performance parameters, and purification process parameters; The dryness of the raw gas is obtained based on the raw gas quality parameters, and the dryness of the raw gas is extracted as the first feature variable. Based on the adsorbent performance parameters, the physical parameters of the adsorbent are obtained, and the peak parameters of the adsorbent are obtained based on the physical parameters of the adsorbent. The peak parameters of the adsorbent are then extracted as the second feature variable. Based on the purification process parameters, real-time purification status parameters are obtained, and these real-time purification status parameters are extracted as a third feature variable. Based on the first feature variable, the second feature variable, and the third feature variable, variable features are obtained, and a hydrogen parameter prediction model is synthesized using the variable features. Based on the hydrogen parameter prediction model, a predicted hydrogen purity value is generated; The hydrogen purity deviation value is obtained by comparing the predicted hydrogen purity value with the preset hydrogen purity value, and it is then determined whether the hydrogen purity deviation value is within the preset value range. If it is in the specified state, the hydrogen purity is determined to meet the requirements; If not, the hydrogen purification parameters will be readjusted based on the hydrogen purity deviation value.
2. The method for improving hydrogen purity according to claim 1, characterized in that, The step of obtaining the dryness of the raw gas based on the raw gas quality parameters and extracting the dryness of the raw gas as the first feature variable includes: The raw gas humidity data is obtained based on the raw gas quality parameters, wherein the raw gas humidity data includes water content concentration and water content temperature; The dryness of the raw gas is obtained based on the water content concentration and the water content temperature. A first correlation coefficient is generated based on the water content concentration, water content temperature, and raw material gas dryness, and a first characteristic variable is generated based on the first correlation coefficient.
3. The hydrogen production method for improving hydrogen purity as described in claim 1, characterized in that, The steps of obtaining the physical parameters of the adsorbent based on the performance parameters of the adsorbent, obtaining the peak parameters of the adsorbent based on the physical parameters of the adsorbent, and extracting the peak parameters of the adsorbent as the second feature variable are as follows: The physical parameters of the adsorbent are obtained based on the adsorbent performance parameters, wherein the physical parameters of the adsorbent include the adsorbent planar area and the adsorbent thickness; The adsorption range is obtained based on the planar area of the adsorbent. The adsorption permeability is obtained based on the thickness of the adsorbent. The adsorbent saturation is obtained based on the adsorption range and the adsorption permeability. The peak parameters of the adsorbent are obtained based on the adsorbent saturation. Establish a table of peak value changes over several time periods based on the peak parameters of the adsorbent; A second correlation coefficient is generated based on the volatility of the peak change table for the aforementioned several time periods, and a second feature variable is generated based on the second correlation coefficient.
4. The hydrogen production method for improving hydrogen purity as described in claim 1, characterized in that, The step of obtaining real-time purification status parameters based on the purification process parameters and extracting the real-time purification status parameters as a third feature variable includes: The purification process parameters include raw material flow rate and process temperature; The inlet and outlet flow rates of the raw material gas are obtained based on the raw material flow rate, and the loss flow rate is obtained based on the inlet and outlet flow rates. The intake time is obtained based on the intake flow rate. The air outlet time is obtained based on the air outlet flow rate. The adsorbent capacity parameter is obtained based on the loss flow rate, the inlet time, and the outlet time. Obtain the intake pressure at the intake time; Obtain the exhaust pressure during the exhaust time; The adsorbent pressure difference parameter is obtained based on the inlet pressure and the outlet pressure. The minimum and maximum temperatures of the adsorbent are obtained based on the process temperature. The temperature range of the adsorbent is obtained based on the lowest and highest temperatures of the adsorbent. A third correlation coefficient is generated based on the adsorbent capacity parameter, the adsorbent differential pressure parameter, and the adsorbent temperature range value, and a third characteristic variable is generated based on the third correlation coefficient.
5. The hydrogen production method for improving hydrogen purity as described in claim 1, characterized in that, The steps for obtaining variable features based on the first, second, and third feature variables, and synthesizing a hydrogen parameter prediction model using these variable features, are as follows: The set of coupling variables is obtained based on the first feature variable, the second feature variable, and the third feature variable; The model training set is obtained according to the generation process of the coupling variable set, and a model training sample library is constructed based on the model training set; Obtain multiple sample data from the model training sample library, and derive variable features based on the mean of the multiple sample data; A hydrogen parameter prediction model is synthesized based on the generation process of the aforementioned variable characteristics.
6. A hydrogen production system for improving hydrogen purity, characterized in that, include The acquisition module is used to acquire various hydrogen purification parameters during the hydrogen purification process, wherein the hydrogen purification parameters include raw gas quality parameters, adsorbent performance parameters, and purification process parameters. The raw gas dryness module is used to obtain the raw gas dryness from the raw gas quality parameters and extract the raw gas dryness as a first feature variable. The adsorbent peak parameter module is used to obtain the physical parameters of the adsorbent from the adsorbent performance parameters, and to obtain the adsorbent peak parameters based on the physical parameters of the adsorbent, and to extract the adsorbent peak parameters as a second feature variable; The real-time purification status parameter module is used to obtain real-time purification status parameters from the purification process parameters and extract the real-time purification status parameters as a third feature variable. The hydrogen parameter prediction model module is used to obtain variable features from the first feature variable, the second feature variable, and the third feature variable, and to synthesize the hydrogen parameter prediction model through the variable features. The identification module is used to generate a predicted hydrogen purity value from the hydrogen parameter prediction model. The judgment module is used to compare the predicted hydrogen purity value with the preset hydrogen purity value to obtain the hydrogen purity deviation value, and to determine whether the hydrogen purity deviation value is within the preset value range.
7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.