Fully automatic online atmospheric pre-concentration method, device, storage medium and electronic equipment

Abstract

The present disclosure relates to a fully automatic online atmospheric pre-concentration method, device, storage medium and electronic device. The method comprises: obtaining the thermal infrared distribution images of the cold trap tube collected by the infrared sensor at multiple predetermined time points within a predetermined time period to obtain a set of thermal infrared distribution images; obtaining the time series of pressure values ​​of the cold trap tube collected by the pressure sensor; performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​respectively to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors; and determining the control decision result based on the deep cross-modal constraint fusion information between the pressure time series feature vectors and the sequence of thermal infrared distribution state feature vectors. In this way, it is possible to intelligently determine when to stop heating, thereby improving the accuracy of the analysis results.

Description

Technical Field

[0001] The present disclosure relates to the technical field of atmospheric pre-concentration, and in particular to a fully automatic online atmospheric pre-concentration method, device, storage medium and electronic equipment. Background Art

[0002] Atmospheric pre-concentration is a commonly used technique for analyzing trace components in atmospheric samples. Due to the low concentration of trace components in the atmosphere, direct analysis often fails to achieve the desired detection effect. Therefore, it is necessary to enrich the trace components through pre-concentration to improve their detection sensitivity in the analyzer.

[0003] In the traditional atmospheric pre-concentration method, the equipment is large in size, the maintenance workload is large, the replacement of refrigerant is troublesome, and liquid nitrogen cannot be realized online. In this regard, patent CN107328637A discloses an online atmospheric pre-concentration device, which uses electronic refrigeration to enrich the atmosphere and record the content traces. However, in the actual working process of the online atmospheric pre-concentration device, many complex operations still need to be performed manually, such as desorption. This manual processing method is easily affected by the operator's technical level, resulting in inaccurate subsequent analysis results. In addition, during the manual operation process, the sample may be affected by the environment, resulting in sample loss or contamination, affecting the analysis results.

[0004] Therefore, an optimized fully automatic online atmospheric pre-concentration method, device, storage medium and electronic equipment are desired. Summary of the invention

[0005] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0006] In a first aspect, the present disclosure provides a fully automatic online atmospheric pre-concentration method, which comprises: a first step of pre-purging, completing cold trap preparation and purging residual substances in the cold trap; a second step of enrichment, enriching the gas in a cold trap tube of the cold trap; a third step of desorption, rapidly heating the cold trap tube; a fourth step of sample injection, switching a six-way valve, eluting the gas out of the cold trap tube, injecting it into an analyzer, and connecting a detector through the analytical column; a fifth step of back-blowing, reversely switching a conversion valve, back-blowing the non-desorbed high-boiling-point gas out of the cold trap tube, and entering the next cycle;

[0007] Characterized in that the third step comprises:

[0008] Acquire thermal infrared distribution images of the cold trap tube at a plurality of predetermined time points within a predetermined time period collected by an infrared sensor to obtain a set of thermal infrared distribution images;

[0009] Acquire a time series of pressure values ​​of the cold trap tube collected by a pressure sensor;

[0010] Performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​respectively to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors;

[0011] A control decision result is determined based on deep cross-modal constraint fusion information between the sequences of the pressure time series feature vector and the thermal infrared distribution state feature vector.

[0012] Optionally, feature extraction is performed on the set of thermal infrared distribution images and the time series of pressure values ​​respectively to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors, including: performing time series analysis on the time series of pressure values ​​to obtain the pressure time series feature vector; performing thermal infrared distribution state feature extraction on the set of thermal infrared distribution images to obtain a sequence of thermal infrared distribution state feature vectors.

[0013] Optionally, a time series analysis is performed on the time series of the pressure values ​​to obtain the pressure time series feature vector, including: arranging the time series of the pressure values ​​into a pressure time series input vector according to the time dimension; and passing the pressure time series input vector through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain the pressure time series feature vector.

[0014] Optionally, the pressure time series input vector is passed through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain the pressure time series feature vector, comprising: processing the pressure time series input vector with the following one-dimensional extended convolution formula to obtain the pressure time series feature vector; wherein the one-dimensional extended convolution formula is: ; Among them, the one-dimensional expanded convolution kernel The size is , is the length of the original convolution kernel, is the expansion rate, Indicates that the pressure time series input vector The length of the first eigenvalue at each position is time window, is the bias term, and , is a nonlinear activation function, It is local convolutional coding feature vector, is the dimension of the pressure time series input vector, is the pressure time series characteristic vector, Indicates cascade.

[0015] Optionally, thermal infrared distribution state feature extraction is performed on the set of thermal infrared distribution images to obtain a sequence of thermal infrared distribution state feature vectors, including: passing the set of thermal infrared distribution images through a thermal infrared distribution state feature extractor based on a convolutional neural network model to obtain a sequence of thermal infrared distribution state feature vectors.

[0016] Optionally, a control decision result is determined based on the deep cross-modal constraint fusion information between the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector, including: passing the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector; performing feature distribution correction on the pressure-heat distribution deep constraint interactive fusion feature vector to obtain a corrected pressure-heat distribution deep constraint interactive fusion feature vector; passing the corrected pressure-heat distribution deep constraint interactive fusion feature vector through a classifier-based desorption control decider to obtain the control decision result, and the control decision result is used to indicate whether to stop heating.

[0017] Optionally, the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector, including: processing the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector with the following deep cross-modal constraint fusion formula to obtain the pressure-heat distribution deep constraint interactive fusion feature vector; wherein the deep cross-modal constraint fusion formula is: ;in, is a constant bias term, It is A bias vector, It is A bias matrix, is the first in the sequence of the thermal infrared distribution state feature vector Thermal infrared distribution state feature vector, is the pressure time series characteristic vector, represents the transpose of a vector, is the first in the pressure-heat distribution depth constraint interactive fusion feature vector feature values.

[0018] In a second aspect, the present disclosure provides a fully automatic online atmospheric pre-concentration device, the device comprising:

[0019] A pre-purge module, used to complete cold trap preparation and purge residual substances in the cold trap;

[0020] An enrichment module, used for enriching the gas in a cold trap tube of the cold trap;

[0021] A desorption module, used for rapidly heating the cold trap tube;

[0022] A sample injection module, used for switching the six-way valve, eluting the gas out of the cold trap tube, injecting it into the analyzer, and connecting it to the detector through the analytical column;

[0023] A backflush module, used for reversely switching the conversion valve to backflush the undesorbed high-boiling-point gas out of the cold trap tube to enter the next cycle;

[0024] Characterized in that the desorption module comprises:

[0025] A thermal infrared distribution image acquisition unit, used to acquire thermal infrared distribution images of the cold trap tube at a plurality of predetermined time points within a predetermined time period collected by an infrared sensor to obtain a set of thermal infrared distribution images;

[0026] A pressure value acquisition unit, used to acquire a time series of pressure values ​​of the cold trap tube collected by a pressure sensor;

[0027] A feature extraction unit, used for performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors;

[0028] A deep cross-modal constraint fusion unit is used to determine a control decision result based on deep cross-modal constraint fusion information between the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector.

[0029] In a third aspect, the present disclosure provides a computer-readable storage medium having computer program instructions stored thereon, which, when executed in a computer, causes the computer to execute the steps of the fully automatic online atmospheric pre-concentration method.

[0030] In a fourth aspect, an electronic device is characterized by comprising:

[0031] processor;

[0032] A memory, in which computer program instructions are stored, and when the computer program instructions are executed by the processor, the processor executes the steps of the fully automatic online atmospheric pre-concentration method.

[0033] The above technical solution is adopted to obtain a set of thermal infrared distribution images by acquiring the thermal infrared distribution images of the cold trap tube collected by the infrared sensor at multiple predetermined time points within a predetermined time period; acquiring the time series of the pressure values ​​of the cold trap tube collected by the pressure sensor; performing feature extraction on the set of thermal infrared distribution images and the time series of the pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors; and determining the control decision result based on the deep cross-modal constraint fusion information between the pressure time series feature vectors and the sequence of thermal infrared distribution state feature vectors. In this way, it is possible to intelligently determine when to stop heating, thereby improving the accuracy of the analysis results.

[0034] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale. In the drawings:

[0036] Figure 1 The present invention is a flow chart showing a fully automatic online atmospheric pre-concentration method according to an exemplary embodiment.

[0037] Figure 2 The figure is a block diagram of a fully automatic online atmospheric pre-concentration device according to an exemplary embodiment.

[0038] Figure 3 It is a block diagram of an electronic device according to an exemplary embodiment.

[0039] Figure 4 It is an application scenario diagram of a fully automatic online atmospheric pre-concentration method according to an exemplary embodiment. DETAILED DESCRIPTION

[0040] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0041] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0042] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0043] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0044] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0045] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0046] In order to solve the above problems, the present disclosure provides a fully automatic online atmospheric pre-concentration method, device, storage medium and electronic device, which obtains a set of thermal infrared distribution images by acquiring the thermal infrared distribution images of the cold trap tube at multiple predetermined time points within a predetermined time period collected by an infrared sensor; obtains the time series of pressure values ​​of the cold trap tube collected by a pressure sensor; performs feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors; and determines the control decision result based on the deep cross-modal constraint fusion information between the pressure time series feature vectors and the sequence of thermal infrared distribution state feature vectors. In this way, it is possible to intelligently determine when to stop heating, thereby improving the accuracy of the analysis results.

[0047] The specific embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0048] Figure 1 is a flow chart of a fully automatic online atmospheric pre-concentration method according to an exemplary embodiment. Figure 1 As shown, the method includes:

[0049] Step 101, the first step of pre-purging, completing the cold trap preparation, and purging the residual substances in the cold trap;

[0050] Step 102, the second step of enrichment, enriching the gas in the cold trap tube of the cold trap;

[0051] Step 103, the third step of desorption, rapidly heating the cold trap tube;

[0052] Step 104, the fourth step of sample injection, switching the six-way valve, eluting the gas out of the cold trap tube, injecting it into the analyzer, and connecting it to the detector through the analytical column;

[0053] Step 105, the fifth step of back-blowing, reverse switching the conversion valve, back-blowing the undesorbed high-boiling-point gas out of the cold trap tube, and entering the next cycle;

[0054] Characterized in that step 103, the third step, comprises:

[0055] Step 1031, acquiring thermal infrared distribution images of the cold trap tube at a plurality of predetermined time points within a predetermined time period collected by an infrared sensor to obtain a set of thermal infrared distribution images;

[0056] Step 1032, obtaining a time series of pressure values ​​of the cold trap tube collected by a pressure sensor;

[0057] Step 1033, performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors;

[0058] Step 1034 : Determine a control decision result based on the deep cross-modal constraint fusion information between the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector.

[0059] Among them, desorption is the reverse process of adsorption. It refers to the operation process in which the adsorbed components are precipitated from the saturated adsorbent to regenerate the adsorbent. Simply put, it is the process in which the substance adsorbed on the interface escapes from the interface and re-enters the bulk phase under certain conditions (such as temperature, pressure, etc.), which is also called desorption. In the above-mentioned atmospheric pre-concentration method, the third step of desorption is to quickly heat up the cold trap tube so that the adsorbed gas can be released from the cold trap tube for subsequent analysis. It should be understood that this process is endothermic, that is, an energy absorption process. Possible problems in this process may be: excessive heating may cause damage to the cold trap tube, and may even cause inaccurate analysis results; insufficient heating time will affect the operating efficiency of the pre-concentration system. The manual control and processing method is easily affected by the operator's technical level, resulting in inaccurate judgment of the time point when heating stops, affecting the accuracy of the analysis results. Therefore, an optimized solution is expected.

[0060] In response to the above technical problems, the technical concept of the present application is to comprehensively utilize infrared sensors and pressure sensors to collect thermal infrared distribution image data and pressure data of the cold trap tube, and use the deep cross-modal constraint interactive fusion features between the two to characterize the desorption process and desorption effect, so as to intelligently judge when to stop heating based on this deep cross-modal constraint interactive fusion features.

[0061] Based on this, in the technical solution of the present application, the third step of desorption, the specific step of rapidly heating the cold trap tube, includes: first, obtaining the thermal infrared distribution images of the cold trap tube collected by the infrared sensor at multiple predetermined time points within a predetermined time period to obtain a set of thermal infrared distribution images; and obtaining the time series of the pressure values ​​of the cold trap tube collected by the pressure sensor. Among them, the thermal infrared sensor can monitor the temperature distribution on the surface of the cold trap tube in real time. During the desorption process, the release of gas will cause temperature changes, and the thermal infrared distribution image collected by the infrared sensor can reflect the temperature changes during the desorption process. That is to say, during the desorption process, the cold trap tube will be rapidly heated, which will cause the adsorbed gas to be released from the cold trap tube. This process can be intuitively observed and monitored through the thermal infrared distribution image collected by the infrared sensor. If the temperature of the cold trap tube is too high, it may cause damage to the cold trap tube, cause gas leakage, affect the desorption process, and may affect the accuracy of the analysis results. In the technical solution of the present application, by analyzing the thermal infrared distribution image, the temperature change of the cold trap tube can be timely understood as one of the bases for judging whether the desorption is going smoothly. The pressure sensor can monitor the pressure change in the cold trap tube in real time. During the desorption process, as the gas is released from the adsorbent, the pressure changes. By collecting the time series of pressure values, the pressure changes during the desorption process can be monitored. In other words, the time series of pressure values ​​can reflect the desorption effect to a certain extent, such as the speed and completeness of gas release. By analyzing the changes in pressure values, the effect of the desorption process can be evaluated.

[0062] In one embodiment of the present disclosure, feature extraction is performed on the set of thermal infrared distribution images and the time series of pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors, including: performing time series analysis on the time series of pressure values ​​to obtain the pressure time series feature vector; performing thermal infrared distribution state feature extraction on the set of thermal infrared distribution images to obtain a sequence of thermal infrared distribution state feature vectors.

[0063] Then, the time series of the pressure values ​​are arranged according to the time dimension as a pressure time series input vector; and the pressure time series input vector is passed through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain a pressure time series feature vector. In this way, the discretely distributed pressure data can be converted into a structured vector representation by arranging according to the time dimension, and the time series of the data is retained, so that the subsequent one-dimensional extended convolution processing process can capture the time correlation and dynamic changes of the pressure data. Among them, the one-dimensional extended convolutional neural network is a neural network structure for processing one-dimensional data, similar to the one-dimensional convolutional neural network. Compared with the one-dimensional convolutional neural network, the one-dimensional extended convolutional neural network introduces a hyperparameter called the expansion rate to control the number of null values ​​0 in the convolution kernel. The expansion rate defines the size of the receptive field when the convolution kernel processes the input data. In this way, during the application process, the one-dimensional extended convolution can adjust the characteristic receptive field of the pressure data according to actual needs, so as to better capture the pressure implicit time series change characteristics and distribution pattern of the pressure time series input vector.

[0064] In one embodiment of the present disclosure, a time series analysis is performed on the time series of the pressure values ​​to obtain the pressure time series feature vector, including: arranging the time series of the pressure values ​​into a pressure time series input vector according to the time dimension; and passing the pressure time series input vector through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain the pressure time series feature vector.

[0065] Further, the pressure time series input vector is passed through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain the pressure time series feature vector, including: processing the pressure time series input vector with the following one-dimensional extended convolution formula to obtain the pressure time series feature vector; wherein the one-dimensional extended convolution formula is: ; Among them, the one-dimensional expanded convolution kernel The size is , is the length of the original convolution kernel, is the expansion rate, Indicates that the pressure time series input vector The length of the first feature value at the position is time window, is a bias term, and , is a nonlinear activation function, It is local convolutional coding feature vector, is the dimension of the pressure time series input vector, is the pressure time series characteristic vector, Indicates cascade.

[0066] At the same time, the set of thermal infrared distribution images is passed through a thermal infrared distribution state feature extractor based on a convolutional neural network model to obtain a sequence of thermal infrared distribution state feature vectors. That is, the excellent performance of the convolutional neural network model in processing images is used to capture the thermal infrared distribution pattern and implicit spatial association of the thermal infrared distribution image.

[0067] In one embodiment of the present disclosure, thermal infrared distribution state feature extraction is performed on the set of thermal infrared distribution images to obtain a sequence of thermal infrared distribution state feature vectors, including: passing the set of thermal infrared distribution images through a thermal infrared distribution state feature extractor based on a convolutional neural network model to obtain a sequence of thermal infrared distribution state feature vectors.

[0068] Then, the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector. It should be understood that during the desorption process, there are important and complex correlations and interactions between the temperature and pressure data inside the cold trap tube. For example, an increase in temperature will increase the thermal motion of the gas molecules in the tube, thereby increasing the collision frequency of the gas molecules, causing the pressure to increase accordingly. However, there is not a simple linear relationship between temperature and pressure. For example, a small change in temperature may cause a drastic change in pressure, or there may be some threshold effects, which make the relationship between temperature and pressure present nonlinear characteristics. In the technical solution of the present application, it is expected that the implicit relationship between the two will be characterized and described through the deep cross-modal constraint fusion network. Moreover, the characterization of this implicit relationship is helpful for the comprehensive description of the desorption process and desorption effect.

[0069] In one embodiment of the present disclosure, a control decision result is determined based on the deep cross-modal constraint fusion information between the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector, including: passing the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector; performing feature distribution correction on the pressure-heat distribution deep constraint interactive fusion feature vector to obtain a corrected pressure-heat distribution deep constraint interactive fusion feature vector; passing the corrected pressure-heat distribution deep constraint interactive fusion feature vector through a classifier-based desorption control decider to obtain the control decision result, and the control decision result is used to indicate whether to stop heating.

[0070] Furthermore, the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector, including: processing the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector with the following deep cross-modal constraint fusion formula to obtain the pressure-heat distribution deep constraint interactive fusion feature vector; wherein the deep cross-modal constraint fusion formula is: ;in, is a constant bias term, It is A bias vector, It is A bias matrix, is the first in the sequence of the thermal infrared distribution state feature vector Thermal infrared distribution state feature vector, is the pressure time series characteristic vector, represents the transpose of a vector, is the first in the pressure-heat distribution depth constraint interactive fusion feature vector feature values.

[0071] In the technical solution of the present application, the pressure time series feature vector expresses the time series correlation pattern characteristics of the pressure value, and the sequence of the thermal infrared distribution state feature vector expresses the image semantic characteristics of each thermal infrared distribution image in the set of thermal infrared distribution images distributed along the time series direction. Therefore, considering the cross-modal differences in source data of the sequences of the pressure time series feature vector and the thermal infrared distribution state feature vector and the differences in feature expression methods in the time series direction, when the sequences of the pressure time series feature vector and the thermal infrared distribution state feature vector are passed through a deep cross-modal constraint fusion network, it is expected to improve the mapping effect of the sequences of the pressure time series feature vector and the thermal infrared distribution state feature vector to the feature distribution domain after feature interaction response association fusion.

[0072] Therefore, the applicant of the present application further optimizes and fuses the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector, which is specifically expressed as follows: the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector are optimized and fused using the following optimization formula to obtain a correction feature vector; wherein the optimization formula is: ;in, is the pressure time series characteristic vector, is the cascade feature vector obtained by cascading the sequence of the thermal infrared distribution state feature vectors, and the pressure time series feature vector Linear interpolation is performed to make it have the same concatenation characteristic vector as the Same length, and The eigenvectors are The mean and standard deviation of the corresponding feature set, and The eigenvectors are The mean and standard deviation of the corresponding feature set, represents the position-wise square root of the eigenvector, and is the logarithmic function with base 2, is the corrected eigenvector, represents positional addition, Represents point multiplication by position; the corrected feature vector The pressure-heat distribution depth constraint interactive fusion feature vector is fused to obtain a corrected pressure-heat distribution depth constraint interactive fusion feature vector.

[0073] Here, in order to improve the mapping effect of the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector to the fusion feature distribution domain in the feature fusion scenario, based on the limitations of the traditional weighted fusion method for inferring the semantic space evolution diffusion mode based on feature superposition, a low-order superposition fusion mode and a high-order superposition fusion mode are adopted to combine the space, and the statistical feature interaction relationship of the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is used to simulate the evolution center and evolution trajectory, so as to reconstruct the semantic space evolution diffusion in the fusion scenario based on asynchronous evolution under the action of different evolution diffusion speed fields, effectively improving the projection effect to the same high-dimensional feature space, thereby improving the fusion effect of the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector. In this way, the correction feature vector used to optimize the fusion is used. Fusion with the pressure-heat distribution deep constraint interactive fusion feature vector improves the deep cross-modal constraint fusion expression effect of the pressure-heat distribution deep constraint interactive fusion feature vector, thereby improving the accuracy of the classification result obtained by the pressure-heat distribution deep constraint interactive fusion feature vector through the classifier.

[0074] Furthermore, the corrected pressure-heat distribution depth constraint interactive fusion feature vector is passed through a classifier-based desorption control decision maker to obtain a control decision result, and the control decision result is used to indicate whether to stop heating.

[0075] In summary, the above scheme is adopted to collect the thermal infrared distribution image data and the pressure data of the cold trap tube by comprehensively utilizing infrared sensors and pressure sensors, and the deep cross-modal constraint interactive fusion features between the two are used to characterize the desorption process and desorption effect, so as to intelligently judge when to stop heating based on this deep cross-modal constraint interactive fusion features.

[0076] Figure 2 FIG. 1 is a block diagram of a fully automatic online atmospheric pre-concentration device according to an exemplary embodiment. Figure 2 As shown, the fully automatic online atmospheric pre-concentration device 200 comprises:

[0077] A pre-purge module 201 is used to complete cold trap preparation and purge residual substances in the cold trap;

[0078] An enrichment module 202, used for enriching the gas in a cold trap tube of the cold trap;

[0079] A desorption module 203, used to rapidly heat up the cold trap tube;

[0080] The sample injection module 204 is used to switch the six-way valve to elute the gas out of the cold trap tube, inject it into the analyzer, and connect it to the detector through the analytical column;

[0081] Backflush module 205, used for reverse switching the switching valve to backflush the undesorbed high boiling point gas out of the cold trap tube and enter the next cycle;

[0082] Characterized in that the desorption module 203 comprises:

[0083] The thermal infrared distribution image acquisition unit 2031 is used to acquire the thermal infrared distribution images of the cold trap tube at multiple predetermined time points within a predetermined time period collected by the infrared sensor to obtain a set of thermal infrared distribution images;

[0084] A pressure value acquisition unit 2032, used to acquire a time series of pressure values ​​of the cold trap tube collected by a pressure sensor;

[0085] A feature extraction unit 2033 is used to perform feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors;

[0086] The deep cross-modal constraint fusion unit 2034 is used to determine the control decision result based on the deep cross-modal constraint fusion information between the sequence of the pressure time series feature vector and the sequence of the thermal infrared distribution state feature vector.

[0087] In one embodiment of the present application, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed in a computer, the computer is caused to execute the fully automatic online atmospheric pre-concentration method.

[0088] In one embodiment of the present application, an electronic device is provided, comprising: a processor; a memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the fully automatic online atmospheric pre-concentration method as described above.

[0089] Reference below Figure 3 , which shows a schematic diagram of the structure of an electronic device 600 suitable for implementing the embodiment of the present disclosure. The terminal device in the embodiment of the present disclosure may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0090] like Figure 3 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0091] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0092] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0093] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0094] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0095] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0096] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0097] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0098] The modules involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a module does not limit the module itself in some cases. For example, a test parameter acquisition module may also be described as a "module for acquiring device test parameters corresponding to a target device".

[0099] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0100] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0101] According to one or more embodiments of the present disclosure, Example 1 provides a fully automatic online atmospheric pre-concentration method, the method comprising:

[0102] The first step is pre-purging, completing the cold trap preparation and purging the residual substances in the cold trap; the second step is enrichment, enriching the gas in the cold trap tube of the cold trap; the third step is desorption, rapidly heating the cold trap tube; the fourth step is sample injection, switching the six-way valve, eluting the gas out of the cold trap tube, injecting it into the analyzer, and connecting the detector through the analytical column; the fifth step is backflush, reverse switching the conversion valve, backflush the non-desorbed high-boiling-point gas out of the cold trap tube, and entering the next cycle;

[0103] Characterized in that the third step comprises:

[0104] Acquire thermal infrared distribution images of the cold trap tube at a plurality of predetermined time points within a predetermined time period collected by an infrared sensor to obtain a set of thermal infrared distribution images;

[0105] Acquire a time series of pressure values ​​of the cold trap tube collected by a pressure sensor;

[0106] Performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​respectively to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors;

[0107] A control decision result is determined based on deep cross-modal constraint fusion information between the sequences of the pressure time series feature vector and the thermal infrared distribution state feature vector.

[0108] According to one or more embodiments of the present disclosure, Example 2 provides the method of Example 1, performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors, including:

[0109] Performing time series analysis on the time series of the pressure values ​​to obtain the pressure time series feature vector;

[0110] Thermal infrared distribution state feature extraction is performed on the set of thermal infrared distribution images to obtain a sequence of thermal infrared distribution state feature vectors.

[0111] According to one or more embodiments of the present disclosure, Example 3 provides the method of Example 2, performing a time series analysis on the time series of the pressure value to obtain the pressure time series feature vector, including:

[0112] Arranging the time series of the pressure values ​​into a pressure time series input vector according to the time dimension;

[0113] The pressure time series input vector is passed through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain the pressure time series feature vector.

[0114] According to one or more embodiments of the present disclosure, Example 4 provides the method of Example 3, wherein the pressure time series input vector is passed through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain the pressure time series feature vector, including:

[0115] The pressure time series input vector is processed using the following one-dimensional extended convolution formula to obtain the pressure time series feature vector; wherein the one-dimensional extended convolution formula is: ; Among them, the one-dimensional expanded convolution kernel The size is , is the length of the original convolution kernel, is the expansion rate, Indicates that the pressure time series input vector The length of the first eigenvalue at each position is time window, is a bias term, and , is a nonlinear activation function, It is local convolutional coding feature vector, is the dimension of the pressure time series input vector, is the pressure time series characteristic vector, Indicates cascade.

[0116] According to one or more embodiments of the present disclosure, Example 5 provides the method of Example 4, performing thermal infrared distribution state feature extraction on the set of thermal infrared distribution images to obtain a sequence of thermal infrared distribution state feature vectors, including:

[0117] The set of thermal infrared distribution images is passed through a thermal infrared distribution state feature extractor based on a convolutional neural network model to obtain a sequence of thermal infrared distribution state feature vectors.

[0118] According to one or more embodiments of the present disclosure, Example 6 provides the method of Example 5, which determines a control decision result based on deep cross-modal constraint fusion information between the pressure time series feature vector and the sequence of the thermal infrared distribution state feature vector, including:

[0119] The sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector;

[0120] Performing feature distribution correction on the pressure-heat distribution depth constraint interactive fusion feature vector to obtain a corrected pressure-heat distribution depth constraint interactive fusion feature vector;

[0121] The corrected pressure-heat distribution depth constraint interactive fusion feature vector is passed through a classifier-based desorption control decision maker to obtain the control decision result, and the control decision result is used to indicate whether to stop heating.

[0122] According to one or more embodiments of the present disclosure, Example 7 provides the method of Example 6, wherein the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector, including:

[0123] The sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is processed by the following deep cross-modal constraint fusion formula to obtain the pressure-heat distribution deep constraint interactive fusion feature vector; wherein the deep cross-modal constraint fusion formula is: ;in, is a constant bias term, It is A bias vector, It is A bias matrix, is the first in the sequence of the thermal infrared distribution state feature vector Thermal infrared distribution state feature vector, is the pressure time series characteristic vector, represents the transpose of a vector, is the first in the pressure-heat distribution depth constraint interactive fusion feature vector feature values.

[0124] According to one or more embodiments of the present disclosure, Example 8 provides a fully automatic online atmospheric pre-concentration device, the device comprising:

[0125] A pre-purge module, used to complete cold trap preparation and purge residual substances in the cold trap;

[0126] An enrichment module, used for enriching the gas in a cold trap tube of the cold trap;

[0127] A desorption module, used for rapidly heating the cold trap tube;

[0128] A sample injection module, used for switching the six-way valve, eluting the gas out of the cold trap tube, injecting it into the analyzer, and connecting it to the detector through the analytical column;

[0129] A backflush module, used for reversely switching the conversion valve to backflush the undesorbed high-boiling-point gas out of the cold trap tube to enter the next cycle;

[0130] Characterized in that the desorption module comprises:

[0131] A thermal infrared distribution image acquisition unit, used to acquire thermal infrared distribution images of the cold trap tube at a plurality of predetermined time points within a predetermined time period collected by an infrared sensor to obtain a set of thermal infrared distribution images;

[0132] A pressure value acquisition unit, used to acquire a time series of pressure values ​​of the cold trap tube collected by a pressure sensor;

[0133] A feature extraction unit, used for performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors;

[0134] A deep cross-modal constraint fusion unit is used to determine a control decision result based on deep cross-modal constraint fusion information between the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector.

[0135] According to one or more embodiments of the present disclosure, Example 9 provides a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed in a computer, the computer is caused to execute the steps of implementing the methods described in Examples 1 to 7.

[0136] According to one or more embodiments of the present disclosure, Example 10 provides an electronic device, characterized in that it includes:

[0137] processor;

[0138] A memory, in which computer program instructions are stored, and when the computer program instructions are executed by the processor, the processor performs the steps of the method described in Example 1 to Example 7.

[0139] Figure 4 FIG. 1 is an application scenario diagram of a fully automatic online atmospheric pre-concentration method according to an exemplary embodiment. Figure 4 As shown, in this application scenario, first, the thermal infrared distribution images of the cold trap tube at multiple predetermined time points within a predetermined time period collected by the infrared sensor are obtained to obtain a set of thermal infrared distribution images (for example, Figure 4 ), obtaining a time series of pressure values ​​of the cold trap tube collected by the pressure sensor (for example, as shown in C1 in FIG. Figure 4 Then, the acquired thermal infrared distribution image and pressure value are input to a server deployed with a fully automatic online atmospheric pre-concentration algorithm (for example, Figure 4 In S) shown in FIG. 1 , the server is capable of processing the thermal infrared distribution image and the pressure value based on a fully automatic online atmospheric pre-concentration algorithm to determine a control decision result.

[0140] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0141] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0142] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Claims

1. A fully automatic online atmospheric pre-concentration method, comprising: The first step is pre-purging, completing the cold trap preparation and purging the residual substances in the cold trap; The second step is enrichment, enriching the gas in the cold trap tube of the cold trap; The third step is desorption, which rapidly heats up the cold trap tube; the fourth step is sample injection, which switches the six-way valve to elute the gas out of the cold trap tube, injects it into the analyzer, and connects the detector through the analytical column; the fifth step is backflush, which switches the conversion valve in reverse to backflush the non-desorbed high-boiling-point gas out of the cold trap tube and enter the next cycle; Characterized in that the third step comprises: Acquire thermal infrared distribution images of the cold trap tube at a plurality of predetermined time points within a predetermined time period collected by an infrared sensor to obtain a set of thermal infrared distribution images; Acquire a time series of pressure values ​​of the cold trap tube collected by a pressure sensor; Performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​respectively to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors; Determining a control decision result based on deep cross-modal constraint fusion information between the pressure time series feature vector and the sequence of the thermal infrared distribution state feature vector; Wherein, determining the control decision result based on the deep cross-modal constraint fusion information between the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector includes: The sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector; Performing feature distribution correction on the pressure-heat distribution depth constraint interactive fusion feature vector to obtain a corrected pressure-heat distribution depth constraint interactive fusion feature vector; The corrected pressure-heat distribution depth constraint interactive fusion feature vector is passed through a classifier-based desorption control decision maker to obtain the control decision result, and the control decision result is used to indicate whether to stop heating; The pressure time series feature vector and the thermal infrared distribution state feature vector are passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector, including: The sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is processed by the following deep cross-modal constraint fusion formula to obtain the pressure-heat distribution deep constraint interactive fusion feature vector; wherein the deep cross-modal constraint fusion formula is: ;in, is a constant bias term, It is A bias vector, It is A bias matrix, is the first in the sequence of the thermal infrared distribution state feature vector Thermal infrared distribution state feature vector, is the pressure time series characteristic vector, represents the transpose of a vector, is the first in the pressure-heat distribution depth constraint interactive fusion feature vector feature values.

2. The fully automatic online atmospheric pre-concentration method according to claim 1, characterized in that: The set of thermal infrared distribution images and the time series of pressure values ​​are respectively subjected to feature extraction to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors, including: Performing time series analysis on the time series of the pressure values ​​to obtain the pressure time series feature vector; Thermal infrared distribution state feature extraction is performed on the set of thermal infrared distribution images to obtain a sequence of thermal infrared distribution state feature vectors.

3. The fully automatic online atmospheric pre-concentration method according to claim 2, characterized in that: Performing time series analysis on the time series of the pressure values ​​to obtain the pressure time series feature vector includes: Arranging the time series of the pressure values ​​into a pressure time series input vector according to the time dimension; The pressure time series input vector is passed through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain the pressure time series feature vector.

4. The fully automatic online atmospheric pre-concentration method according to claim 3, characterized in that: The pressure time series input vector is passed through a pressure time series feature extractor based on a one-dimensional extended convolutional neural network model to obtain the pressure time series feature vector, including: The pressure time series input vector is processed using the following one-dimensional extended convolution formula to obtain the pressure time series feature vector; wherein the one-dimensional extended convolution formula is: ; Among them, the one-dimensional expanded convolution kernel The size is , is the length of the original convolution kernel, is the expansion rate, Indicates that the pressure time series input vector The length of the first feature value at the position is time window, is the bias term, and , is a nonlinear activation function, It is local convolutional coding feature vector, is the dimension of the pressure time series input vector, is the pressure time series characteristic vector, Indicates cascade.

5. The fully automatic online atmospheric pre-concentration method according to claim 4, characterized in that: Extracting thermal infrared distribution state features from the set of thermal infrared distribution images to obtain a sequence of thermal infrared distribution state feature vectors includes: The set of thermal infrared distribution images is passed through a thermal infrared distribution state feature extractor based on a convolutional neural network model to obtain a sequence of thermal infrared distribution state feature vectors.

6. A fully automatic online atmospheric pre-concentration device, characterized in that: include: A pre-purge module, used to complete cold trap preparation and purge residual substances in the cold trap; An enrichment module, used for enriching the gas in a cold trap tube of the cold trap; A desorption module, used for rapidly heating the cold trap tube; A sample injection module, used for switching the six-way valve, eluting the gas out of the cold trap tube, injecting it into the analyzer, and connecting it to the detector through the analytical column; A backflush module, used for reversely switching the conversion valve to backflush the undesorbed high-boiling-point gas out of the cold trap tube to enter the next cycle; Characterized in that the desorption module comprises: A thermal infrared distribution image acquisition unit, used to acquire thermal infrared distribution images of the cold trap tube at a plurality of predetermined time points within a predetermined time period collected by an infrared sensor to obtain a set of thermal infrared distribution images; A pressure value acquisition unit, used to acquire a time series of pressure values ​​of the cold trap tube collected by a pressure sensor; A feature extraction unit, used for performing feature extraction on the set of thermal infrared distribution images and the time series of pressure values ​​to obtain a sequence of pressure time series feature vectors and thermal infrared distribution state feature vectors; A deep cross-modal constraint fusion unit, used to determine a control decision result based on deep cross-modal constraint fusion information between the pressure time series feature vector and the sequence of the thermal infrared distribution state feature vector; Wherein, determining the control decision result based on the deep cross-modal constraint fusion information between the sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector includes: The sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector; Performing feature distribution correction on the pressure-heat distribution depth constraint interactive fusion feature vector to obtain a corrected pressure-heat distribution depth constraint interactive fusion feature vector; The corrected pressure-heat distribution depth constraint interactive fusion feature vector is passed through a classifier-based desorption control decision maker to obtain the control decision result, and the control decision result is used to indicate whether to stop heating; The pressure time series feature vector and the thermal infrared distribution state feature vector are passed through a deep cross-modal constraint fusion network to obtain a pressure-heat distribution deep constraint interactive fusion feature vector, including: The sequence of the pressure time series feature vector and the thermal infrared distribution state feature vector is processed by the following deep cross-modal constraint fusion formula to obtain the pressure-heat distribution deep constraint interactive fusion feature vector; wherein the deep cross-modal constraint fusion formula is: ;in, is a constant bias term, It is A bias vector, It is A bias matrix, is the first in the sequence of the thermal infrared distribution state feature vector Thermal infrared distribution state feature vector, is the pressure time series characteristic vector, represents the transpose of a vector, is the first in the pressure-heat distribution depth constraint interactive fusion feature vector feature values.

7. A computer-readable storage medium having computer program instructions stored thereon, wherein when the computer program instructions are executed in a computer, the computer is caused to execute the fully automatic online atmospheric pre-concentration method according to any one of claims 1 to 5.

8. An electronic device, characterized in that: include: processor; A memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the fully automatic online atmospheric pre-concentration method according to any one of claims 1 to 5.

Images