Nitrogen content measurement method and apparatus, and storage medium

By establishing and fitting soft measurement models, real-time prediction and fusion of nitrogen content in argon fraction, the problem of real-time online detection in the air separation device is solved, efficient and flexible online measurement and control are achieved, and the stability of the device is improved and argon yield.

WO2024235216A1PCT designated stage expired Publication Date: 2024-11-21SHANGHAI LIFENGAS CO LTD
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
PCT/CN2024/093024
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-15
Filing Date
2024-05-14
Publication Date
2024-11-21

AI Technical Summary

Technical Problem

The existing air separation device failed to realize real-time online detection of nitrogen content in the argon fraction, resulting in nitrogen plug problems and low argon yield, which could not meet the online control requirements.

Method used

By establishing an initial model based on historical running data, fitting the soft measurement model, combining the detection values ​​of the associated variables to predict the nitrogen content in real time, and fusing it with the actual nitrogen content obtained by periodic sampling, real-time online measurement is achieved.

Benefits of technology

Real-time online measurement is realized, the response speed can reach second level, reduce costs, improve detection accuracy and flexibility, meet the requirements of online closed-loop control, reduce the probability of nitrogen plug generation and improve argon gas production.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2024093024_21112024_PF_FP_ABST
    Figure CN2024093024_21112024_PF_FP_ABST
Patent Text Reader

Abstract

A nitrogen content measurement method and apparatus, and a storage medium. The method comprises the following steps: determining an associated variable related to a nitrogen content; establishing an initial model, which is based on historical operation data, then using the associated variable as an independent variable, using the nitrogen content as a dependent variable, and performing calculation to obtain parameters of the initial model; on the basis of the parameters of the initial model, performing fitting to obtain parameters of a soft measurement model; predicting the nitrogen content in real time according to a measured value of the acquired associated variable and on the basis of the soft measurement model; and obtaining a fusion measurement result on the basis of the predicted nitrogen content and an actual nitrogen content that is acquired by means of periodic sampling.
Need to check novelty before this filing date? Find Prior Art

Description

A nitrogen content measurement method, device and storage medium Technical Field

[0001] The present invention relates to the technical field of air separation, and in particular to a nitrogen content measuring method, device and storage medium. Background Art

[0002] Argon is an inert gas with stable chemical properties and is not easily reactive with other substances. In industry, argon is widely used in metal welding, smelting and processing, semiconductor production, laser cutting, and medical imaging. Argon is crucial in these applications because it provides a protective, stable, and inert environment.

[0003] When the nitrogen content in the argon fraction of an air separation unit is excessively high and persists for an extended period, it can cause nitrogen plugging in the crude argon column condenser, disrupting the unit's distillation process and affecting normal argon production. Excessively high nitrogen content in the argon fraction can affect the unit's stable operation; however, low nitrogen content in the argon fraction indicates a low argon extraction rate and potential for increased production, requiring measures to improve argon production. Therefore, measuring the nitrogen content in the argon fraction is crucial for increasing argon production in air separation units.

[0004] Most current air separation units do not measure the nitrogen content of the argon fraction online. The few that do use specialized TCD or PDD analyzers. While these analyzers address the issue of online measurement, their detection cycles are long (10 to 30 minutes), resulting in inadequate nitrogen content feedback and insufficient timeliness to meet online control requirements.

[0005] Summary of the Invention

[0006] The purpose of the present invention is to provide a nitrogen content measurement method, device and storage medium, which adjust the detection results of the analyzer through a soft measurement method to achieve automatic online calibration, which is used to prevent nitrogen blockage in air separation units.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A method for measuring nitrogen content comprises the following steps:

[0009] Identify associated variables related to nitrogen content;

[0010] Establishing an initial model based on historical operating data, with associated variables as independent variables and nitrogen content as dependent variables, and calculating parameters of the initial model;

[0011] Based on the parameters of the initial model, fitting the parameters of the soft sensor model is performed;

[0012] Predicting nitrogen content in real time based on the obtained detection values ​​of the associated variables and the soft sensor model;

[0013] Based on the predicted nitrogen content and the actual nitrogen content obtained by periodic sampling, a fusion detection result is obtained.

[0014] In some embodiments, obtaining a fusion detection result based on the predicted nitrogen content and the actual nitrogen content obtained by sampling further includes:

[0015] Calculation AI1 K =j·AI1 K-1 +(1-j)·AI1 b +ΔAI1 aK

[0016] Among them, AI1 K is the fusion detection result at time K, AI1 b is the actual nitrogen content obtained, ΔAI1 aK The difference between the nitrogen content predicted at time K and the nitrogen content predicted at time K-1,

[0017] Wherein, j={0,1}, when time K is the sampling time of the actual nitrogen content, set j=0, otherwise, set j=1.

[0018] In some embodiments, the soft sensor model is:

[0019] in: is the predicted nitrogen content; x1~x m They correspond to m associated variables respectively; β0~β m are the soft-sensor parameters obtained by fitting.

[0020] In some embodiments, establishing the initial model further comprises:

[0021] The historical operation data is processed and expressed as: y i =β i0 +β i1 x i1 +β i2 x i2 +β i3 x i3 +β i4 x i4 +β i5 x i5 +β i6 x i6 ...+β in x in +ε i

[0022] The variable relationship for representing the obtained n-th group of historical operation data, wherein i=1, 2, 3...n, n is the total number of groups of historical operation data, ε i is a random error;

[0023] The obtained n groups of historical operation data are expressed in matrix form as follows: Y = Cβ + ε

[0024] Where Y=[y1…y i …y n ] T ε=[ε1…ε i …ε n ] T .

[0025] In some embodiments, fitting the parameters of the soft sensor model based on the parameters of the initial model includes:

[0026] Using the least squares regression algorithm, we can solve:

[0027] based on Calculate the parameters of the soft sensor model β0~β m .

[0028] In some embodiments, the associated variables include at least one of liquid air purity, argon fraction amount, air amount, argon fraction purity, upper column pressure, and argon fraction temperature.

[0029] In some embodiments, the source of the historical operation data includes at least one of an existing database, a real-time detection system, simulation, and experiment.

[0030] In some embodiments, the historical operating data is data under multiple operating conditions, wherein the operating conditions include at least one of a maximum oxygen production operating condition, a maximum liquid production operating condition, a minimum gas production operating condition, and a minimum liquid production operating condition.

[0031] A nitrogen content measuring device comprises a memory, a processor, and a program stored in the memory, wherein the processor implements the method described above when executing the program.

[0032] A storage medium stores a program thereon, and when the program is executed, the method described above is implemented.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] (1) Low cost and easy to deploy: The fusion of the soft measurement model of the present invention and the detection results of the analyzer is realized by computer, without the need for additional equipment. In addition, the soft measurement technology only needs to use existing data for modeling and calculation, which is easy to deploy.

[0035] (2) Strong scalability: The soft measurement model can adapt to different processes and variables and can be adjusted as the process changes. Therefore, it has higher flexibility and scalability. When the environment changes, better adaptation can be achieved by changing the representation of the soft measurement model.

[0036] (3) Higher timeliness: The soft measurement model can predict the measurement results in real time based on the actual operation data, with a response speed of seconds, enabling real-time online measurement.

[0037] (4) High detection accuracy: By combining the soft measurement model and the analyzer, the present invention can utilize the high timeliness of the soft measurement model to overcome the defect of the long detection time of the analyzer and meet the requirements of online closed-loop control. At the same time, the soft measurement model is regularly calibrated using the detection results of the analyzer to overcome the problem of accuracy degradation caused by the long-term accumulation of soft measurement errors and improve the reliability of the detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] FIG1 is a flow chart of a method according to an embodiment of the present invention;

[0039] FIG2 is a schematic diagram of a measurement process according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0041] This embodiment provides a nitrogen content measurement method for predicting the nitrogen content of the argon fraction based on easily measurable variables with online data in an air separation unit, and then obtaining a fusion detection result based on the predicted nitrogen content and the actual nitrogen content obtained by periodic sampling.

[0042] As shown in FIG1 , the method includes the following steps:

[0043] S1, determine the associated variables related to nitrogen content;

[0044] S2, establishing an initial model, wherein the initial model is based on historical operating data, takes the associated variable as an independent variable, and the nitrogen content as a dependent variable, and calculates the parameters of the initial model;

[0045] S3, fitting the parameters of the soft sensor model based on the parameters of the initial model;

[0046] S4, predicting the nitrogen content in real time according to the obtained detection values ​​of the associated variables and based on the soft sensor model; and

[0047] S5, based on the predicted nitrogen content and the actual nitrogen content obtained by periodic sampling, obtain a fusion detection result.

[0048] Optionally, S1: determining a correlation variable related to the nitrogen content in the argon fraction.

[0049] Correlated variables include at least one of: liquid air purity, argon fraction, air volume, argon fraction purity, upper column pressure, and argon fraction temperature. The measured values ​​of these correlated variables can be easily obtained through measurement, such as online instrumentation. Air separation units have complex processes and significant coupling, making the selection of correlated variables for soft sensing crucial. Choosing appropriate correlated variables can reduce the difficulty of soft-sensing fitting and improve measurement accuracy.

[0050] The associated variables selected in this embodiment are shown in Table 1.

[0051] Table 1 Correlation variables

[0052] Optionally, S2: Establish an initial model of a soft-sensing model for nitrogen content in the argon fraction, wherein the initial model uses historical operating data of the associated variable as an independent variable and the historical operating data of nitrogen content as a dependent variable. The soft-sensing model uses the detected value of the associated variable as an independent variable and the predicted nitrogen content as a dependent variable.

[0053] This embodiment uses a multivariate linear regression approach to establish a soft measurement model, and the model format is as follows:

[0054] in: is the nitrogen content estimated by the soft measurement; x1-x6 correspond to the associated variables numbered 1-6 in Table 1; β0-β6 are the parameters of the soft measurement model. This embodiment uses six associated variables as an example, but it can also be 1, 2, 3, or m associated variables.

[0055] Optionally, S3: acquiring historical operation data, and fitting soft sensor model parameters using the historical operation data.

[0056] The sources of historical operating data include at least one of an existing database, a real-time monitoring system, simulation, and experimentation. Soft measurement requires a large amount of historical operating data to train the model, so data related to the soft measurement target must be collected. This data can be obtained from an existing database or real-time monitoring system, or generated through experimentation or simulation. To ensure the accuracy and robustness of the soft measurement model, as many sets of historical operating data as possible should be collected, especially multiple sets of historical operating data under different operating conditions.

[0057] To this end, this embodiment takes the following operations:

[0058] ① Calibrate all instruments of the associated variables in Table 1 to ensure the accuracy of their detection;

[0059] ②Adjust the operating conditions of the device

[0060] Because air separation units experience relatively low load variations, some units, despite long operating years, operate under relatively simple conditions, making it difficult to accumulate sufficient data from a wide range of operating conditions. Therefore, it is necessary to adjust the unit's operating conditions, including but not limited to maximum oxygen production, maximum liquid production, minimum gas production, and minimum liquid production, to accumulate sufficient and valuable actual data.

[0061] ③Read data from the real-time database

[0062] This embodiment uses Microsoft Excel software to access the actual DCS real-time database online through the OPC protocol to obtain historical operating data.

[0063] In order to achieve the best fitting effect, before using historical operation data to fit the soft sensor model parameters, the historical operation data is processed and expressed as: i =β i0 +β i1 x i1 +β i2 x i2 +β i3 x i3 +β i4 x i4 +β i5 x i5 +β i6 x i6 +ε i

[0064] The above formula represents the variable relationship of the actual operating data obtained under each set of different working conditions, for example, the historical operating data of the nth group, where i = 1, 2, 3...n, n is the total number of groups of historical operating data, ε i is a random error;

[0065] The above formula is expressed in matrix form as follows: Y = Cβ + ε

[0066] Where, Y=[y1…y i …y n ] T ε=[ε1…ε i …εn ] T

[0067] The parameters of the soft sensor model are fitted using historical operation data and the least squares regression algorithm is used to solve the problem:

[0068] The fitted soft sensor model is:

[0069] based on Calculate the parameters of the soft sensor model β0~β m .

[0070] Compared with online analyzers, soft measurement methods have the following advantages:

[0071] 1) Lower cost: Online analyzers require the purchase, installation, maintenance, and updating of equipment, while soft sensing technology can use existing process data for modeling and calculation, so the cost is lower.

[0072] 2) Easy to deploy: Online analyzers require installation and debugging, while soft sensing technology only needs to use existing data for modeling and calculation, so it is easier to deploy.

[0073] 3) Scalability: Soft sensing technology can adapt to different processes and variables and can be adjusted as the process changes, so it has higher flexibility and scalability.

[0074] 4) Higher timeliness: Soft measurement methods can predict measurement results in real time based on actual operating data, with a response speed of seconds.

[0075] Therefore, this embodiment adopts a combination of soft measurement and analyzer to accelerate the sampling frequency of component detection, meet the requirements of online closed-loop control, and achieve real-time and reliable online detection while retaining high detection accuracy, thereby providing a reliable data source for real-time online control of air separation devices.

[0076] The analyzer detection results obtained by periodic sampling can be used to periodically correct the fusion detection results.

[0077] The detection process of the analyzer is generally long. The sampling period of the analyzer detection used in this embodiment is 30 minutes. If only the analyzer is used, the demand for real-time online control cannot be met.

[0078] Optionally, S4: acquiring in real time the associated variable value related to the nitrogen content in the argon fraction in the air separation unit, and predicting the current nitrogen content in the argon fraction based on the soft sensor model established in S3.

[0079] Optionally, S5: adjusting the current analyzer detection result in real time based on the nitrogen content predicted by the soft measurement model to obtain a fusion detection result of the nitrogen content.

[0080] This embodiment uses a periodic calibration method to fuse the soft measurement results with the analyzer results. In the actual fusion process, it is hoped that the online analyzer can calibrate the fusion results after each test, so an incremental algorithm is required. Specifically: AI1 K =j·AI1 K-1 +(1-j)·AI1 b +ΔAI1 aK

[0081] Among them, AI1 K is the fusion detection result at time K, AI1 b is the actual nitrogen content obtained, ΔAI1 aK The difference between the nitrogen content predicted at time K and the nitrogen content predicted at time K-1,

[0082] Wherein, j={0,1}, when time K is the sampling time of the actual nitrogen content, set j=0, otherwise, set j=1.

[0083] For example, each time the analyzer completes sampling, j=0, and each soft measurement calculation cycle after the analyzer completes sampling, j=1.

[0084] As shown in Figure 2, the analyzer test results are updated every half hour, that is, the actual nitrogen content obtained by periodic sampling. After the analyzer test results are updated, before a soft measurement calculation, AI1 K =AI1 b +ΔAI1 aK

[0085] That is, the fusion test result at this time is based on the actual nitrogen content AI1 b As a benchmark.

[0086] In a soft measurement calculation cycle (usually 1 second), based on the detection values ​​of the associated variables, the nitrogen content is predicted in real time, the soft measurement model output is updated, and the fusion detection result is obtained as follows: AI1 K =AI1 K-1 +ΔAI1 aK

[0087] The fusion detection result AI1 at the previous moment K-1 As a benchmark, the difference between the two predicted nitrogen contents ΔAI1 is used. aK Make adjustments.

[0088] Therefore, half an hour before the next analyzer completes sampling and feeds back the test results, the nitrogen content test results can be continuously adjusted every second according to the soft measurement model, with strong real-time performance.

[0089] Based on the above measurement method, online commissioning and cross-validation are carried out.

[0090] Online commissioning:

[0091] (1) Determine the process data collected in real time: Before the soft measurement model is put into use online, it is necessary to determine the process data collected in real time so that the model can be monitored and controlled in real time.

[0092] (2) Real-time monitoring and control: Apply the soft measurement model to the real-time collected process data for real-time monitoring and control. Make control decisions based on the output of the soft measurement model.

[0093] Cross-validation:

[0094] (1) Model validation: Apply the test set to the soft sensor model, calculate the error between the model prediction results and the actual results, and evaluate the generalization ability of the model.

[0095] (2) Evaluate the degree of matching between the model and the actual situation. Generally, if the matching degree between the soft measurement results and the actual detection results is above 95%, it is considered to be usable. Otherwise, the model needs to be revised. However, the specific requirements are also related to the measurement accuracy requirements of the actual process and cannot be generalized.

[0096] Based on the above method, the equipment required in the implementation of this embodiment is shown in Table 2.

[0097] Table 2 Equipment required for implementation

[0098] In one embodiment, the method is applied to a large air separation unit in a steel plant for online control. After commissioning, the probability of nitrogen plugging in the air separation unit is reduced from an average of 3 times / month to 1 time / month, thereby improving the stability and reliability of the unit operation. At the same time, the argon extraction rate is increased by 3% (approximately 2 tons / day), increasing the economic benefits of the unit by approximately RMB 700,000 / year.

[0099] When an air separation unit is equipped with an analyzer for nitrogen content in the argon fraction, the present invention, by combining soft measurement with an online analyzer, can accelerate the sampling frequency of component detection, significantly improving detection speed and meeting the requirements of online closed-loop control. This also reduces the accuracy degradation caused by long-term soft measurement error accumulation, avoids the need for frequent maintenance (ranging from several weeks to six months), and improves the reliability of detection results. It can achieve online adjustment of the nitrogen content in the argon fraction, automatically optimize the operating conditions of the air separation unit's argon system, improve the unit's automation level, enhance its operational stability, and increase its economic benefits.

[0100] In another embodiment, when there is no analyzer in the device to detect the nitrogen content of the argon fraction, the soft measurement method of the present invention can be used as a reference to assist the operator to understand the working conditions in a timely manner, intervene in the working conditions early, reduce the probability of nitrogen blockage in the argon system, and increase the argon production of the device.

[0101] The present invention is applicable to all air separation units with argon and has strong scalability and replicability. The present invention can be deployed independently or in a combination of soft and hard cyclic correction, which can increase the argon production of air separation units and has strong economic value.

[0102] In another embodiment, the soft measurement algorithm can be improved by combining more actual device operating parameters to improve the accuracy of soft measurement, reduce dependence on analytical instruments, and thus further reduce costs.

[0103] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0104] The above describes in detail the preferred embodiments of the present invention. It should be understood that those skilled in the art can make numerous modifications and variations based on the concepts of the present invention without inventive effort. Therefore, any technical solutions that can be derived by those skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A method for measuring nitrogen content, characterized in that: The following steps are involved: Identify associated variables related to nitrogen content; Establishing an initial model, wherein the initial model is based on historical operation data, takes the associated variable as an independent variable, and the nitrogen content as a dependent variable, and calculates the parameters of the initial model; Based on the parameters of the initial model, fitting obtains the parameters of the soft sensor model; According to the obtained detection values ​​of the associated variables and based on the soft sensor model, the nitrogen content is predicted in real time; Based on the predicted nitrogen content and the actual nitrogen content obtained by periodic sampling, a fusion detection result is obtained.

2. The method according to claim 1, characterized in that The step of obtaining a fusion detection result based on the predicted nitrogen content and the actual nitrogen content obtained by sampling further includes: calculate AI1 K =j·AI1 K-1 +(1-j)·AI1 b +ΔAI1 aK Among them, AI1 K is the fusion detection result at time K, AI1 b is the actual nitrogen content obtained, ΔAI1 aK is the difference between the nitrogen content predicted at time K and the nitrogen content predicted at time K-1, Wherein, j={0,1}, when the K moment is the sampling moment of the actual nitrogen content, j=0 is set, otherwise, j=1 is set.

3. The method for measuring nitrogen content in argon fraction by combining soft sensor and analyzer according to claim 1, characterized in that: The soft sensor model is: in: is the predicted nitrogen content; x1~x m They correspond to m associated variables respectively; β0~β m are the soft-sensor parameters obtained by fitting.

4. The method according to claim 3, characterized in that The establishing of the initial model further includes: processing the historical operation data and expressing it as: y i =β i0 +β i1 x i1 +β i2 x i2 +β i3 x i3 +β i4 x i4 +β i5 x i5 +β i6 x i6 ……+β in x in +ε i The variable relationship for representing the obtained nth group of historical operation data, wherein i=1, 2, 3...n, n is the total number of groups of historical operation data, ε i is a random error; The obtained n groups of historical operation data are expressed in matrix form as follows: Y=Cβ+ε Among them, Y = [y1 … y i … y n T ​ ε=[ε1 … ε i … eh n ] T 。 5. The method according to claim 4, characterized in that The fitting of the parameters of the soft sensor model based on the parameters of the initial model includes: Using the least squares regression algorithm, we can solve: based on The parameters β0~β of the soft sensor model are calculated m .

6. The method according to any one of claims 1 to 5, characterized in that: The associated variables include: at least one of: liquid air purity, argon fraction amount, air amount, argon fraction purity, upper column pressure and argon fraction temperature.

7. The method according to any one of claims 1 to 6, characterized in that: The source of the historical operation data includes: at least one of an existing database, a real-time detection system, simulation and experiment.

8. The method according to any one of claims 1 to 7, characterized in that: The historical operation data is data under multiple operating conditions, and the operating conditions include: at least one of a maximum oxygen production condition, a maximum liquid production condition, a minimum gas production condition, and a minimum liquid production condition.

9. A nitrogen content measuring device, comprising a memory, a processor, and a program stored in the memory, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 8 is implemented.

10. A storage medium having a program stored thereon, characterized in that: When the program is executed, the method according to any one of claims 1 to 8 is implemented.

Citation Information

Patent Citations

  • Method for preventing and controlling nitrogen blockage of argon preparation system of space division device

    CN101634837A

  • Nitrogen blockage prevention and control method for argon generation system of air separation device

    CN106642992A

  • Forecasting method of SO2 concentration value in city

    CN108228538A

  • Model predictive control method for an air-seperation system

    CN1126307A

  • Air separation plant argon fraction argon content correlation analysis method based on time-lag Pearson's

    CN115310254A