Gas classification method and device based on pooling aggregation and joint attention enhanced LSTM

Abstract

The application discloses a gas classification method and equipment based on pooling aggregation and joint attention enhanced LSTM, wherein the gas classification method first performs low-parameter dimension reduction on original data of a multi-sensor array through a pooling aggregation module; secondly, a joint attention module composed of a channel attention submodule and a time attention submodule is introduced to respectively adaptively weight different sensor channels and time sequence positions, so that the effective gas response features are strengthened and the environmental noise is suppressed, and the signal and noise can be distinguished without relying on baseline calibration; finally, the screened features are input into a single-layer lightweight long short-term memory network module for time sequence modeling, so that an efficient recognition architecture adapting to short sequence data is constructed, the local and global features in a short time sequence are deeply mined, the effective signal and noise are accurately distinguished under the condition of no baseline, the model calculation complexity is greatly reduced, and the rapid reasoning and lightweight deployment requirements of a portable device are met.

Description

Technical Field

[0001] This invention relates to the field of electronic nose gas recognition technology, specifically to a gas classification method and device based on pooling aggregation and joint attention-enhanced LSTM. Background Technology

[0002] Electronic nose technology simulates the biological olfactory system, using a gas sensor array to collect gas response signals and combining them with pattern recognition algorithms to achieve rapid detection and identification of gas types. In recent years, electronic nose devices with "short time series and no baseline output" have been widely used in lightweight applications such as portable on-site detection, rapid screening in the fresh food distribution process, real-time monitoring of leaks in chemical industrial parks, and gas safety early warning in home environments.

[0003] The core data characteristics of such lightweight electronic nose devices differ fundamentally from those of multi-channel long-time-series devices: the time sequence length of a single detection is typically only a few dozen time steps, and due to the lightweight design of the device, a baseline output function is not configured. This leads to a significant change in the core data processing requirements: (1) within a limited short time window, it is necessary to efficiently mine the intrinsic relationship between local and global features to avoid missing key gas response features; (2) under conditions without baseline calibration, it is necessary to achieve accurate differentiation between effective gas response signals and environmental noise; (3) it is necessary to significantly reduce the computational complexity and number of parameters of the model to adapt to the rapid inference and lightweight deployment requirements of portable devices.

[0004] Current traditional electronic nose data processing models suffer from numerous technical shortcomings and application limitations when dealing with short-term, baseline-free data: First, most deep learning models are designed for long-term data and struggle to achieve deep correlations between local and global features within short time windows, leading to insufficient feature mining; second, traditional models rely on baseline calibration for noise removal, lacking an effective signal-noise differentiation mechanism in baseline-free scenarios and exhibiting insufficient anti-interference capabilities; third, existing complex network architectures have a large number of parameters, failing to meet the requirements of millisecond-level fast inference and lightweight embedded deployment for portable devices, severely restricting their engineering application in rapid on-site detection scenarios.

[0005] Furthermore, existing electronic noses lack any air intake filtration mechanism, thus relying on direct inhalation and detection of the target gas during the testing process. The applicant of this application discovered during application that certain non-target gases and impurities can interfere with the gas sensor's recognition signal, thereby causing issues with detection accuracy.

[0006] Furthermore, due to the different detection targets or purposes, electronic noses require specific combinations of gas sensors for detection. Therefore, please refer to Chinese patent applications with publication numbers CN119936122A and CN119959474A, etc. Existing electronic noses are all specialized devices, capable of detecting only one type of target or a broad category of targets, and classifying them into smaller subcategories based on the collected information. This results in very poor versatility. Different electronic noses need to be prepared specifically for different detection targets or purposes, incurring huge costs and limiting the application and promotion of electronic noses. Moreover, existing miniaturized and compact electronic noses lack redundant detection capabilities (redundant arrangement). To meet processing algorithm requirements and ensure detection accuracy, multiple repeated measurements are required, leading to low detection efficiency and poor operational convenience.

[0007] Solving these problems is now a top priority. Summary of the Invention

[0008] To address the technical challenges faced by electronic nose devices with "short time series and no baseline output" in gas identification, such as insufficient feature mining, weak noise discrimination ability, and insufficient model lightweighting, this invention provides a gas classification method and device based on pooling aggregation and joint attention-enhanced LSTM.

[0009] The technical solution is as follows:

[0010] The first aspect of this application relates to a gas classification method based on pooling aggregation and joint attention-enhanced LSTM, which is carried out according to the following steps:

[0011] S1. The electronic nose inhales the target sample. The gas sensor array of the electronic nose collects the gas response data of the target sample and converts the gas response data into a continuous analog signal sequence.

[0012] S2. The pooling aggregation module uses a shared global average pooling layer to perform global average pooling on the continuous analog signal sequence output by each gas sensor array. The pooling aggregation module includes a time pooling module and a channel pooling module. The time pooling module is used to perform global pooling on the time dimension of all gas sensors, and the channel pooling module is used to perform global pooling on the channel dimension within a single gas sensor array.

[0013] S3. Input the feature vectors output by the pooling aggregation module after performing global average pooling operation into the joint attention module composed of the channel attention submodule and the temporal attention submodule respectively.

[0014] Specifically, the channel attention submodule uses learnable parameters to construct an adaptive channel importance weight matrix to perform a channel importance weighting operation on the feature vectors corresponding to each gas sensor array output by the channel pooling module, obtaining the channel feature vectors corresponding to each gas sensor array. Then, the channel feature vectors corresponding to each gas sensor array are summed to obtain the channel feature vector integrating the output signals of all gas sensor arrays. The time attention submodule uses learnable parameters to construct an adaptive time step importance weight vector to perform a time step differential weighting operation on the different time step features of the feature vector output by the time pooling module, obtaining the time-series feature vector integrating the output signals of all gas sensor arrays.

[0015] The channel feature vector and the time-series feature vector are summed to generate a fused time-channel joint feature vector;

[0016] S4. Input the time-channel joint feature vector into a single-layer lightweight long short-term memory network module to perform temporal feature modeling, generate the final output feature vector, and classify the target sample according to the final output feature vector to obtain the classification result of the target sample.

[0017] The gas classification method based on pooling aggregation and joint attention-enhanced LSTM achieved the following technical results:

[0018] 1. High efficiency in dimensionality reduction with low parameter count: The shared global average pooling layer can complete the dimensionality reduction of multi-gas sensor array data with only a very small number of parameters, while preserving the global core features and significantly reducing the overall computational burden of the model.

[0019] 2. Precise feature selection in two dimensions: The collaborative work of the channel attention submodule and the time attention submodule enables the enhancement of effective features and the suppression of redundant noise from both channel and time dimensions, which significantly improves the overall quality and discriminativeness of features and solves the technical problem of difficult feature localization in key time steps in short time series data.

[0020] 3. Lightweight and fast inference capability: The minimalist single-layer LSTM structure design greatly simplifies the network architecture and reduces the number of model parameters, enabling fast inference of short-sequence electronic nose data, perfectly adapting to the lightweight deployment requirements of portable devices.

[0021] 4. Noise discrimination capability in baseline-free scenarios: Without relying on baseline calibration, the model can accurately distinguish between effective gas response signals and environmental noise through a dual attention mechanism, which greatly improves the model's anti-interference capability in baseline-free scenarios.

[0022] 5. Deep feature mining within short time series: It realizes deep correlation and mining of local and global features within a short time series window, which fully meets the gas recognition requirements of short-sequence electronic nose data and ensures the recognition accuracy of the model.

[0023] A second aspect of this application relates to an electronic device, including at least one processor and a memory communicatively connected to the at least one processor, the memory storing a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the gas classification method described above.

[0024] Using the above electronic equipment, all the advantages of the gas classification methods mentioned above are achieved.

[0025] A third aspect of this application relates to a computer-readable medium storing computer instructions for causing at least one processor to execute the gas classification method described above.

[0026] Using the above computer-readable media, all the advantages of the gas classification methods described above are achieved.

[0027] The fourth aspect of this application relates to an electronic nose, which includes a universal gas detection assembly and an electronic nose air intake mechanism;

[0028] The electronic nose air intake mechanism includes an air pump with inlet and outlet ports. A filter assembly is connected to the outer end of the inlet and outlet ports. A pressure sensor is installed inside the air pump. The filter assembly includes an assembly housing, which integrates an inlet nozzle, an outlet nozzle, an inlet / outlet duct communicating with the inlet and outlet ports, an exhaust channel connecting the inlet / outlet duct and the outlet nozzle, and a filter channel communicating with the inlet / outlet duct. At least two stages of filters are sequentially arranged on the filter channel. The inlet end of each filter is connected to the inlet nozzle through a corresponding inlet channel. An on / off actuator is installed on the assembly housing to switch the on / off state of the exhaust channel and each inlet channel. When the on / off actuator keeps the exhaust channel open, all inlet channels are cut off by the on / off actuator; when the on / off actuator keeps any inlet channel open, the exhaust channel and the remaining inlet channels are cut off by the on / off actuator.

[0029] The universal gas detection assembly includes a detection box and an intake splitter assembly and an exhaust assembly respectively disposed at both ends of the detection box. The exhaust nozzle is connected to the intake end of the intake splitter assembly through a pipeline. The detection box is provided with a central container and multiple side mounting slots distributed circumferentially around the central mounting slot. A first gas sensor array is installed in the central container. The first gas sensor array and the central container together form a universal sensor detection chamber that connects the intake splitter assembly and the exhaust assembly. A second gas sensor array is installed in each of the side mounting slots. Each second gas sensor array and the corresponding side mounting slot together form a quick-change sensor detection chamber that connects the intake splitter assembly and the exhaust assembly. The outer wall of the detection box is equipped with quick-change sealing plates that can be detachably covered at the opening of the corresponding side mounting slot.

[0030] The electronic nose employing the above-mentioned design not only allows the air intake mechanism to connect and disconnect the corresponding air intake channel based on the type of gas being tested and the detection environment, ensuring that the gas being tested is filtered through various filters as it is drawn into the air pump, removing interfering gases and impurities, thus effectively improving the recognition accuracy of subsequent gas sensors and consequently enhancing the detection accuracy of the electronic nose; but it also allows for pressurization of the gas being tested according to its type and detection requirements, meeting the need for specific gases to be detected at specific pressures, further improving the applicability and versatility of the electronic nose; and it also possesses the capability to filter different levels of filters. The filter element's self-inspection and self-cleaning functions further ensure detection accuracy. Furthermore, by incorporating a universal sensor detection chamber with a built-in first gas sensor array and multiple quick-change sensor detection chambers with built-in second gas sensor arrays, the versatile first gas sensor array, which requires less frequent replacement, is installed in the central container. The second gas sensor arrays, which need to be adaptively adjusted for different detection targets, are installed in the quick-change sensor detection chambers surrounding the universal sensor detection chamber. Simultaneously, the easily removable quick-change sealing plates allow for convenient access to the second gas sensor arrays in the side mounting slots. The system features quick-change functionality and simple, convenient operation. By combining the types of the first gas sensor array with various second gas sensor arrays, a modular sensor array combination capable of accurately analyzing different target gases can be formed. This adaptable combination is suitable for detecting different targets, offering excellent versatility. This makes the electronic nose a universal device applicable to various targets, significantly reducing the investment cost and facilitating its application, promotion, and widespread adoption. Furthermore, the system allows for identical settings of the type and parameters of the second gas sensor arrays in two or more quick-change sensor detection chambers, enabling a single test to achieve the equivalent of multiple redundant tests. The system achieves the desired effect, satisfying the data acquisition requirements of the processing algorithm and obtaining high-precision detection results, while significantly improving testing efficiency and operational convenience. Furthermore, the first gas sensor array and each of the second gas sensor arrays are located in their own independent chambers to detect gases, without any interference between them, thus further improving detection accuracy. The gas universal detection assembly and the electronic nose air intake mechanism adopt a split structure, with both parts being extremely compact. Different air intake electronic nose air intake mechanisms can be replaced according to the detection needs of the gas being tested, meeting the application requirements of miniaturization and modularization, with high integration and excellent portability.

[0031] The fifth aspect of this application relates to a control method for the aforementioned electronic nose, the control method comprising the following steps:

[0032] A1. Check the condition of the filter elements of each stage of the electronic nose air intake mechanism to ensure they are functioning properly, following these steps:

[0033] A11. Air intake nozzle connects to air;

[0034] A12. The on / off actuator keeps the air intake passage connected to the first-stage filter open until the air pump is full of air.

[0035] A13. The on / off actuator cuts off the exhaust passage and all intake passages;

[0036] A14. The air pump compresses the internal air until the pressure sensor detects that the internal air pressure of the air pump has reached the self-test set value.

[0037] A15. The on / off actuator keeps the air intake passage connected to the first-stage filter open until the internal air pressure of the air pump returns to normal pressure, and determines whether the time to return to normal pressure is greater than the time threshold: if yes, proceed to step A2; if no, proceed to step A3.

[0038] A2. Clean the filter elements of each stage of the electronic nose air intake mechanism according to the following steps:

[0039] A21. The on / off actuator connects the air intake passage that is connected to the last stage filter;

[0040] A22. After the air pump is filled with air, the air pump quickly empties the air inside and determines whether the number of repetitions of step A22 is greater than the repetition threshold: if yes, after the air pump is filled with air, proceed to step A23; otherwise, repeat step A22.

[0041] A23. After the on / off actuator cuts off the exhaust passage and all intake passages, the air pump compresses the internal air until the pressure sensor detects that the internal air pressure of the air pump has reached the self-test set value.

[0042] A24. Connect the on / off actuator to the air intake channel connected to the last stage filter until the internal air pressure of the air pump returns to normal pressure, and determine whether the time to return to normal pressure is greater than the time threshold: if yes, replace the filter element of the last stage filter first, and then connect the on / off actuator to the air intake channel connected to the previous stage filter, and proceed to step A25; if no, after connecting the on / off actuator to the air intake channel connected to the previous stage filter, proceed to step A25.

[0043] A25. After the air pump is filled with air, the air pump quickly empties the air inside and determines whether the number of repetitions of step A25 is greater than the repetition threshold: if yes, after the air pump is filled with air, proceed to step A26; if no, repeat step A25.

[0044] A26. After the on / off actuator cuts off the exhaust passage and all intake passages, the air pump compresses the internal air until the pressure sensor detects that the internal air pressure of the air pump has reached the self-test set value.

[0045] A27. The on / off actuator connects the previously connected air intake channel until the internal air pressure of the air pump returns to normal pressure, and determines whether the time to return to normal pressure is greater than the time threshold: if yes, replace the filter element of the first-stage filter closest to the unobstructed air intake channel and proceed to step A28; otherwise, proceed to step A28.

[0046] A28. Determine whether the primary filter closest to the unobstructed air intake channel is the first-stage filter: If yes, after the air pump empties the internal air, proceed to step A3; if no, after the air pump empties the internal air, the on / off actuator connects the air intake channel connected to the previous primary filter, and then returns to step A25.

[0047] A3. To test the gas, follow these steps:

[0048] A31. Connect the air inlet nozzle to the gas being tested;

[0049] A32. Based on the type of gas being tested, determine whether the combination of the second gas sensor array installed in the mounting slots on each side meets the testing requirements: if yes, proceed to step A33; if no, replace the combination of the second gas sensor array corresponding to the gas being tested and proceed to step A33.

[0050] A33. Depending on the type of gas being tested, the on / off actuator keeps one of the air intake channels corresponding to the gas being tested open until the air pump is filled with the gas being tested.

[0051] A34. Based on the type of gas being tested, determine whether the gas inside the compressed air pump is needed: if yes, proceed to step A35; otherwise, proceed to step A37.

[0052] A35. The on / off actuator cuts off the exhaust passage and all intake passages;

[0053] A36. The air pump compresses the gas being tested inside until the volume of the gas being tested is compressed to the detection set value required for its detection.

[0054] A37. The on / off actuator keeps the exhaust passage open until the air pump empties the gas being tested inside. During this process, all the first gas sensor arrays and the second gas sensor arrays detect the gas being tested.

[0055] The control method of the above electronic nose not only has all the advantages of the above electronic nose, but also enables the electronic nose to perform self-inspection and self-cleaning of the filter elements of each stage of the filter, as well as to reasonably select the combination of redundant detection modules of fast-changing gas sensors and adjust the detection pressure of the gas to be detected according to the type of gas being detected. Attached Figure Description

[0056] Figure 1This is a schematic diagram illustrating the principle of a gas classification method.

[0057] Figure 2 A three-dimensional structural diagram of the electronic nose air intake mechanism;

[0058] Figure 3 This is a schematic diagram of the planar structure of the electronic nose air intake mechanism;

[0059] Figure 4 for Figure 3 Sectional view at point AA;

[0060] Figure 5 for Figure 3 Sectional view at point BB;

[0061] Figure 6 This is a schematic diagram of the general-purpose gas detection assembly;

[0062] Figure 7 This is a cross-sectional view of the general gas detection assembly when using the second gas sensor array in Embodiment 1.

[0063] Figure 8 This is a schematic diagram of the gas universal detection assembly after the quick-change cover is removed when using the second gas sensor array in Embodiment 1.

[0064] Figure 9 for Figure 8 A schematic diagram of the structure after removing the quick-change gas sensor circuit board;

[0065] Figure 10 for Figure 9 A schematic diagram of the structure after removing the rubber gasket pressure plate and sealing gasket;

[0066] Figure 11 This is a cross-sectional view of the general gas detection assembly when using the first gas sensor array in Embodiment 2.

[0067] Figure 12 This is a schematic diagram of the gas universal detection assembly after the quick-change cover is removed when using the first gas sensor array implementation method 2.

[0068] Figure 13 This is a schematic diagram of the volume adjustment component.

[0069] Figure 14 This is a schematic diagram of the structure of the central container. Detailed Implementation

[0070] The present invention will be further described below with reference to the embodiments and accompanying drawings.

[0071] Example 1:

[0072] like Figure 1As shown, a gas classification method based on pooling aggregation and joint attention-enhanced LSTM is performed according to the following steps:

[0073] S1. The electronic nose inhales the target sample. The gas sensor arrays of the electronic nose collect the gas response data of the target sample and convert the gas response data into a continuous analog signal sequence, denoted as... ,in, Indicates the number of gas sensor arrays. This indicates the number of sensors in each gas sensor array. This represents the time step length of the continuous analog signal sequence for each gas sensor array.

[0074] S2. The pooling aggregation module uses a shared global average pooling layer to perform global average pooling on the continuous analog signal sequence output by each gas sensor array.

[0075] Specifically, the core objective of the pooling aggregation module is to quickly complete the dimensionality reduction and global feature extraction of data from various gas sensor arrays under the constraint of a very small number of parameters, so as to provide high-quality feature input for the subsequent attention screening process.

[0076] Based on the working characteristics of gas sensor arrays based on electronic noses, the sensor units within the same gas sensor array usually have similar response trends and feature distributions for the target gas. The data they collect has natural spatial redundancy. Without building a complex network architecture, effective dimensionality reduction and global feature extraction can be achieved through simple pooling operations.

[0077] The pooling aggregation module includes a time pooling module and a channel pooling module. The time pooling module is used to perform global pooling in the time dimension of all gas sensors, while the channel pooling module is used to perform global pooling in the channel dimension within a single gas sensor array.

[0078] The continuous analog signal sequences output by each gas sensor array are input into the channel pooling module. The calculation formula for performing one-dimensional global average pooling operation on each gas sensor array in the channel dimension is as follows:

[0079] ;

[0080] In the above formula, Indicates the corresponding to the first The first gas sensor array Features obtained from a gas sensor through global average pooling. Indicates the first The data from the gas sensor array of the first The first gas sensor Data at each time step.

[0081] The channel feature vector corresponding to each gas sensor array is obtained through the channel pooling module. .

[0082] Simultaneously, the continuous analog signal sequences output by each gas sensor array are input into the time pooling module. The calculation formula for performing a one-dimensional global average pooling operation on all gas sensor arrays in the time dimension is as follows:

[0083] ;

[0084] In the above formula, Indicates corresponding to the first Features obtained at each time step through global average pooling Indicates the first The data from the gas sensor array of the first The first gas sensor Data at each time step, This indicates the number of gas sensors in the gas sensor array.

[0085] The temporal feature vectors of all gas sensor arrays are obtained through the time pooling module. .

[0086] Therefore, global average pooling can compress high-dimensional gas sensor array data into low-dimensional feature vectors while preserving the global feature information of the gas sensor array. Furthermore, this operation has very few parameters and does not introduce additional computational burden to the model. After processing by this pooling aggregation module, the original data of [data missing] is [data missing]. The data was compressed to a size of The high-dimensional output data of each gas sensor array can be transformed into low-dimensional feature vectors. This process not only achieves efficient compression of data dimensions, but also preserves the core global features in the data, laying a solid high-quality data foundation for the subsequent attention screening process.

[0087] S3. Input the feature vectors output by the pooling aggregation module after performing global average pooling operation into the joint attention module composed of the channel attention submodule and the temporal attention submodule.

[0088] Specifically, the channel attention submodule uses learnable parameters to construct an adaptive channel importance weight matrix to perform channel importance weighting on the feature vectors output by the channel pooling module corresponding to each gas sensor array, thereby obtaining the channel feature vectors corresponding to each gas sensor array. Then, the channel feature vectors corresponding to each gas sensor array are summed to obtain the channel feature vector that integrates the output signals of all gas sensor arrays.

[0089] Specifically, the channel attention submodule uses learnable parameters to construct an adaptive channel importance weight matrix, and performs a channel importance weighting operation on the channel feature vectors corresponding to each gas sensor array using the following formula:

[0090] ;

[0091] In the above formula, Indicates corresponding to the first Channel feature vectors of a gas sensor array weighted by channel importance; Indicates corresponding to the first The learnable channel importance weight matrix for each gas sensor array has elements ranging from [0,1].

[0092] The channel feature vectors corresponding to each gas sensor array, after being weighted according to channel importance, are summed to generate a channel feature vector that integrates the output signals of all gas sensor arrays. The calculation formula is as follows:

[0093] ;

[0094] In the above formula, This represents the channel feature vector that integrates the output signals of all gas sensor arrays.

[0095] Meanwhile, the time attention submodule uses learnable parameters to construct an adaptive time step importance weight vector to perform time step differential weighting on the feature vector output by the time pooling module, thereby obtaining a time-series feature vector that integrates the output signals of all gas sensor arrays.

[0096] Specifically, the formula for the time-step differential weighting operation of the temporal feature vector output by the time-pooling module on the temporal attention submodule is as follows:

[0097] ;

[0098] In the above formula, This represents the time-series feature vector after integrating the output signals of all sensor arrays after time-step differential weighting. This represents the learnable time-step importance weight vector. It represents the Hadamardi (or Hadama) stack.

[0099] Finally, the temporal feature vector output by the temporal attention submodule and the channel feature vector output by the channel attention submodule are summed to obtain the temporal-channel joint feature vector, and the calculation formula is as follows:

[0100] ;

[0101] In the above formula, This represents the time-channel joint feature vector.

[0102] Therefore, in order to achieve accurate screening and enhancement of effective features in short-time, baseline-free electronic nose data, the joint attention module innovatively constructs a dual attention architecture in which the channel attention submodule and the time attention submodule work together to complete the weighted enhancement and redundancy suppression of features from the channel dimension and the time dimension, respectively.

[0103] Through the aforementioned weight allocation mechanism, the time attention submodule can accurately identify and enhance key time step features that contribute significantly to gas identification without allocating additional parameters to filter channel features. At the same time, it weakens the interference of noisy time steps, further improving the feature quality of the time series dimension and solving the technical problem of difficult localization of key time step features in short time series data.

[0104] S4. Input the time-channel joint feature vector into a single-layer lightweight long short-term memory network module to perform temporal feature modeling, generate the final output feature vector, and classify the target sample according to the final output feature vector to obtain the classification result of the target sample.

[0105] Specifically, Long Short-Term Memory (LSTM) networks, with their unique three-gate control mechanism of input gate, forget gate, and output gate, can efficiently and stably capture long-term and short-term dependencies in time-series data. They are a classic and efficient model choice for time-series classification tasks. The four core mechanisms are: input gate, forget gate, output gate, and candidate cell state.

[0106] First, there's the forget gate, which concatenates the current input and the hidden state from the previous time step into a vector. After passing through the sigmoid activation function, it generates a gate value between 0 and 1, which determines how much historical information to retain or discard. Its calculation is as follows:

[0107] ;

[0108] In the above formula, Indicates the first The output of the forget gate at each time step This represents the sigmoid activation function. This represents the forget gate weight matrix. This represents the forget gate bias vector. Indicates the first The hidden state vector at each time step, This represents a vector concatenation operation. The first eigenvector in the time-channel joint eigenvector represents the th eigenvector. An element at each time step.

[0109] Simultaneously, the input gate performs another sigmoid activation function operation on the concatenated vector to control the intake of new information. The operation formula is as follows:

[0110] ;

[0111] In the above formula, Indicates the first The output of the input gate at each time step, This represents the input gate weight matrix. This represents the input gate bias vector.

[0112] At the same time, the candidate cell state is updated, and the operation is as follows:

[0113] ;

[0114] In the above formula, Indicates the first Output features of candidate cell states at each time step This represents the candidate cell state weight matrix. This represents the candidate cell state bias vector. This represents the hyperbolic tangent function.

[0115] Each training session updates the cell state:

[0116] ;

[0117] In the above formula, Indicates the first Output features of candidate cell states at each time step Indicates the first Output features of candidate cell states at each time step.

[0118] The output gate also applies a sigmoid activation function to the concatenated vector to control the final output information. Its operation is as follows:

[0119] ;

[0120] In the above formula, Indicates the first The output of the gate at each time step This represents the output gate weight matrix. This represents the output gate bias vector.

[0121] Therefore, the formula for modeling temporal features of the feature vector after assigning differentiated weights in a single-layer lightweight long short-term memory network module is as follows:

[0122] ;

[0123] In the above formula, Indicates the first The final output feature vector at each time step.

[0124] To address the characteristics of short-sequence electronic nose data, this single-layer lightweight Long Short-Term Memory (LSTM) network module employs a single-layer LTM network structure for temporal correlation modeling. This simplifies the network architecture to the maximum extent possible while ensuring the model's ability to model temporal features. After dimensionality reduction processing by the pooling aggregation module and precise selection by the dual attention module, the input features possess high channel validity and temporal importance. Inputting these high-quality features into the single-layer lightweight LTM network module eliminates the need for additional hidden layers, enabling rapid deep modeling of short-sequence features and accurate gas type identification. This single-layer LTM network structure significantly reduces the number of model parameters and overall computational complexity, enabling rapid inference of short-sequence electronic nose data and perfectly adapting to the lightweight deployment requirements of portable electronic nose devices.

[0125] Example 2:

[0126] An electronic device includes at least one processor and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, which enables the at least one processor to perform the aforementioned gas classification method.

[0127] It should be noted that electronic devices are intended to represent various forms of digital computers, such as laptops, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. Electronic devices can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0128] A processor can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processors include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor performs the various methods and processes described above, such as gas classification methods.

[0129] In some embodiments, the gas classification method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on an electronic device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by a processor, one or more steps of the gas classification method described above may be performed. Alternatively, in other embodiments, the processor may be configured to perform the gas classification method by any other suitable means (e.g., by means of firmware).

[0130] Various implementations of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0131] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0132] Example 3:

[0133] Computer-readable storage media can be tangible media that may contain or store computer programs for use by or in conjunction with an instruction execution system, apparatus, or device. Computer-readable storage media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0134] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0135] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0136] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0137] Example 4:

[0138] Please see Figures 2-14 An electronic nose, comprising an electronic nose air intake mechanism and a universal gas detection assembly.

[0139] The electronic nose air intake mechanism includes an air pump 9, which has a pressure sensor 11 installed inside. By setting the pressure sensor 11, the pressure of the gas inside the air pump 9 can be monitored in real time. The air pump 9 is provided with an inlet and outlet port 9a for air intake and exhaust, and the outer end of the inlet and outlet port 9a is connected to a filter assembly 10 for filtering the inhaled gas.

[0140] The filter assembly 10 includes an assembly housing 10a, on which an intake nozzle 10c for air intake and an exhaust nozzle 10d for air exhaust are mounted. The assembly housing 10a internally integrates an intake / exhaust passage 10b communicating with the intake / exhaust port 9a, an exhaust passage 10e communicating with the intake / exhaust passage 10b and the exhaust nozzle 10d, and a filter passage 10f communicating with the intake / exhaust passage 10b. At least two stages of filters 10g are sequentially arranged on the filter passage 10f. Specifically, the filter passage 10f may have only one stage of filters 10g, or it may have multiple stages of filters 10g. The filter 10g furthest from the intake / exhaust passage 10b is the first stage filter 10g, and the filter 10g closest to the intake / exhaust passage 10b is the last stage filter 10g. Simultaneously, the intake end of each filter 10g is connected to the intake nozzle 10c through a corresponding intake passage 10h. An on / off actuator 10i is installed on the assembly housing 10a to switch the on / off state of the exhaust passage 10e and each intake passage 10h.

[0141] Therefore, when the on / off actuator 10i keeps the exhaust passage 10e open, all intake passages 10h are cut off by the on / off actuator 10i. At this time, the gas inside the air pump 9 is discharged to the outside through the exhaust nozzle 10d. When the on / off actuator 10i keeps any intake passage 10h open, the exhaust passage 10e and the other intake passages 10h are all cut off by the on / off actuator 10i. At this time, outside gas is drawn into the air pump 9 through the intake nozzle 10c and the open intake passage 10h. If the open intake passage 10h is the intake passage connected to the first-stage filter 10g... In channel 10h, the inhaled gas will be filtered sequentially through all filters 10g before being drawn into the air pump 9. If the unobstructed intake channel 10h is connected to the last filter 10g, the inhaled gas will be filtered sequentially through only the last filter 10g before being drawn into the air pump 9. If the unobstructed intake channel 10h is connected to one of the intermediate filters 10g, the inhaled gas will be filtered sequentially through the corresponding first filter 10g to the last filter 10g before being drawn into the air pump 9.

[0142] Therefore, the electronic nose's air intake mechanism can not only connect and disconnect the corresponding air intake channel 10h via the actuator 10i according to the type of gas being tested and the testing environment, allowing the gas being tested to be filtered sequentially through the corresponding filters 10g while being drawn into the air pump 9, removing interfering gases and impurities, thereby effectively improving the recognition accuracy of subsequent gas sensors and thus improving the detection accuracy of the electronic nose; it can also precisely pressurize the gas being tested according to the type of gas being tested and the testing requirements, meeting the needs of some specific gases that require testing under specific pressures, further improving the applicability and versatility of the electronic nose; it also has the function of self-testing and self-cleaning the filter elements of each level of the filters 10g; please refer to the following text for details.

[0143] Regarding the 10g filters at each level, the most versatile 10g filter is placed in the last stage, and the least versatile 10g filter is placed in the first stage. The versatility of the intermediate 10g filters increases progressively towards the last stage. The filter element of the 10g filter can be a physical filter element capable of removing solids, liquids, and aerosols, or a chemical adsorbent filter element capable of removing specific chemical gases, such as activated carbon.

[0144] It should be noted that the assembly housing 10a is equipped with a removable housing cover, which allows the filter element of the filter 10g to be replaced by removing the housing cover.

[0145] Please see Figure 4The on / off actuator 10i includes an actuator push rod 10i1 and an actuator motor 10i2 for controlling the reciprocating movement of the actuator push rod 10i1. The actuator push rod 10i1 has an exhaust port 10i11 and an intake port 10i12. When the actuator motor 10i2 connects the exhaust port 10i11 of the actuator push rod 10i1 to the exhaust channel 10e, the intake port 10i12 is not connected to any of the intake channels 10h. When the actuator motor 10i2 connects the intake port 10i12 of the actuator push rod 10i1 to any one of the intake channels 10h, the exhaust port 10i11 is not connected to the exhaust channel 10e. Furthermore, the actuator motor 10i2 can also cause the actuator push rod 10i1 to simultaneously block the exhaust channel 10e and all the intake channels 10h, thereby achieving precise control of the on / off state of the exhaust channel 10e and each intake channel 10h.

[0146] Please see Figures 2-5 The air pump 9 includes an air cylinder 9b with an air chamber 9b1. In this embodiment, the air cylinder 9b is preferably cylindrical. The inlet and outlet 9a are located at one end of the air chamber 9b1, and the inner end of the inlet and outlet 9a communicates with the air chamber 9b1. A piston 9c and a piston actuator 9d for moving the piston 9c closer to or away from the inlet and outlet 9a are installed in the air chamber 9b. A pressure balancing valve 9e communicating with the air chamber 9b1 is installed at the end of the air cylinder 9b away from the inlet and outlet 9a. Therefore, by controlling the movement of the piston 9c through the piston actuator 9d, the air chamber 9b1 can be inhaled and exhausted. Specifically, when the piston actuator 9d moves the piston 9c away from the inlet and outlet 9a, the outside gas enters the air chamber 9b1 sequentially through the filter assembly 10 and the inlet and outlet 9a; when the piston actuator 9d moves the piston 9c closer to the inlet and outlet 9a, the gas in the air chamber 9b1 is discharged sequentially through the inlet and outlet 9a and the filter assembly 10.

[0147] A motor mounting plate 9b2 is provided at the end of the air cylinder 9b furthest from the inlet / outlet port 9a. The piston actuator 9d is a piston drive motor mounted on the motor mounting plate 9b2. The motor shaft of this piston drive motor is an extended motor shaft, the outer end of which is rotatably mounted on the end of the air cylinder 9b near the inlet / outlet port 9a, thus ensuring the stability and reliability of the extended motor shaft's rotation. Simultaneously, external threads are formed on the outer circumferential surface of the extended motor shaft, and the piston 9c has an internally threaded hole that forms a lead screw and nut moving pair with the extended motor shaft. A translation guide rod 9f parallel to the extended motor shaft is also provided in the air chamber 9b1, and the piston 9c slides in conjunction with the translation guide rod 9f. Therefore, by rotating the extended motor shaft back and forth, the position of the piston 9c in the air chamber 9b1 can be precisely controlled.

[0148] Furthermore, by setting a pressure balancing valve 9e, the internal pressure of the air cylinder 9b can be balanced. Specifically, the vent hole 9b21 on the motor mounting plate 9b2 allows external gas to enter sequentially through the pressure balancing valve 9e and the vent hole 9b21 when the piston 9c is close to the inlet / outlet port 9a, and internal gas to be discharged sequentially through the pressure vent hole 9b21 and the balancing valve 9e when the piston 9c is away from the inlet / outlet port 9a.

[0149] Further, please see Figure 5 The piston 9c is equipped with a safety valve 9f that connects the two sides of the piston 9c. When the pressure in the air chamber 9b1 exceeds the limit, the gas in the air chamber 9b1 will be automatically discharged to protect the safety of the air cylinder 9b.

[0150] Please see Figure 4 and Figure 5 The pressure sensor 11 is installed at one end of the air chamber 9b1 where the air inlet and outlet 9a are provided. The air cylinder 9b is equipped with an air inlet and outlet control circuit board 12. The on / off actuator 10i, the piston actuator 9d and the pressure sensor 11 are all electrically connected to the air inlet and outlet control circuit board 12, thereby enabling the linkage control of various electrical components.

[0151] like Figures 6-14 As shown, the general gas detection assembly includes a detection box 1 and an intake splitter assembly 2 and an exhaust assembly 3 respectively disposed at both ends of the detection box 1. The exhaust nozzle 10d is connected to the intake end of the intake splitter assembly 2 through a pipeline.

[0152] The testing box 1 includes a central container 1a and multiple side mounting slots 1b, which are circumferentially distributed around the central mounting slot 1a. In this embodiment, each side mounting slot 1b is recessed into the circumferential side wall of the testing box 1 surrounding the central mounting slot 1a. Simultaneously, quick-change sealing plates 8 are installed on the outer wall of the testing box 1, each capable of being detachably covered at the opening of its corresponding side mounting slot 1b. Therefore, opening the quick-change sealing plate 8 exposes the corresponding side mounting slot 1b. Taking the overall structure of the testing box 1 as a cuboid as an example, the intake splitter assembly 2 and the exhaust assembly 3 are respectively located at both ends of the testing box 1. The remaining four circumferential side walls of the testing box 1 are all recessed to form side mounting slots 1b, which are covered by quick-change sealing plates 8. Thus, the testing box 1 contains one central container 1a and four side mounting slots 1b. By installing and removing the quick-change sealing plates 8, the equipment and components in the side mounting slots 1b can be easily operated, replaced, and maintained.

[0153] In this embodiment, a first gas sensor array 4 is installed in the central container 1a. The first gas sensor array 4 and the central container 1a together form a universal sensor detection chamber 1a1 that connects the intake split assembly 2 and the exhaust assembly 3. A second gas sensor array 5 is installed in each of the side mounting slots 1b. Each second gas sensor array 5 and the corresponding side mounting slot 1b together form a quick-change sensor detection chamber 1b1 that connects the intake split assembly 2 and the exhaust assembly 3.

[0154] Therefore, in this embodiment, the first gas sensor array 4, which is versatile and does not require frequent replacement, is installed in the central container 1a (replacing it would be cumbersome). The second gas sensor array 5, which requires adaptive adjustment for different detection targets, is installed in the quick-change sensor detection chamber 1b1 located around the general sensor detection chamber 1a1. By setting up a quick-change sealing plate 8 that is easy to install and remove, the second gas sensor array 5 in the side mounting slot 1b can be quickly replaced very conveniently. The operation is simple and convenient. Thus, by combining the types of the first gas sensor array 4 and each of the second gas sensor arrays 5 (the gas sensor arrays of the aforementioned electronic nose are composed of the first gas sensor array 4 and each of the second gas sensor arrays 5), a modular sensor array combination that can accurately analyze different target gases is formed, which is suitable for the detection of different targets and has excellent versatility. This makes the electronic nose a universal device that can be applied to different targets, greatly reducing the investment cost of the electronic nose and facilitating its application, promotion and popularization. Meanwhile, the type and parameters of the second gas sensor array 4 in two or more quick-change sensor detection chambers 1b1 can be set to be exactly the same, so that a single test can detect space 9b1, which is equivalent to the effect of multiple redundant tests. This not only meets the data acquisition requirements of the processing algorithm and obtains high-precision detection results, but also greatly improves the testing efficiency and the convenience of the detection operation.

[0155] Furthermore, the first gas sensor array 4 and each of the second gas sensor arrays 5 are located in their respective independent chambers to detect the gas without any interference between them, thereby further improving the detection accuracy. Moreover, the gas universal detection assembly and the electronic nose air intake mechanism adopt a split structure, and the structures of both parts are extremely compact. The electronic nose air intake mechanism with different air intake volumes can be replaced according to the detection needs of the gas being detected, meeting the application requirements of miniaturization and modularization, with high integration and excellent portability.

[0156] Different second gas sensor arrays 5 have different requirements for the detection space; a reasonable detection space volume can effectively improve detection accuracy. Therefore, please refer to... Figure 7 and Figure 11Each quick-change sensor detection chamber 1b1 is equipped with a volume adjustment component 6 for adjusting the volume of the quick-change sensor detection chamber 1b1. By setting the volume adjustment component 6, the volume of the quick-change sensor detection chamber 1b1 can be adaptively adjusted according to the different targets being detected and the current requirements of the second gas sensor array 5, thereby not only improving the detection accuracy but also further enhancing the versatility of the gas universal detection assembly.

[0157] In this embodiment, each quick-change sensor detection chamber 1b1 has a volume adjustment section with the same width at each position. The volume adjustment component 6 is installed in the volume adjustment section so that the volume adjustment component 6 can maintain good airtightness when adjusting the volume of the quick-change sensor detection chamber 1b1.

[0158] Specifically, each volume adjustment assembly 6 includes a slider mounting plate 6a installed at one end of the corresponding volume adjustment section along its length, and a volume adjustment slider 6b slidably mounted on the slider mounting plate 6a along the length of the volume adjustment section. The mounting end face and the outer edges on both sides in the width direction of the slider mounting plate 6a are fixedly connected to the cavity wall of the corresponding volume adjustment section without gaps. That is, the slider mounting plate 6a is fixedly installed at one end of the volume adjustment section along its length, and extends along the length of the volume adjustment section. Furthermore, there are no gaps between the mounting end (inner end) and the outer edges on both sides in the width direction of the slider mounting plate 6a and the cavity wall of the corresponding volume adjustment section. The volume adjustment slider 6b is slidably mounted on the outer end of the slider mounting plate 6a along its length direction, and its bottom surface and the outer edges on both sides in the width direction are slidably fitted to the cavity wall of the corresponding volume adjustment section without gaps.

[0159] Therefore, when it is necessary to reduce the volume of the quick-change sensor detection chamber 1b1, the volume adjustment slider 6b slides away from the slider mounting plate 6a (slides outward); when it is necessary to increase the volume of the quick-change sensor detection chamber 1b1, the volume adjustment slider 6b slides towards the slider mounting plate 6a (slides inward). This design not only allows for flexible adjustment of the volume of the quick-change sensor detection chamber 1b1 by sliding the volume adjustment slider 6b, but also ensures good airtightness.

[0160] Furthermore, the bottom surface of the volume adjustment slider 6b and the outer edges of both sides in the width direction are equipped with a sealing layer that is interference-fitted with the cavity wall of the corresponding volume adjustment section. This not only ensures good airtightness, but also makes it difficult for the volume adjustment slider 6b to slide under gas pressure (the volume adjustment slider 6b will only slide when a large force is applied by a person). This allows the volume of the sensor detection chamber 1b1 to remain stable after it has been filled, further improving the recognition accuracy.

[0161] Further, please see Figure 7 , Figure 10 and Figure 13 The volume adjustment section has recessed guide grooves 1b3 extending along its length on both sides of the groove wall in the width direction. The outer edges of both sides of the volume adjustment slider 6b in the width direction have sliding guide ribs 6b2 that are adapted to the corresponding guide grooves 1b3. Each sliding guide rib 6b2 is slidably embedded in its corresponding guide groove 1b3, effectively improving the stability and reliability of the volume adjustment slider 6b's forward and backward sliding. The outer wall of the sliding guide rib 6b2 also has a sealing layer that is interference-fitted with the corresponding guide groove 1b3, with the effect described above.

[0162] Furthermore, one or both sides of the guide slide 1b3 are provided with adjustment scales distributed along its length (not shown in the figure), so that the operator can accurately know the adjusted volume of the quick-change sensor detection chamber 1b1 according to the adjustment scale position of the volume adjustment slider 6b, which not only improves the convenience of operation, but also enhances the accuracy of volume adjustment.

[0163] Furthermore, a support boss 6b1 is formed on the bottom of the volume adjustment slider 6b away from the slider mounting plate 6a. The outer circumferential edge of the support boss 6b1 has a sealing layer that is interference-fitted with the cavity wall of the corresponding volume adjustment section. This not only ensures good airtightness, but also makes it difficult for the volume adjustment slider 6b to slide under gas pressure. This allows the volume of the sensor detection chamber 1b1 to remain stable after it has been filled. At the same time, it reduces the contact area between the volume adjustment slider 6b and the cavity wall of the volume adjustment section, thus avoiding jamming.

[0164] A detection groove is formed in the middle of the bottom of the side mounting groove 1b. At least two circuit board mounting seats 1b2 are provided at the bottom of the side mounting groove 1b surrounding the detection groove. The second gas sensor array 5 has the following two implementation methods:

[0165] Second gas sensor array 5, implementation method 1: Please refer to Figures 7-10 The second gas sensor array 5 includes a quick-change gas sensor circuit board 5a, a sealing gasket 5b, and a gasket pressure plate 5c. The quick-change gas sensor circuit board 5a is detachably mounted on all circuit board mounting bases 1b2 (typically using screws or clips). The sealing gasket 5b covers the opening of the detection groove, and the gasket pressure plate 5c presses the sealing gasket 5b firmly against the bottom of the detection groove. The sealing gasket 5b and the detection groove together form a quick-change sensor detection chamber 1b1. At least one dedicated gas sensor 5a1 is integrated on the side of the quick-change gas sensor circuit board 5a closest to the detection groove. The dedicated gas sensor 5a1 is a conventional gas sensor, such as a metal oxide gas sensor.

[0166] The sealing gasket 5b has sealing through holes 5b1 that are adapted to each dedicated gas sensor 5a1. The gasket pressure plate 5c is located on the side of the sealing gasket 5b away from the bottom of the side mounting groove 1b. The gasket pressure plate 5c is detachably connected to the bottom of the side mounting groove 1b (usually by screws or clips). The gasket pressure plate 5c has clearance through holes 5c1 that are adapted to each dedicated gas sensor 5a1. The diameter of each sealing through hole 5b1 is smaller than the diameter of the corresponding clearance through hole 5c1. Each dedicated gas sensor 5a1 passes through the corresponding clearance through hole 5c1 and sealing through hole 5b1 in sequence and then extends into the quick-change sensor detection chamber 1b1. Each dedicated gas sensor 5a1 is interference-fitted with the hole wall of the corresponding sealing through hole 5b1.

[0167] Therefore, by designing that the diameter of the sealing through-hole 5b1 is smaller than the diameter of the corresponding clearance through-hole 5c1 and that each dedicated gas sensor 5a1 is press-fitted with the wall of the corresponding sealing through-hole 5b1, the detection part of the dedicated gas sensor 5a1 can be inserted into the quick-change sensor detection chamber 1b1 while ensuring the airtightness of the quick-change sensor detection chamber 1b1.

[0168] Second gas sensor array 5, embodiment 2: Please refer to Figure 11 and Figure 12 The second gas sensor array 5 includes a gas detection thin-film sensor 5d and a thin-film sensor pressure plate 5e. The gas detection thin-film sensor 5d is a thin-film gas sensor, which is a gas detection instrument based on metal oxide semiconductor material and belongs to the category of semiconductor gas sensors. The gas detection thin-film sensor 5d covers the opening of the detection groove. The thin-film sensor pressure plate 5e is used to press the gas detection thin-film sensor 5d tightly against the bottom of the detection groove. The gas detection thin-film sensor 5d and the detection groove together form a quick-change sensor detection chamber 1b1. The thin-film sensor pressure plate 5e is located on the side of the gas detection thin-film sensor 5d away from the bottom of the side mounting groove 1b and is detachably connected to the bottom of the side mounting groove 1b (usually by screws or clips). The second gas sensor array 5 using this embodiment can ensure that the quick-change sensor detection chamber 1b1 has excellent airtightness. It should be noted that each gas detection thin-film sensor 5d usually integrates multiple gas sensors, and the types of gas sensors can be the same or different. The gas sensors on each gas detection thin-film sensor 5d constitute a gas sensor array.

[0169] Please see Figure 7 , Figure 11 and Figure 12The central container 1a has a sensor mounting port on the side near the exhaust assembly 3. The first gas sensor array 4 includes a universal gas sensor circuit board 4a that covers the sensor mounting port. The universal gas sensor circuit board 4a and the central container 1a together form a universal sensor detection chamber 1a1. At least one universal gas sensor 4b extending into the universal sensor detection chamber 1a1 is integrated on the side of the universal gas sensor circuit board 4a near the central container 1a. This not only ensures the reliable installation of the first gas sensor array 4, but also allows the first gas sensor array 4 to be replaced. However, the replacement steps are more complicated than those of the second gas sensor array 5.

[0170] Furthermore, the universal sensor detection chamber 1a1 is divided into a diffusion space 1a11 and a detection space 1a12 by a centrally located air distribution plate 1a2. The air distribution plate 1a2 has an array of small holes 1a21. The diffusion space 1a11 includes an expansion section 1a111 that gradually increases in size from the air inlet of the universal sensor detection chamber 1a1 and a stable section 1a112 that extends from the end of the expansion section 1a111 away from the air inlet to the air distribution plate 1a2. The expansion section 1a111 has a conical structure, and the stable section 1a112 has a cylindrical structure. The universal gas sensor 4b is located in the detection space 1a12, and the detection space 1a12 is connected to the air outlet of the universal sensor detection chamber 1a1. With this structure, the gas being tested, after entering the detection chamber 1a1 of the universal sensor, can first diffuse rapidly in the expansion section 1a111 of the conical structure, then achieve a relatively stable state in the stable section 1a112 of the cylindrical structure, and then enter the detection space 1a12 through the array of small holes 1a21 on the air distribution plate 1a2. This results in extremely high uniformity of the gas being tested filling the detection space 1a12, which can effectively improve the detection accuracy of the universal gas sensor 4b.

[0171] Please see Figures 6-12 The air intake splitter assembly 2 includes a front housing 2a connected to the detection housing 1, an air intake connector 2b installed on the side of the front housing 2a away from the detection housing 1, and a gas distributor 2c installed on the side of the front housing 2a close to the detection housing 1. The front housing 2a has an air intake channel 2a1 connecting the air intake connector 2b and the gas distributor 2c. The gas distributor 2c integrates exhaust ports 2c1 corresponding to the universal sensor detection chamber 1a1 and the quick-change sensor detection chamber 1b1, respectively. Each exhaust port 2c1 is connected to the air intake port of the corresponding universal sensor detection chamber 1a1 or the air intake port of the quick-change sensor detection chamber 1b1 through a pipeline. Therefore, the gas to be tested enters from the air intake connector 2b and is then guided into the universal sensor detection chamber 1a1 and each quick-change sensor detection chamber 1b1 by the gas distributor 2c and the corresponding pipeline, which is simple and reliable.

[0172] The exhaust assembly 3 includes a rear housing 3a connected to the detection box 1 and several exhaust connectors 3b installed on the side of the rear housing 3a away from the detection box 1. Each exhaust connector 3b is connected to the corresponding universal sensor detection chamber 1a1 or quick-change sensor detection chamber 1b1 through an exhaust channel 3c provided in the rear housing 3a. Therefore, the gas to be detected in the universal sensor detection chamber 1a1 and each quick-change sensor detection chamber 1b1 is guided to the exhaust connector 3b by the corresponding exhaust channel 3c and then discharged by the exhaust connector 3b, which is simple and reliable.

[0173] Meanwhile, a detection control circuit board 7 is installed in the rear housing 3a. The detection control circuit board 7 is electrically connected to the first gas sensor array 4 and each of the second gas sensor arrays 5. The detection control circuit board 7 can control the first gas sensor array 4 and each of the second gas sensor arrays 5 to detect the gas to be tested, process the detection information of the first gas sensor array 4 and each of the second gas sensor arrays 5, and transmit the processed information to the outside.

[0174] In this embodiment, the test box 1, the intake split assembly 2, and the exhaust assembly 3 all adopt a split structure to facilitate the maintenance and replacement of internal equipment and electrical components. Specifically, the test box 1, the intake split assembly 2, and the exhaust assembly 3 are all plate structures connected by dedicated connectors.

[0175] Example 5:

[0176] A control method for an electronic nose according to Embodiment 4 is performed according to the following steps:

[0177] A1. Check the condition of the 10g filter cartridges in each stage of the electronic nose air intake mechanism to ensure they are functioning properly, following these steps:

[0178] A11, Air intake nozzle 10c is connected to air;

[0179] A12. The on / off actuator 10i keeps the air intake passage 10h, which is connected to the first-stage filter 10g, unobstructed until the air pump 9 is filled with air.

[0180] A13. The on / off actuator 10i cuts off the exhaust passage 10e and all intake passages 10h;

[0181] A14. The air pump 9 compresses the internal air until the pressure sensor 11 detects that the air pressure inside the air pump 9 has reached the self-test set value.

[0182] A15. The on / off actuator 10i keeps the air intake channel 10h connected to the first-stage filter 10g unobstructed until the internal air pressure of the air pump 9 returns to normal pressure, and determines whether the time to return to normal pressure is greater than the time threshold: if yes, it means that the filter element of filter 10g is not in good condition, and proceed to step A2; if no, it means that the filter element of filter 10g is in good condition, and proceed to step A3.

[0183] A2. Clean the 10g filter cartridges of each stage of the electronic nose air intake mechanism according to the following steps:

[0184] A21. The on / off actuator 10i connects to the intake passage 10h, which is connected to the last stage filter 10g;

[0185] A22. After the air pump 9 is filled with air, the air pump 9 quickly empties the air inside and determines whether the number of repetitions of step A22 is greater than the repetition threshold: if yes, after the air pump 9 is filled with air, proceed to step A23; if no, repeat step A22. The purpose of the air pump 9 quickly emptying the air inside is to be able to quickly reverse the discharge of the residue on the filter element of the last stage filter 10g through the air pressure impact force.

[0186] A23. After the on / off actuator 10i cuts off the exhaust passage 10e and all intake passages 10h, the air pump 9 compresses the internal air until the pressure sensor 11 detects that the internal air pressure of the air pump 9 reaches the self-test set value.

[0187] A24. The on / off actuator 10i connects to the air intake channel 10h connected to the last stage filter 10g until the internal air pressure of the air pump 9 returns to normal pressure. It is then determined whether the time to return to normal pressure is greater than the time threshold. If yes, it means that the filter element of the last stage filter 10g cannot be restored to a good state through self-cleaning. Therefore, the filter element of the last stage filter 10g is replaced first. Then, after the on / off actuator 10i connects to the air intake channel 10h connected to the previous stage filter 10g, the process proceeds to step A25. If no, it means that the filter element of the last stage filter 10g has been restored to a good state through self-cleaning. After the on / off actuator 10i connects to the air intake channel 10h connected to the previous stage filter 10g, the process proceeds to step A25.

[0188] A25. After the air pump 9 is filled with air, the air pump 9 quickly empties the air inside and determines whether the number of repetitions of step A25 is greater than the repetition threshold: if yes, after the air pump 9 is filled with air, proceed to step A26; if no, repeat step A25. The purpose of the air pump 9 quickly emptying the air inside is to be able to quickly expel the residue on the filter element of the primary filter 10g, which is closest to the unobstructed air intake channel 10h, through the air pressure impact force.

[0189] A26. After the on / off actuator 10i cuts off the exhaust passage 10e and all intake passages 10h, the air pump 9 compresses the internal air until the pressure sensor 11 detects that the internal air pressure of the air pump 9 reaches the self-test set value.

[0190] A27. The on / off actuator 10i connects the previously connected air intake channel 10h until the internal air pressure of the air pump 9 returns to normal pressure, and determines whether the time to return to normal pressure is greater than the time threshold: if yes, it means that the filter element of the first-stage filter 10g closest to the unobstructed air intake channel 10h cannot be restored to a good state through self-cleaning, so after replacing the filter element of the first-stage filter 10g closest to the unobstructed air intake channel 10h, proceed to step A28; if no, it means that the filter element of the first-stage filter 10g closest to the unobstructed air intake channel 10h has been restored to a good state through self-cleaning, and proceed to step A28.

[0191] A28. Determine whether the primary filter 10g closest to the unobstructed air intake channel 10h is the first-stage filter 10g: If yes, it means that the filter elements of all filters 10g are in good condition. After the air pump 9 empties the internal air, proceed to step A3. If no, after the air pump 9 empties the internal air, the on / off actuator 10i connects the air intake channel 10h connected to the previous primary filter 10g, and then returns to step A25.

[0192] A3. To test the gas, follow these steps:

[0193] A31. Connect the air inlet nozzle 10c to the gas being tested;

[0194] A32. Based on the type of the gas being tested, determine whether the combination of the second gas sensor array 5 installed in the mounting slots 1b on each side meets the detection requirements: if yes, proceed to step A33; if no, replace the combination of the second gas sensor array 5 corresponding to the gas being tested, and then proceed to step A33; thereby, through the type combination of the first gas sensor array 4 and each second gas sensor array 5, a modular sensor array capable of accurately analyzing the gas being tested is formed.

[0195] A33. Based on the type of gas being tested, determine how many stages of filters 10g are needed for filtration. Then, switch the actuator 10i on and off to keep one of the air intake channels 10h corresponding to the gas being tested unobstructed until the air pump 9 is filled with the gas being tested.

[0196] A34. Based on the type of gas being tested, determine whether the gas inside the compressed air pump 9 is needed: if yes, proceed to step A35; otherwise, proceed to step A37.

[0197] A35, the on / off actuator 10i cuts off the exhaust passage 10e and all intake passages 10h;

[0198] A36. Air pump 9 compresses the internal gas to be tested until the volume of the gas to be tested is compressed to the detection set value required for its detection.

[0199] A37. The on / off actuator 10i keeps the exhaust passage 10e unobstructed until the air pump 9 empties the gas being tested. During this process, the first gas sensor array 4 and each of the second gas sensor arrays 5 of the gas universal detection assembly detect the gas being tested.

[0200] Through the above control method, the electronic nose logic can simply and reasonably realize the functions of self-inspection and self-cleaning of the filter elements of each level of filter, as well as the reasonable selection of the second gas sensor array combination and adjustment of the detection pressure of the gas being tested according to the type of gas being tested.

[0201] Furthermore, step A32 also includes adjusting the volume of the corresponding quick-change sensor detection chamber 1b1 according to the type of gas being detected using each volume adjustment component 6. This allows the volume of the quick-change sensor detection chamber 1b1 to be adaptively adjusted according to the different targets being detected and the current requirements of the second gas sensor array 5, thereby not only improving the detection accuracy but also further enhancing the versatility of the gas universal detection assembly.

[0202] Furthermore, the control method for the electronic nose also includes step A4, which involves self-cleaning the interior of the universal gas detection assembly and the electronic nose air intake mechanism, performed according to the following steps:

[0203] A41, Air intake nozzle 10c is connected to air;

[0204] A42. The on / off actuator 10i keeps the intake passage 10h, which is connected to the last stage filter 10g, unobstructed.

[0205] A43. After the air pump 9 is filled with air, the air pump 9 empties the air inside and determines whether the response values ​​of all the first gas sensor array 4 and the second gas sensor array 5 are the reference values: if yes, it means that the self-cleaning of the air pump 9 is completed and the machine is stopped; if no, it means that there is still residual gas to be detected in the detection space 9b1 and step A43 is repeated.

[0206] By adding step A4, the internal parts of the gas universal detection assembly and the electronic nose air intake mechanism can be self-cleaned using air, avoiding interference from residual gas from the previous test on the next test, and effectively improving the detection accuracy.

[0207] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention. Those skilled in the art, under the guidance of the present invention, can make various similar representations without departing from the spirit and claims of the present invention, and such modifications all fall within the protection scope of the present invention.

Claims

1. A gas classification method based on pooling aggregation and joint attention-enhanced LSTM, characterized in that, Follow these steps: S1. The electronic nose inhales the target sample. The gas sensor array of the electronic nose collects the gas response data of the target sample and converts the gas response data into a continuous analog signal sequence. S2. The pooling aggregation module uses a shared global average pooling layer to perform global average pooling on the continuous analog signal sequence output by each gas sensor array. The pooling aggregation module includes a time pooling module and a channel pooling module. The time pooling module is used to perform global pooling on the time dimension of all gas sensors, and the channel pooling module is used to perform global pooling on the channel dimension within a single gas sensor array. S3. Input the feature vectors output by the pooling aggregation module after performing global average pooling operation into the joint attention module composed of the channel attention submodule and the temporal attention submodule respectively. Specifically, the channel attention submodule uses learnable parameters to construct an adaptive channel importance weight matrix to perform a channel importance weighting operation on the feature vectors corresponding to each gas sensor array output by the channel pooling module, obtaining the channel feature vectors corresponding to each gas sensor array. Then, the channel feature vectors corresponding to each gas sensor array are summed to obtain the channel feature vector integrating the output signals of all gas sensor arrays. The time attention submodule uses learnable parameters to construct an adaptive time step importance weight vector to perform a time step differential weighting operation on the different time step features of the feature vector output by the time pooling module, obtaining the time-series feature vector integrating the output signals of all gas sensor arrays. The channel feature vector and the time-series feature vector are summed to generate a fused time-channel joint feature vector; S4. Input the time-channel joint feature vector into a single-layer lightweight long short-term memory network module to perform temporal feature modeling, generate the final output feature vector, and classify the target sample according to the final output feature vector to obtain the classification result of the target sample.

2. The gas classification method according to claim 1, characterized in that, In step S2, the continuous analog signal sequence output by each gas sensor array is input into the channel pooling module. The calculation formula for performing one-dimensional global average pooling operation on each gas sensor array in the channel dimension is as follows: ； In the above formula, Indicates the corresponding to the first The first gas sensor array Features obtained from a gas sensor through global average pooling. Indicates the first The data from the gas sensor array of the first The first gas sensor Data at each time step, This represents the time step length of a continuous analog signal sequence from a gas sensor array. Indicates the number of gas sensor arrays; The channel feature vector corresponding to each gas sensor array is obtained through the channel pooling module. ; The continuous analog signal sequences output by each gas sensor array are input into the time pooling module. The calculation formula for performing a one-dimensional global average pooling operation on all gas sensor arrays in the time dimension is as follows: ； In the above formula, Indicates the corresponding to the first Features obtained at each time step through global average pooling Indicates the first The data from the gas sensor array of the first The first gas sensor Data at each time step, This indicates the number of gas sensors in the gas sensor array; The temporal feature vectors of all gas sensor arrays are obtained through the time pooling module. .

3. The gas classification method according to claim 2, characterized in that, In step S3, the channel attention submodule uses learnable parameters to construct an adaptive channel importance weight matrix to perform a channel importance weighting operation on the channel feature vectors corresponding to each gas sensor array. The calculation formula is as follows: ； In the above formula, Indicates the corresponding to the first Channel feature vectors of a gas sensor array weighted by channel importance; Indicates the corresponding to the first A learnable channel importance weight matrix for a gas sensor array, with elements ranging from [0,1]. The channel feature vectors corresponding to each gas sensor array, after being weighted according to channel importance, are summed to generate a channel feature vector that integrates the output signals of all gas sensor arrays. The calculation formula is as follows: ； In the above formula, This represents the channel feature vector that integrates the output signals of all gas sensor arrays; The formula for the time-step differential weighting operation of the temporal feature vector output by the time-pooling module on the temporal attention submodule is as follows: ； In the above formula, This represents the time-series feature vector after integrating the output signals of all sensor arrays after time-step differential weighting. This represents the learnable time-step importance weight vector. It represents the Hadamardi (or Hadama) stack; The temporal feature vector output by the temporal attention submodule and the channel feature vector output by the channel attention submodule are summed to obtain the temporal-channel joint feature vector, and the calculation formula is as follows: ； In the above formula, This represents the time-channel joint feature vector.

4. The gas classification method according to claim 3, characterized in that, In step S4, the formula for the single-layer lightweight long short-term memory network module to perform temporal feature modeling on the time-channel joint feature vector is as follows: ； In the above formula, Indicates the first The final output feature vector at each time step, The first eigenvector in the time-channel joint feature vector Elements at each time step , , and These represent the output gate weight matrix, forget gate weight matrix, input gate weight matrix, and candidate cell state weight matrix, respectively. , , and Let represent the output gate bias vector, forget gate bias vector, input gate bias vector, and candidate cell state bias vector, respectively. This represents the sigmoid activation function. Indicates the first The hidden state at each time step. Indicates the first Output features of candidate cell states at each time step This represents the hyperbolic tangent function.

5. An electronic device, characterized in that, It includes at least one processor and a memory communicatively connected to the at least one processor, the memory storing a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the gas classification method according to any one of claims 1 to 4.

6. A computer-readable medium, characterized in that, The computer-readable medium stores computer instructions for causing the at least one processor of claim 5 to perform the gas sorting method of any one of claims 1 to 4.

7. An electronic nose according to any one of claims 1 to 4, characterized in that, The electronic nose includes a universal gas detection assembly and an electronic nose air intake mechanism; The electronic nose air intake mechanism includes an air pump with inlet and outlet ports. A filter assembly is connected to the outer end of the inlet and outlet ports. A pressure sensor is installed inside the air pump. The filter assembly includes an assembly housing, which integrates an inlet nozzle, an outlet nozzle, an inlet / outlet duct communicating with the inlet and outlet ports, an exhaust channel connecting the inlet / outlet duct and the outlet nozzle, and a filter channel communicating with the inlet / outlet duct. At least two stages of filters are sequentially arranged on the filter channel. The inlet end of each filter is connected to the inlet nozzle through a corresponding inlet channel. An on / off actuator is installed on the assembly housing to switch the on / off state of the exhaust channel and each inlet channel. When the on / off actuator keeps the exhaust channel open, all inlet channels are cut off by the on / off actuator; when the on / off actuator keeps any inlet channel open, the exhaust channel and the remaining inlet channels are cut off by the on / off actuator. The universal gas detection assembly includes a detection box and an intake splitter assembly and an exhaust assembly respectively disposed at both ends of the detection box. The exhaust nozzle is connected to the intake end of the intake splitter assembly through a pipeline. The detection box is provided with a central container and multiple side mounting slots distributed circumferentially around the central mounting slot. A first gas sensor array is installed in the central container. The first gas sensor array and the central container together form a universal sensor detection chamber that connects the intake splitter assembly and the exhaust assembly. A second gas sensor array is installed in each of the side mounting slots. Each second gas sensor array and the corresponding side mounting slot together form a quick-change sensor detection chamber that connects the intake splitter assembly and the exhaust assembly. The outer wall of the detection box is equipped with quick-change sealing plates that can be detachably covered at the opening of the corresponding side mounting slot.

8. The electronic nose according to claim 7, characterized in that, Each of the quick-change sensor detection chambers has a volume adjustment section of the same width at each location. Each volume adjustment section is equipped with a volume adjustment component for adjusting the volume of the quick-change sensor detection chamber. Each volume adjustment component includes a slider mounting plate installed at one end of the corresponding volume adjustment section along its length and a volume adjustment slider that is slidably mounted on the slider mounting plate along the length of the volume adjustment section. The mounting end face and the outer edges of both sides of the slider mounting plate in the width direction are fixedly connected to the cavity wall of the corresponding volume adjustment section without gaps. The bottom surface and the outer edges of both sides of the volume adjustment slider in the width direction are slidably fitted to the cavity wall of the corresponding volume adjustment section without gaps.

9. The electronic nose according to claim 7, characterized in that, A detection groove is formed in the middle of the bottom of the side mounting groove. The second gas sensor array includes a gas detection thin film sensor covering the opening of the detection groove and a thin film sensor pressure plate for pressing the gas detection thin film sensor against the bottom of the detection groove. The gas detection thin film sensor and the detection groove together form the quick-change sensor detection chamber. The thin film sensor pressure plate is located on the side of the volume detection thin film sensor away from the bottom of the side mounting groove and is detachably connected to the bottom of the side mounting groove. The central container has a sensor mounting port on the side near the exhaust assembly. The first gas sensor array includes a universal gas sensor circuit board that covers the sensor mounting port. The universal gas sensor circuit board and the central container together form the universal sensor detection chamber. At least one universal gas sensor extending into the universal sensor detection chamber is integrated on the side of the universal gas sensor circuit board near the central container. The universal sensor detection chamber is divided into a diffusion space and a detection space by a centrally located air distribution plate. The air distribution plate has an array of small holes. The diffusion space includes an expansion section that gradually increases in size from the air inlet of the universal sensor detection chamber and a stable section that extends from the end of the expansion section away from the air inlet to the air distribution plate. The expansion section has a conical structure, and the stable section has a cylindrical structure. The universal gas sensor is located in the detection space, and the detection space is connected to the air outlet of the universal sensor detection chamber.

10. A control method for an electronic nose according to any one of claims 7 to 9, characterized in that, The control method is carried out according to the following steps: A1. Check the condition of the filter elements of each stage of the electronic nose air intake mechanism to ensure they are functioning properly, following these steps: A11. Air intake nozzle connects to air; A12. The on / off actuator keeps the air intake passage connected to the first-stage filter open until the air pump is full of air. A13. The on / off actuator cuts off the exhaust passage and all intake passages; A14. The air pump compresses the internal air until the pressure sensor detects that the internal air pressure of the air pump has reached the self-test set value. A15. The on / off actuator keeps the air intake passage connected to the first-stage filter open until the internal air pressure of the air pump returns to normal pressure, and determines whether the time to return to normal pressure is greater than the time threshold: if yes, proceed to step A2; if no, proceed to step A3. A2. Clean the filter elements of each stage of the electronic nose air intake mechanism according to the following steps: A21. The on / off actuator connects the air intake passage that is connected to the last stage filter; A22. After the air pump is filled with air, the air pump quickly empties the air inside and determines whether the number of repetitions of step A22 is greater than the repetition threshold: if yes, after the air pump is filled with air, proceed to step A23; otherwise, repeat step A22. A23. After the on / off actuator cuts off the exhaust passage and all intake passages, the air pump compresses the internal air until the pressure sensor detects that the internal air pressure of the air pump has reached the self-test set value. A24. Connect the on / off actuator to the air intake channel connected to the last stage filter until the internal air pressure of the air pump returns to normal pressure, and determine whether the time to return to normal pressure is greater than the time threshold: if yes, replace the filter element of the last stage filter first, and then connect the on / off actuator to the air intake channel connected to the previous stage filter, and proceed to step A25; if no, after connecting the on / off actuator to the air intake channel connected to the previous stage filter, proceed to step A25. A25. After the air pump is filled with air, the air pump quickly empties the air inside and determines whether the number of repetitions of step A25 is greater than the repetition threshold: if yes, after the air pump is filled with air, proceed to step A26; if no, repeat step A25. A26. After the on / off actuator cuts off the exhaust passage and all intake passages, the air pump compresses the internal air until the pressure sensor detects that the internal air pressure of the air pump has reached the self-test set value. A27. The on / off actuator connects the previously connected air intake channel until the internal air pressure of the air pump returns to normal pressure, and determines whether the time to return to normal pressure is greater than the time threshold: if yes, replace the filter element of the first-stage filter closest to the unobstructed air intake channel and proceed to step A28; otherwise, proceed to step A28. A28. Determine whether the primary filter closest to the unobstructed air intake channel is the first-stage filter: If yes, after the air pump empties the internal air, proceed to step A3; if no, after the air pump empties the internal air, the on / off actuator connects the air intake channel connected to the previous primary filter, and then returns to step A25. A3. To test the gas, follow these steps: A31. Connect the air inlet nozzle to the gas being tested; A32. Based on the type of gas being tested, determine whether the combination of the second gas sensor array installed in the mounting slots on each side meets the testing requirements: if yes, proceed to step A33; if no, replace the combination of the second gas sensor array corresponding to the gas being tested and proceed to step A33. A33. Depending on the type of gas being tested, the on / off actuator keeps one of the air intake channels corresponding to the gas being tested open until the air pump is filled with the gas being tested. A34. Based on the type of gas being tested, determine whether the gas inside the compressed air pump is needed: if yes, proceed to step A35; otherwise, proceed to step A37. A35. The on / off actuator cuts off the exhaust passage and all intake passages; A36. The air pump compresses the gas being tested inside until the volume of the gas being tested is compressed to the detection set value required for its detection. A37. The on / off actuator keeps the exhaust passage open until the air pump empties the gas being tested inside. During this process, all the first gas sensor arrays and the second gas sensor arrays detect the gas being tested.

Images