Nitrogen production regulation system, method, device and storage medium

Abstract

The embodiment of the present specification provides a nitrogen production regulation system, method, device and storage medium, the system comprises an air inlet unit, a gas production unit and a processor, the air inlet unit comprises an air inlet channel, a first sensing unit group, a second sensing unit group and a compression device, a cooling device and a drying device. The processor is configured to obtain the gas characteristics of the raw gas from the first sensing unit group; generate the compression parameter according to the gas characteristics of the raw gas; control the compression device to process the raw gas according to the compression parameter to obtain the first intermediate gas; obtain the gas characteristics of the first intermediate gas from the second sensing unit group; generate the cooling parameter and the drying parameter according to the gas characteristics of the first intermediate gas; control the cooling device and the drying device to process the first intermediate gas according to the cooling parameter and the drying parameter to obtain the second intermediate gas. The gas production unit is configured to produce nitrogen based on the second intermediate gas.

Description

TECHNICAL FIELD

[0001] The present specification relates to the technical field of nitrogen production, and in particular to a nitrogen production regulation system, method, device and storage medium. BACKGROUND

[0002] In the technical field of nitrogen production equipment, the pretreatment sub-equipment is the core link to ensure the efficiency and quality of nitrogen production. In the prior art, the pretreatment sub-equipment usually includes air compressor, cooler, freeze dryer, filter and other components, which are responsible for compression, cooling, drying and filtering of air or industrial waste gas. However, in the compression, cooling and drying stages of the existing pretreatment sub-equipment, the parameters of each stage (such as compression pressure, cooling temperature) are fixedly set or adjusted depending on manual experience, and cannot be dynamically optimized according to real-time working conditions.

[0003] In view of this, it is desirable to provide an improved nitrogen production regulation system, method, device and storage medium, which adjusts the parameters of the compression device, cooling device and drying device in the pretreatment sub-equipment, to improve the efficiency of nitrogen production, shorten the nitrogen production cycle and ensure the quality of nitrogen production. SUMMARY

[0004] One or more embodiments of the present specification provide a nitrogen production regulation system, which includes an air inlet unit, a gas production unit and a processor, the air inlet unit includes an air inlet channel, a first sensing unit group, a second sensing unit group and a compression device, a cooling device and a drying device; the processor is configured to execute a nitrogen production regulation method; the gas production unit is configured to produce nitrogen based on the second intermediate gas.

[0005] One or more embodiments of the present specification provide a nitrogen production regulation method, which includes: obtaining the gas characteristics of the original gas from the first sensing unit group; generating compression parameters according to the gas characteristics of the original gas; controlling the compression device to process the original gas according to the compression parameters to obtain the first intermediate gas; obtaining the gas characteristics of the first intermediate gas from the second sensing unit group; generating cooling parameters and drying parameters according to the gas characteristics of the first intermediate gas; controlling the cooling device and the drying device to process the first intermediate gas according to the cooling parameters and the drying parameters to obtain the second intermediate gas; and controlling the gas production unit to produce nitrogen based on the second intermediate gas.

[0006] One or more embodiments of the present specification provide a nitrogen production regulation device, which includes a processor configured to execute the nitrogen production regulation method as described in any one of the above.

[0007] The one or more embodiments of the present specification provide a computer readable storage medium storing computer instructions, when the computer reads the computer instructions in the storage medium, the computer executes the nitrogen production regulation method according to any one of the above. BRIEF DESCRIPTION OF DRAWINGS

[0008] The present specification will be further described in the manner of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same reference numbers denote the same structures, wherein:

[0009] Figure 1 is a system schematic diagram of a nitrogen production regulation system according to some embodiments of the present specification;

[0010] Figure 2 is an exemplary flowchart of a nitrogen production regulation method according to some embodiments of the present specification;

[0011] Figure 3 is an exemplary schematic diagram of generating cooling parameters and drying parameters according to some embodiments of the present specification;

[0012] Figure 4 is an exemplary flowchart of controlling the operation of the recovery unit according to some embodiments of the present specification. DETAILED DESCRIPTION

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present specification, the drawings required to be used in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some examples or embodiments of the present specification, and for those skilled in the art, the present specification can be applied to other similar scenarios without creative labor on the basis of these drawings. Unless it is obvious from the language environment or otherwise stated, the same reference numbers in the drawings represent the same structures or operations.

[0014] Flowcharts are used in the present specification to illustrate the operations performed by the system according to the embodiments of the present specification. It should be understood that the preceding or subsequent operations are not necessarily performed in sequence. On the contrary, each step can be processed in reverse order or simultaneously. At the same time, other operations can be added to these processes, or one or more steps of operations can be removed from these processes.

[0015] Figure 1 is a system schematic diagram of a nitrogen production regulation system according to some embodiments of the present specification.

[0016] In some embodiments, the nitrogen production regulation system 100 can include an air inlet unit 110, a gas production unit 120, and a processor 130.

[0017] The gas inlet unit 110 refers to a unit module for collecting raw gas into the nitrogen production control system. In some embodiments, the gas inlet unit 110 can include a gas inlet passage 111, a first sensing unit group 112, a second sensing unit group 113, a compression device 114, a cooling device 115, and a drying device 116.

[0018] The gas inlet passage 111 refers to a gas passage for collecting raw gas into the nitrogen production control system.

[0019] In some embodiments, the first sensing unit group 112 can be configured to collect physical quantities of the raw gas. For example, the first sensing unit group 112 can include a flow sensor, a temperature sensor, a humidity sensor, a pressure sensor, etc.

[0020] In some embodiments, the second sensing unit group 113 can be configured to collect physical quantities of the first intermediate gas. For example, the second sensing unit group 113 can include a flow sensor, a temperature sensor, a humidity sensor, a pressure sensor, etc.

[0021] In some embodiments, the second sensing unit group 113 can further include an oil content sensor. The oil content sensor refers to a sensor for detecting the content of oil in the gas. The content of oil can affect the dew point of the first intermediate gas, thereby affecting the subsequent determined cooling parameters.

[0022] By detecting the content of oil in the first intermediate gas, the content of oil can be considered when determining the cooling parameters subsequently, thereby improving the accuracy of the determined cooling parameters.

[0023] The compression device 114 refers to a device for compressing gas and increasing the pressure of the gas. In some embodiments, the compression device 114 can include a piston compression device, a screw compression device, etc.

[0024] The cooling device 115 refers to a device for reducing the temperature of the compressed gas. In some embodiments, the cooling device 115 can include a freeze dryer, an air cooler, etc.

[0025] The drying device 116 refers to a device for drying gas. In some embodiments, the drying device 116 can include an adsorption dryer, a freeze dryer, etc.

[0026] The gas production unit 120 refers to a unit module for producing nitrogen. In some embodiments, the gas production unit 120 can be configured to produce nitrogen based on the second intermediate gas.

[0027] In some embodiments, the nitrogen production control system 100 can further include a recovery unit 140 and a third sensing unit group 150.

[0028] The recovery unit 140 refers to a unit module for recovering the separated gas. In some embodiments, the recovery unit 140 can be configured to recover the separated gas.

[0029] More details about the separated gas can be found in the related description of Figure 4 .

[0030] The third sensing unit group 150 refers to a combination of sensors for processing the separated gas. In some embodiments, the third sensing unit group 150 can be configured to collect the local component features of the separated gas. For example, the third sensing unit group 150 can include semiconductor gas sensors, electrochemical sensors, and thermal conductivity sensors, etc.

[0031] In some embodiments, the sensors in the third sensing unit group 150 can also be configured to detect and measure specific gas components. For example, to collect the content of low chemical inertness and gas components with identifiable concentrations in the separated gas. Wherein, the identifiable concentration refers to the concentration that can be detected by the sensor.

[0032] The processor 130 refers to a computing device for processing instructions, processing data, and performing operations, etc. In some embodiments, the processor 130 can be configured to obtain the gas features of the raw gas from the first sensing unit group; generate compression parameters according to the gas features of the raw gas; control the compression device to process the raw gas according to the compression parameters to obtain the first intermediate gas; obtain the gas features of the first intermediate gas from the second sensing unit group; generate cooling parameters and drying parameters according to the gas features of the first intermediate gas; and control the cooling device and the drying device to process the first intermediate gas according to the cooling parameters and the drying parameters to obtain the second intermediate gas.

[0033] More details about the components of the nitrogen production regulation system can be found in the related description of Figures 2-4 .

[0034] It should be noted that the above description of the nitrogen production regulation system and its modules is for convenience of description only, and cannot limit the scope of the embodiments. It can be understood that, for those skilled in the art, after understanding the principle of the system, the modules can be combined arbitrarily or connected with other modules to form a subsystem without departing from the principle. In some embodiments, Figure 1 The intake unit, the gas production unit, the processor, the recovery unit, and the third sensing unit group disclosed in the specification can be different modules in a system, or one module can implement the functions of two or more modules described above. For example, each module can share a storage module, and each module can also have its own storage module. Variations such as this are within the scope of protection of the specification.

[0035] Figure 2 is an exemplary flowchart of a nitrogen production regulation method according to some embodiments of the present specification. As shown in Figure 2 , the flow 200 includes the following steps. In some embodiments, the flow 200 can be executed by a processor.

[0036] Step 210, obtaining a gas feature of a raw gas from a first sensor unit group.

[0037] The raw gas refers to the raw material for nitrogen production. For example, air, industrial gas, etc. In some embodiments, the raw gas can be obtained in various ways. For example, air is extracted by a blower, obtained from a factory, etc.

[0038] In some embodiments, the gas feature can include the flow rate, pressure, temperature, humidity, etc. of the gas.

[0039] Step 220, generating a compression parameter according to the gas feature of the raw gas.

[0040] The compression parameter refers to the working parameter of the compression device. For example, the pressure value that the raw gas needs to reach after compression, the speed of the compression device, etc.

[0041] In some embodiments, the processor can construct a first data table based on historical experience and historical production data, the first data table including historical gas features, historical compression parameters and their corresponding relationship. In some embodiments, the processor can determine the same or similar historical gas features based on the current gas feature by querying the first data table, and determine the corresponding historical compression parameter of the same or similar historical gas features as the current compression parameter.

[0042] In some embodiments, the processor is further configured to determine a loss distribution, and generate the compression parameter based on the loss distribution and the gas feature of the raw gas.

[0043] The loss distribution refers to the distribution of pressure loss at each preset point in the compression device during the process of the raw gas entering the compression device from the outside. Different preset points can have different materials, sizes, etc. Different preset points can have different resistance values, so the raw gas can have pressure loss due to resistance after passing through the preset point.

[0044] In some embodiments, the loss distribution can be determined by the following steps:

[0045] Step S1, determining a plurality of preset points on the pipeline in the compression device.

[0046] The preset points are points selected in advance. For example, multiple points in the air inlet channel of the compression device or the compressed air pipe. The preset distance can be set between adjacent preset points. The preset points and the preset distance are preset values, which can be set according to actual needs.

[0047] In some embodiments, the preset distance can be inversely related to the pipe complexity.

[0048] The pipe complexity is used to represent the complexity of the pipe. In some embodiments, the pipe complexity can be determined based on the number of bends in the pipe. The more the number of bends, the higher the pipe complexity, and the shorter the preset distance can be.

[0049] Step S2, installing a pressure gauge at the preset points.

[0050] Step S3, applying pressure to the pipe, and obtaining the pressure values of the preset points based on the pressure gauges.

[0051] Step S4, determining the pressure loss values of the preset points based on the pressure difference between adjacent pressure gauges of the preset points.

[0052] For example, 500 Pa of pressure is applied to the inlet of the pipe, the pressure value monitored by the pressure sensor corresponding to the first preset point closest to the inlet is 499 Pa, and the pressure value monitored by the pressure sensor corresponding to the second preset point second closest to the inlet is 497 Pa. The pressure loss value of the first preset point is 1 Pa, and the pressure loss value of the second preset point is 2 Pa.

[0053] Step S5, repeating steps S3 and S4, the processor statistics multiple pressure loss values corresponding to each preset point, and calculates the average value, which can be determined by the processor as the final pressure loss value of the preset point.

[0054] In some embodiments, the number of repetitions and the pressure applied can be preset values, which can be set according to actual needs.

[0055] Step S6, constructing a set of final pressure loss values of each preset point to obtain the loss distribution.

[0056] In some embodiments, the processor can determine the compression parameters based on the loss distribution and the gas characteristics of the original gas through vector matching.

[0057] In some embodiments, the processor can construct a compression vector based on the current gas feature, the altitude, the loss distribution, the compressor type, the maximum temperature allowed by the cooler, and the target nitrogen concentration. The altitude refers to the altitude of the location where the compressor is located, which can be queried through a third-party database. The compressor type refers to the type of the compressor, for example, piston, screw, etc. The maximum temperature allowed by the cooler refers to the maximum temperature of the fluid allowed to enter the cooler. The compressor type and the maximum temperature allowed by the cooler can be obtained from the manufacturer. The target nitrogen concentration refers to the concentration of the nitrogen obtained finally. The target nitrogen concentration is a preset value, which can be set according to actual needs.

[0058] The processor can construct a first vector library based on the historical data. The first vector library includes a plurality of first feature vectors and corresponding first vector labels. The processor can construct a first feature vector based on the historical gas feature, the historical altitude, the historical loss distribution, the historical compressor type, the historical maximum temperature allowed by the cooler, and the historical target nitrogen concentration in the historical data. The first vector label includes the historical compression parameter corresponding to the first feature vector.

[0059] In some embodiments, the processor can select data in the historical data in which the purity of the obtained nitrogen reaches the target nitrogen purity requirement and the nitrogen production time is lower than the preset time, and construct a feature vector based on the historical gas feature, the historical altitude, the historical loss distribution, the historical compressor type, the historical maximum temperature allowed by the cooler, and the historical target nitrogen concentration corresponding to the data. The processor can take the pressure value monitored by the pressure sensor of the second sensor unit group as the first vector label.

[0060] In some embodiments, the processor can determine the historical compression parameter corresponding to the first feature vector as the current compression parameter by vector matching in the first vector library to match the first feature vector with the highest similarity.

[0061] In some embodiments, the processor can generate the compression parameter based on the loss distribution, the gas feature of the raw gas, and the first content distribution.

[0062] For more information about the first content distribution, see the related description of step 310.

[0063] In some embodiments, the compression vector can also include the first content distribution, and the first feature vector can also include the corresponding historical first content distribution.

[0064] Different gas molecules have different compressibilities and different loads on the compressor. By considering the composition of the raw gas and the first content distribution when determining the compression parameter, the accuracy of determining the compression parameter can be improved, and better compression effect can be achieved when the raw gas is subsequently compressed.

[0065] The pressure of the original gas gradually decreases in the process of circulation due to the pressure loss of the original gas caused by pipeline friction, so that the compression effect of the compression device on the original gas after pressurization cannot meet the expectation, and therefore, when determining the compression parameters, the compression parameters can be determined based on the actual situation in the process of circulation of the original gas by considering the loss distribution, which is beneficial to improve the accuracy of determining the compression parameters. The efficiency of determining the compression parameters can be improved by vector matching, and the accuracy of determining the compression parameters can be improved in combination with the actual completed nitrogen production data in the historical data.

[0066] In step 230, the compression device is controlled to process the original gas according to the compression parameters to obtain a first intermediate gas.

[0067] The first intermediate gas refers to the gas after the original gas is compressed by the compression device.

[0068] In step 240, the gas characteristics of the first intermediate gas are obtained from the second sensing unit group.

[0069] In step 250, the cooling parameters and the drying parameters are generated according to the gas characteristics of the first intermediate gas.

[0070] The cooling parameters refer to the working parameters of the cooling device. In some embodiments, the cooling parameters can include the temperature of the cooling medium, the temperature of the first intermediate gas after cooling, etc.

[0071] The drying parameters refer to the working parameters of the drying device. In some embodiments, the drying parameters can include the refrigerant evaporation temperature of the refrigeration type drying device, and the adsorbent regeneration period and the adsorbent regeneration temperature of the adsorption type drying device, etc.

[0072] In some embodiments, the processor can construct a second data table based on historical experience and historical production data, and the second data table includes the corresponding historical gas characteristics, historical cooling parameters, historical drying parameters and their corresponding relationships of the historical first intermediate gas. In some embodiments, the processor can determine the same or similar historical gas characteristics based on the current gas characteristics of the first intermediate gas, and determine the corresponding historical cooling parameters and historical drying parameters as the current cooling parameters and drying parameters by querying the second data table.

[0073] In step 260, the cooling device and the drying device are controlled to process the first intermediate gas according to the cooling parameters and the drying parameters to obtain a second intermediate gas.

[0074] The second intermediate gas refers to the gas used for nitrogen production after the first intermediate gas is cooled and dried.

[0075] At step 270, the gas production unit produces nitrogen based on the second intermediate gas. For more information about the gas production unit, please refer to the description of the gas production unit in Figure 1 .

[0076] In some embodiments, the gas production unit can produce nitrogen in various ways, such as deep cooling air separation, molecular sieve air separation, membrane air separation, etc.

[0077] The nitrogen production regulation system and method provided by some embodiments of the present specification can automatically detect the gas characteristics of the raw gas and automatically generate the compression parameters, automatically detect the gas characteristics of the first intermediate gas and automatically generate the cooling parameters and drying parameters, which can improve the automation degree of the pretreatment of the raw gas. And according to the gas characteristics, the compression parameters, the cooling parameters and the drying parameters can be automatically regulated, which is beneficial to improve the accuracy and efficiency of the pretreatment and provide high-quality gas in the subsequent nitrogen production.

[0078] Figure 3 is an example flowchart for generating cooling parameters and drying parameters according to some embodiments of the present specification. As Figure 3 indicated, the flow 300 includes the following steps. In some embodiments, the flow 300 can be executed by a processor.

[0079] At step 310, the first content distribution of the raw gas is determined.

[0080] The first content distribution refers to the composition of the raw gas and the proportion of each component. For example, the composition can include nitrogen, oxygen, carbon dioxide, water vapor, etc., and the first content distribution also includes the proportion of each component in the raw gas.

[0081] In some embodiments, the processor can obtain the first content distribution in various ways. For example, obtaining user input, looking up a table, etc.

[0082] For example, when the raw gas is air, the processor can construct a third data table based on historical data or a third-party database. The third data table can include geographical location, air quality, weather conditions, first content distribution and their corresponding relationships. The processor can determine the first content distribution by querying the third data table based on the current geographical location, air quality, and weather conditions. The geographical location refers to the location of collecting air, such as province, etc. The air quality can include different pm values. The weather conditions can include sunny, cloudy, rainy, etc.

[0083] Exemplarily, when the original gas is an industrial gas, the processor can construct a fourth data table based on historical data or a third-party database. The fourth data table can include industrial reactant types, raw gas compositions, process operation conditions, first content distributions, and their corresponding relationships. The processor can determine the first content distribution by querying the fourth data table based on the current industrial reactant type, the raw gas composition, and the process operation condition. The industrial reactant type refers to the type of chemical reaction in industrial production, for example, ammonia synthesis, petroleum cracking, steel smelting, etc. The raw gas composition refers to the composition of the gas raw material participating in the reaction in industrial production, for example, hydrogen, nitrogen, hydrocarbon gas, etc. The process operation condition refers to the condition for carrying out the reaction in industrial production, for example, high-pressure synthesis, high-temperature cracking, etc.

[0084] In some embodiments, the first content distribution further includes an oil content.

[0085] The oil content refers to the content of oil in the original gas. The oil content can be obtained by an oil sensor. For updates on the oil sensor, please refer to the relevant description of step 240.

[0086] By determining the oil content, adjusting the compression parameters according to the oil content during the compression of the original gas can effectively reduce the vaporization or residue of oil in the compression process, thereby reducing the pollution of oil to subsequent equipment and processes.

[0087] Step 320, generating a dew point probability distribution according to the first content distribution and the gas characteristics of the first intermediate gas.

[0088] The dew point refers to the temperature at which the gas cools to the first drop of liquid under a certain gas pressure.

[0089] The dew point probability is used to characterize the probability of the first drop of liquid of the compressed gas at different temperatures when the compressed gas is cooled to a preset temperature range.

[0090] The dew point probability distribution is used to characterize the distribution of the dew point probability in the preset temperature range. In some embodiments, the dew point probability is negatively correlated with the temperature. The dew point probability distribution can be a preset value.

[0091] The preset temperature range is a preset value, which can be set according to actual needs. The dew point probability can be obtained through historical data or experimental data.

[0092] Exemplarily, the preset temperature range is -40℃ to -70℃, and the dew point probability distribution is [(-40℃, G1), (-41℃, G2), …, (-70℃, G n )]. G1…G n is the dew point probability at the corresponding temperature.

[0093] In some embodiments, the processor can construct a dew point vector based on the first content distribution and the gas characteristic of the first intermediate gas.

[0094] In some embodiments, the processor can construct a second vector library based on historical data or experimental data. The second vector library comprises second feature vectors and corresponding second vector labels of the second feature vectors.

[0095] The second feature vectors comprise historical first content distributions of historical raw gases in the historical data or experimental data and corresponding historical gas characteristics of historical first intermediate gases of the historical raw gases, and the second vector labels are historical dew point probability distributions corresponding to the second feature vectors. In some embodiments, the processor can determine actual dew points of the historical gases when the historical first intermediate gases are cooled under certain pressure conditions. In some embodiments, when the processor constructs the historical dew point probability distributions corresponding to the historical first intermediate gases, the probability corresponding to the actual dew point is set to 1, and the probabilities corresponding to the remaining temperatures are set to 0. For example, the preset temperature range is -40°C to -70°C, the actual dew point of the historical gas is -50°C, and the second vector label is [(-40°C, 0), (-41°C, 0), …, (-50°C, 1), …, (-70°C, 0)].

[0096] In some embodiments, the processor can determine the current dew point probability distribution by matching the second feature vector with the highest similarity in the second vector library based on the dew point vector through vector matching. The processor can calculate the similarity through a preset algorithm, such as cosine distance, Euclidean distance, etc.

[0097] In some embodiments, the processor can generate the dew point probability distribution based on a dew point prediction model according to the first content distribution and the gas characteristic of the first intermediate gas.

[0098] The dew point prediction model refers to a model for determining the dew point probability distribution. In some embodiments, the dew point prediction model is a machine learning model, such as a neural network model (Neural Networks, NN), etc.

[0099] In some embodiments, the input of the dew point prediction model further comprises the oil content of the first intermediate gas.

[0100] For more information about the oil content, please refer to the relevant description of step 310.

[0101] When the oil content is high, in order to make the oil fully condense into a liquid state for removal, the temperature of the first intermediate gas needs to be reduced to a lower level, thereby requiring the use of a cooling medium with a lower temperature to improve the oil removal effect

[0102] When the dew point probability distribution is determined, the oil content is considered, and in subsequent determination of the cooling parameter based on the dew point probability distribution, the oil content is also considered, so that the determined cooling parameter can effectively improve the condensation efficiency of the oil, thereby enhancing the oil removal effect. It is beneficial to reduce the oil residue in the gas in the subsequent, improve the purity of nitrogen and the stability of equipment operation.

[0103] In some embodiments, the processor can train the dew point prediction model based on the first sample data set.

[0104] The first sample data set includes first training samples and their corresponding first labels.

[0105] In some embodiments, the first training sample includes a sample first content distribution and a sample gas feature of a sample first intermediate gas. The first label is the dew point probability distribution corresponding to each first training sample.

[0106] In some embodiments, the processor can change the first content distribution and the gas feature in advance through a series of experiments, determine the dew point corresponding to different first content distributions and gas features based on experimental accurate measurement, construct the dew point probability distribution based on the dew point, and determine the first content distribution and the gas feature collected as the first training sample, and the dew point probability distribution corresponding thereto as the first training label corresponding to the first training sample.

[0107] For example, the preset temperature range is-40℃ to-70℃, the experimentally accurately measured dew point is-50℃, and the processor can set the probability corresponding to the measured dew point to 1 by exhaustion, and the probabilities of the remaining dew points to 0, that is, the processor constructs the first training label as [(-40℃, 0), (-41℃, 0), …, (-50℃, 1), …, (-70℃, 0)]. The preset temperature range can be set based on artificial experience.

[0108] In some embodiments, the processor can perform multiple rounds of iterations, and at least one round of iteration includes: selecting one or more first training samples from the first sample data set, inputting the one or more first training samples into the initial dew point prediction model to obtain the model prediction output corresponding to the one or more first training samples; according to the model prediction output corresponding to the one or more first training samples and the first label of the one or more first training samples, substituting into the formula of the pre-defined loss function to calculate the value of the loss function; according to the value of the loss function, reversely updating the model parameters in the initial dew point prediction model; this step can be performed using various methods. For example, the update can be performed based on the gradient descent method. When the iteration end condition is met, the iteration is ended, and the trained dew point prediction model is obtained.

[0109] The dew point probability distribution is determined by the dew point prediction model, which can improve the accuracy and efficiency of determining the dew point probability distribution. The dew point prediction model is trained by using experimental data, which can improve the accuracy of the dew point prediction model.

[0110] At step 330, the cooling parameters and the drying parameters are generated based on the first content distribution and the dew point probability distribution.

[0111] In some embodiments, the processor can generate the cooling parameters and the drying parameters by the parameter prediction model based on the first content distribution and the dew point probability distribution.

[0112] The parameter prediction model refers to a model for determining the cooling parameters and the drying parameters.

[0113] In some embodiments, the parameter prediction model can include a cooling parameter determination layer, a drying mode determination layer, and a drying parameter determination layer.

[0114] The cooling parameter determination layer refers to a model for determining the cooling parameters. In some embodiments, the cooling parameter determination layer is a machine learning model, such as a neural network model (NN). For more information about the cooling parameters, please refer to the relevant description of step 250.

[0115] The input of the cooling parameter determination layer includes the first content distribution, the sample gas characteristics of the first intermediate gas, and the sample dew point probability distribution, and the output includes the cooling parameters.

[0116] In some embodiments, the processor can train the cooling parameter determination layer based on the second sample data set.

[0117] The second sample data set includes second training samples and their corresponding second labels.

[0118] In some embodiments, the second training samples include sample first content distributions of sample raw gas, sample gas characteristics of sample first intermediate gas corresponding to the sample raw gas, and sample dew point probability distributions. The second labels are cooling parameters corresponding to the sample first intermediate gas.

[0119] In some embodiments, the processor can set different cooling parameters through pre-experiments. After the second intermediate gas cooled by different cooling parameters is actually used to produce nitrogen, the corresponding nitrogen purity and nitrogen production time are determined. The historical data that the nitrogen purity meets the target nitrogen purity requirement and the nitrogen production time is lower than the preset time length is selected, and the first content distribution, the dew point probability distribution, and the gas characteristics corresponding to the historical data are used as the second training samples. The cooling parameters corresponding to the historical data are determined as the second labels.

[0120] The drying mode determination layer refers to a model for determining the drying mode. In some embodiments, the drying mode determination layer is a machine learning model. For example, a neural network model (NN), etc.

[0121] The input of the drying mode determination layer includes the first content distribution, the sample gas feature of the first intermediate gas, the sample dew point probability distribution, and the cooling parameter, and the output includes the drying mode.

[0122] In some embodiments, the processor can train the drying mode determination layer based on a third sample data set.

[0123] The third sample data set includes a third training sample and a corresponding third label.

[0124] In some embodiments, the third training sample includes a sample first content distribution of a sample raw gas, a sample gas feature of a sample first intermediate gas, a sample dew point probability distribution, and a sample cooling parameter. The third label is the drying mode corresponding to the sample first intermediate gas.

[0125] In some embodiments, the processor can set different drying modes through pre-experiments. After the second intermediate gas dried by different drying modes is actually used to produce nitrogen, the corresponding nitrogen purity and nitrogen production time are determined. Select the historical data whose nitrogen purity reaches the target nitrogen purity requirement and whose nitrogen production time is lower than the preset time length, and use the first content distribution, dew point probability distribution, gas feature, and cooling parameter corresponding to the historical data as the third training sample. The drying mode corresponding thereto is determined as the third label.

[0126] The drying parameter determination layer refers to a model for determining the drying parameter. In some embodiments, the drying parameter determination layer is a machine learning model. For example, a neural network model (NN), etc. For more information about the cooling parameter, please refer to the relevant description of step 250.

[0127] The input of the drying parameter determination layer includes the first content distribution, the sample gas feature of the first intermediate gas, the sample dew point probability distribution, the cooling parameter, and the drying mode, and the output includes the drying parameter.

[0128] In some embodiments, the processor can train the drying parameter determination layer based on a fourth sample data set.

[0129] The fourth sample data set includes a fourth training sample and a corresponding fourth label.

[0130] In some embodiments, the fourth training sample includes the sample first content distribution of the sample raw gas, the sample gas feature of the sample first intermediate gas, the sample dew point probability distribution, the sample cooling parameter, and the sample drying mode. The fourth label is the drying parameter corresponding to the sample first intermediate gas.

[0131] In some embodiments, the processor can set different drying parameters through pre-experiments. After the second intermediate gas dried by different drying parameters is actually used to produce nitrogen, the corresponding nitrogen purity and nitrogen production time are determined. The historical data in which the nitrogen purity reaches the target nitrogen purity requirement and the nitrogen production time is lower than the preset time length are selected, and the first content distribution, the dew point probability distribution, the gas feature, the cooling parameter, and the drying mode corresponding to the historical data are taken as the fourth training sample. The drying parameter corresponding thereto is determined as the fourth label.

[0132] In some embodiments, the cooling parameter determination layer, the drying mode determination layer, and the drying parameter determination layer can be trained separately or jointly. The training process of the cooling parameter determination layer, the drying mode determination layer, and the drying parameter determination layer is similar to the training process of the dew point prediction model. For the training process of the cooling parameter determination layer, the drying mode determination layer, and the drying parameter determination layer, refer to the related description of the training process of the dew point prediction model.

[0133] In some embodiments, the output of the cooling parameter determination layer can be taken as the input of the drying mode determination layer, and the output of the drying mode determination layer can be taken as the input of the drying parameter determination layer.

[0134] In some embodiments, the training sample for joint training includes the sample first content distribution of the sample raw gas, the sample gas feature of the sample first intermediate gas, and the sample dew point probability distribution, and the label is the corresponding cooling parameter of the sample first intermediate gas of the training sample. The processor can input the sample first content distribution, the sample gas feature of the sample first intermediate gas, and the sample dew point probability distribution into the initial cooling parameter determination layer to obtain the cooling parameter output by the cooling parameter determination layer. The cooling parameter is taken as a training sample, and is input into the initial drying mode determination layer together with the sample first content distribution, the sample gas feature of the sample first intermediate gas, and the sample dew point probability distribution to obtain the drying mode output by the drying mode determination layer. The drying mode is taken as a training sample, and is input into the initial drying mode determination layer together with the sample first content distribution, the sample gas feature of the sample first intermediate gas, the sample dew point probability distribution, and the cooling parameter to obtain the drying parameter output by the drying parameter determination layer. A loss function is constructed based on the drying parameter and the label corresponding to the training sample, and the model parameters in the initial cooling parameter determination layer, the drying mode determination layer, and the drying parameter determination layer are updated in reverse according to the value of the loss function to obtain the trained cooling parameter determination layer, the drying mode determination layer, and the drying parameter determination layer.

[0135] In actual production, the humidity, temperature, pressure and other parameters of the original gas may dynamically change due to environmental changes or fluctuations in raw materials. The processor can predict the dew point of the first intermediate gas when the first intermediate gas is cooled, and obtain the dew point probability distribution. When determining the cooling parameters and drying parameters subsequently, more robust cooling parameters and drying parameters can be obtained, improving the response capability of the nitrogen production control system to fluctuations in the composition and gas characteristics of the original gas, thereby improving the application range of the nitrogen production control system. At the same time, the dew point probability distribution can quantify the risk of different dew points, for example, different oil contents in the original gas may result in different dew points of the first intermediate gas. By considering the dew point probability distribution, the cooling parameters and drying parameters can be redundantly designed to improve the ability of the designed cooling parameters and drying parameters to cope with risks, thereby effectively reducing the impact of subsequent nitrogen production on process stability and ensuring the quality of nitrogen production.

[0136] Figure 4 is an example flowchart of controlling the operation of a recovery unit according to some embodiments of the present specification. As shown in Figure 4 , the flow 400 includes the following steps. In some embodiments, the flow 400 can be executed by a processor.

[0137] Step 410, obtaining local component features of the separated gas from the third sensing unit group.

[0138] The separated gas refers to the remaining gas separated after the second intermediate gas is subjected to nitrogen production by the gas production unit. In some embodiments, the separated gas can include methane (CH4), hydrogen (H2), carbon monoxide (CO), hydrogen sulfide (H2S), carbon dioxide (CO2), and helium (He), etc.

[0139] For more information about the second intermediate gas, please refer to the related description of Figure 2 .

[0140] The local component features refer to quantitative or qualitative information of part of the gas components in the separated gas. In some embodiments, the local component features can include gas categories, gas proportions, etc.

[0141] For example only, the local component features of combustible gases such as methane (CH4) and hydrogen (H2) can be obtained based on semiconductor gas sensors.

[0142] For example only, the local component features of gases such as carbon monoxide (CO), hydrogen sulfide (H2S), oxygen (O2), and nitrogen dioxide (NO2) can be obtained based on electrochemical sensors.

[0143] For example only, the local component features of gases such as hydrogen (H2), helium (He), and carbon dioxide (CO2) can be obtained based on thermal conductivity sensors.

[0144] At step 420, a second content distribution of the separated gas is predicted based on the local component features and the first content distribution of the original gas.

[0145] For more information about the original gas and the first content distribution, please refer to the relevant description of Figures 2-3 .

[0146] The second content distribution refers to the content distribution of the gas components in the separated gas.

[0147] In some embodiments, the processor can construct a demand feature vector based on the local component features of the separated gas and the first content distribution of the original gas, and predict the second content distribution of the separated gas through the third vector database based on the demand feature vector.

[0148] In some embodiments, the third vector database includes multiple groups of reference demand feature vectors and their corresponding reference second content distributions. For example, the third vector database can construct a reference demand feature vector based on the first content distribution of the original gas and the corresponding local component features of the separated gas in the historical production data, and take the actual second content distribution corresponding to the local component features of the separated gas and the first content distribution of the original gas as the reference second content distribution corresponding to the reference demand feature vector.

[0149] In some embodiments, the processor can be annotated by a human based on part of the local component features of the separated gas and the first content distribution of the original gas. The processor can also obtain the first content distribution of the original gas based on a preset table, obtain the separated gas based on a nitrogen production operation, determine the local component features of the separated gas based on a semiconductor gas sensor, an electrochemical sensor, a thermal conductivity sensor, etc., and construct the local component features and the first content distribution as a reference demand feature vector. The reference demand feature vectors constructed based on the local component features with artificial annotation and the first content distribution are clustered, and for each cluster obtained by clustering, the second content distribution with the highest proportion is taken as the reference second content distribution corresponding to all reference demand feature vectors in the cluster. For more information about the preset table, please refer to the relevant description of Figure 3 .

[0150] In some embodiments, the processor can construct a target vector based on the current first content distribution and the local component features, select a reference demand feature vector that meets the preset requirements, and take the reference second content distribution corresponding to the reference demand feature vector as the second content distribution. The preset requirements can include that the vector similarity between the reference demand feature vector and the target vector is higher than a preset similarity threshold, etc., wherein the vector similarity can be negatively correlated with the vector distance between the target vector and the reference demand feature vector, and the vector distance can be determined based on the cosine distance, the Euclidean distance, etc. The preset similarity threshold is set according to experience.

[0151] In some embodiments of the present disclosure, since the cost of manual annotation is high, by manually annotating part of the data and then annotating another part of the data based on clustering, the cost of manual annotation can be reduced, the vector database can be enriched, and the prediction accuracy of the second content distribution can be improved.

[0152] In some embodiments, the processor can further predict the second content distribution of the separated gas based on the local component feature and the first content distribution through a content prediction model.

[0153] The content prediction model refers to a model for predicting the second content distribution of the separated gas. In some embodiments, the content prediction model can be a machine learning model, for example, a recurrent neural network (RNN).

[0154] In some embodiments, the input of the content prediction model can include the local component feature and the first content distribution. In some embodiments, the output of the content prediction model can be the second content distribution.

[0155] In some embodiments, the content prediction model can be trained through a plurality of fifth training samples with fifth labels. Each group of fifth training samples includes sample historical local component features of sample separated gas and sample historical first content distribution of sample raw gas, the sample separated gas is the remaining gas after the sample raw gas is separated by nitrogen production, and the corresponding fifth label is the sample historical second content distribution of the sample separated gas.

[0156] The fifth training sample can be obtained based on experiments. In some embodiments, the sample historical first content distribution of the sample raw gas and the sample separated gas can be obtained in advance based on a series of experiments, and the sample historical local component features in the sample separated gas can be determined by using a semiconductor gas sensor, an electrochemical sensor, and a thermal conductivity sensor.

[0157] For more information about the sensor determining the sample historical local component features in the sample separated gas, please refer to the related description above. In some embodiments, the processor can construct the fifth training sample based on the above sensor data integration.

[0158] In some embodiments, the fifth label can be based on the detection of the sample separated gas by the gas chromatography sensor, and the gas components in the obtained sample separated gas and the content distribution of different gases can be taken as the fifth label of the fifth training sample.

[0159] For more information about the training process of the content prediction model, please refer to the related description of the training process of the dew point prediction model in Figure 3

[0160] ​In some embodiments, the input of the content prediction model can further include cooling parameters and drying parameters. More information about the cooling parameters and drying parameters can be found in the related description of Figure 2

[0161] In some embodiments of the present specification, the cooling parameters and drying parameters can be combined with the first content distribution to infer the second content distribution through chemical reactions, etc., so as to improve the prediction accuracy of the second content distribution.

[0162] At step 430, it is determined whether the separated gas needs to be recovered according to the second content distribution.

[0163] In some embodiments, the processor can determine that the separated gas needs to be recovered in response to the content of a specific gas (e.g., oxygen) in the second content distribution being not less than a preset threshold. The preset threshold can be set based on experiments. For example, the processor can determine that the separated gas needs to be recovered in response to the content of oxygen in the second content distribution being not less than 30%.

[0164] At step 440, the recovery unit is controlled to operate in response to the need to recover the separated gas.

[0165] In some embodiments, the processor can control the recovery unit to operate and recover the separated gas in response to the need to recover the separated gas.

[0166] More information about the recovery unit can be found in the related description of Figure 1

[0167] In some embodiments of the present specification, the separated gas can be recovered by the recovery unit and the third sensing unit group, so as to realize sustainable use of nitrogen production resources and improve the efficiency of nitrogen production.

[0168] The above has described the basic concepts. Obviously, for those skilled in the art, the above detailed disclosure is only used as an example, and does not constitute a limitation on the present specification. Although it is not explicitly stated here, those skilled in the art can make various modifications, improvements and corrections to the present specification. Such modifications, improvements and corrections are suggested in the present specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of the present specification.

[0169] Finally, it should be understood that the embodiments described in the present specification are only used to illustrate the principles of the embodiments of the present specification. Other variations can also belong to the scope of the present specification. Therefore, as an example but not limitation, alternative configurations of the embodiments of the present specification can be considered consistent with the teachings of the present specification. Accordingly, the embodiments of the present specification are not limited to the embodiments explicitly introduced and described in the present specification.​​

Claims

1. A nitrogen generation and control system, comprising an intake unit, a gas generation unit, a recovery unit, a third sensing unit group, and a processor. The intake unit includes an intake channel, a first sensing unit group, a second sensing unit group, a compression device, a cooling device, and a drying device; The third sensing unit group is configured to collect local component characteristics of the separated gas, and the recovery unit is configured to recover the separated gas; The processor is configured to: The gas characteristics of the original gas are obtained from the first sensing unit group; Based on the gas characteristics of the original gas, compression parameters are generated; Based on the compression parameters, the compression equipment is controlled to process the raw gas to obtain a first intermediate gas; The gas characteristics of the first intermediate gas are obtained from the second sensing unit group; Determine the first content distribution of the original gas; Based on the first content distribution and the gas characteristics of the first intermediate gas, a dew point probability distribution is generated; Based on the first content distribution and the dew point probability distribution, cooling parameters and drying parameters are generated; Based on the cooling parameters and the drying parameters, the cooling equipment and the drying equipment are controlled to process the first intermediate gas to obtain the second intermediate gas; The gas generation unit is configured to generate nitrogen based on the second intermediate gas; The processor is further configured to: Based on the local component characteristics, the first content distribution, the cooling parameters, and the drying parameters, the second content distribution of the separated gas is predicted by a content prediction model, which is a machine learning model. Based on the second content distribution, determine whether the separated gas needs to be recovered; In response to the need to recover the separated gas, the recovery unit is controlled to operate.

2. The system according to claim 1, characterized in that, The processor is further configured to: Determine the loss distribution; The compression parameters are generated based on the loss distribution and the gas characteristics of the original gas.

3. A nitrogen production control method, characterized in that, The method includes: The gas characteristics of the original gas are obtained from the first sensing unit group; Based on the gas characteristics of the original gas, compression parameters are generated; Based on the compression parameters, the compression equipment is controlled to process the raw gas to obtain a first intermediate gas; The gas characteristics of the first intermediate gas are obtained from the second sensing unit group; Determine the first content distribution of the original gas; Based on the first content distribution and the gas characteristics of the first intermediate gas, a dew point probability distribution is generated; Based on the first content distribution and the dew point probability distribution, cooling parameters and drying parameters are generated; Based on the cooling parameters and the drying parameters, the cooling equipment and the drying equipment are controlled to process the first intermediate gas to obtain the second intermediate gas; The controlled gas generation unit generates nitrogen based on the second intermediate gas; The processor is further configured as follows: Local component characteristics of the separated gas are obtained from the third sensing unit group; Based on the local component characteristics, the first content distribution, the cooling parameters, and the drying parameters, the second content distribution of the separated gas is predicted by a content prediction model, which is a machine learning model. Based on the second content distribution, determine whether the separated gas needs to be recovered; In response to the need to recover the separated gas, the recovery unit is controlled to operate.

4. The method according to claim 3, characterized in that, The step of generating compression parameters based on the gas characteristics of the original gas includes: Determine the loss distribution; The compression parameters are generated based on the loss distribution and the gas characteristics of the original gas.

5. A nitrogen generation control device, comprising a processor, the processor being used to execute the nitrogen generation control method according to any one of claims 3 to 4.

6. A computer-readable storage medium storing computer instructions, wherein when a computer reads the computer instructions in the storage medium, the computer executes the nitrogen production control method as described in any one of claims 3 to 4.

Images