A complex atmosphere gas detection method and electronic nose based on GC-MS

Abstract

The application discloses a GC-MS-based complex atmosphere gas detection method and electronic nose; wherein the detection method comprises: using GC-MS to quantitatively analyze the required detection environment gas containing target gas; configuring multiple groups of mixed standard gas, screening multiple gas-sensitive sensor arrays through each group of mixed standard gas to obtain a detection array; each sensor in the gas-sensitive sensor array can produce differential response to the target gas; associating the response signal output by the detection array in the environment of each mixed standard gas with the concentration of the corresponding target gas to form sample data; training a concentration detection model according to the sample data and a support vector machine algorithm; placing the detection array in the environment of the gas to be detected, collecting real-time response signals and inputting the real-time response signals into the concentration detection model to obtain the detection concentration of the target gas. The detection method can avoid the influence of similar interference gas during the detection of the concentration of the target gas, and the accuracy and stability of the detection result are higher.

Description

Technical Field

[0001] This invention belongs to the field of gas detection technology, and specifically relates to a method for detecting gases in complex atmospheres based on GC-MS and an electronic nose. Background Technology

[0002] In various fields such as industrial production, environmental monitoring, safety protection, and healthcare, the rapid and accurate detection of target gases is of paramount practical significance. The detection results directly impact the scientific basis of decision-making in critical scenarios such as production safety, environmental quality assessment, and human health protection. Currently, portable detectors have become the mainstream field detection equipment for target gas detection due to their ease of operation and portability. The core detection principle of these devices relies primarily on gas sensors, which achieve quantitative or qualitative detection of gas content through the specific response of the gas sensor to the target gas.

[0003] However, in practical applications, the on-site environment is often complex and variable, with frequent fluctuations in temperature and humidity and the coexistence of multiple interfering gases. When other gases with similar properties to the target gas are present in the detection environment, severe cross-interference occurs, causing the sensor to fail to accurately identify the target gas. This results in significant deviations in the detection results, making it difficult to meet the requirements of high-precision detection scenarios and greatly limiting the reliable application of portable detectors in complex environments. Summary of the Invention

[0004] To address the aforementioned problems, this invention discloses a method for detecting gases in complex atmospheres based on GC-MS and an electronic nose, in order to overcome or at least partially solve the aforementioned problems.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] This invention discloses a method for detecting gases in complex atmospheres based on GC-MS, the detection method comprising: GC-MS is used to detect the ambient gas containing the target gas, obtain the main gas components, and other gases other than the target gas in the main gas components are regarded as interfering gases. Multiple sets of mixed standard gases with different target gas concentrations and different interfering gas concentration gradients are configured. Multiple gas-sensitive sensor arrays are screened using the mixed standard gases. The gas-sensitive sensor array that produces the most differentiated response to changes in the target gas concentration and the interfering gas concentration is selected as the detection array. Each sensor in the gas-sensitive sensor array can produce differentiated responses to the target gas. The response signal output by the detection array under each of the mixed standard gases is correlated with the concentration of the corresponding target gas to form sample data; Based on the sample data and the support vector machine algorithm, a concentration detection model that can reflect the relationship between the response signal output by the detection array and the concentration of the target gas is trained. The detection array is placed in the gas environment to be tested, real-time response signals are collected and input into the concentration detection model to obtain the detection concentration of the target gas in the gas environment to be tested.

[0007] Further, the step of correlating the response signal output by the detection array in each of the mixed standard gases with the concentration of the corresponding target gas to form sample data includes: The mixed standard gas is detected using the detection array, and a continuous sequence of electrical signals is output. Feature parameters are extracted from the electrical signal sequence output by the detection array, and the feature parameters are combined according to the order of each gas sensor in the detection array to form a multidimensional feature vector; The multidimensional feature vector is correlated with the corresponding concentration of the target gas to form sample data.

[0008] Furthermore, the execution of training a concentration detection model based on the sample data and the support vector machine algorithm to reflect the relationship between the response signal output by the detection array and the concentration of the target gas includes: The multidimensional feature vectors corresponding to the mixed standard gases in each group are used as input samples, and the concentrations of the corresponding target gases are used as output labels. The optimal classification hyperplane is constructed in the feature space using the support vector machine algorithm, so that samples with different target gas concentrations can achieve maximum separation in the high-dimensional feature space, thereby obtaining the concentration detection model.

[0009] Furthermore, the characteristic parameters include response amplitude, response time, and curve integral value.

[0010] Furthermore, the detection method further includes: Record historical testing data generated under normal testing conditions to form a standard database; The real-time feature parameters acquired by the current detection array are compared with the historical feature parameters in the standard database. When the feature parameters output by a gas sensor in the detection array continuously deviate from its historical normal distribution range, the gas sensor is determined to be an abnormal sensor. Delete the feature parameters corresponding to the abnormal sensor and perform data compensation for the missing feature parameters.

[0011] Furthermore, the process of deleting the feature parameters corresponding to the abnormal sensor and performing data compensation for the missing feature parameters includes: The feature parameters corresponding to the abnormal sensor are removed from the multidimensional feature vector. Based on the feature parameters generated by the other gas sensors in the detection array, an interpolation estimation method is used to perform feature compensation on the removal position of the multidimensional feature vector.

[0012] Furthermore, the detection method further includes: The detection array is placed in a standard target gas environment of known concentration, and the baseline signal and characteristic parameters generated by the detection array are collected, and the corresponding calibration coefficients are generated accordingly. During the detection process, the calibration coefficients are used to perform baseline correction, sensitivity compensation, or scale normalization on the real-time acquired sensor signals to eliminate the influence of environmental changes and sensor drift on the detection results.

[0013] Another aspect of the present invention discloses an electronic nose, the electronic nose comprising: The gas-sensitive sensor array is used to detect the gas to be measured and output a response signal. Each sensor in the gas-sensitive sensor array can produce a differentiated response to the target gas. The signal acquisition module is used to acquire the response signal and perform preprocessing. The storage device is used to store the executable instructions of the processor and the concentration detection model; The processor is electrically connected to the gas sensor array, the signal acquisition module, and the memory, respectively, and is used to acquire the response signal output by the gas sensor array and input the response signal into the concentration detection model to output the concentration detection result of the target gas, so that the electronic nose realizes the complex atmosphere gas detection method based on GC-MS described above.

[0014] The advantages and beneficial effects of this invention are: In the gas detection method of this invention, GC-MS provides a real and reliable reference for gas components. Combined with the multi-dimensional feature capture capability of the gas sensor array, and then a precise mapping model is established through the support vector machine algorithm, the influence of similar interfering gases is eliminated in principle. This makes the accuracy and stability of the detection results far exceed those of traditional single sensor solutions, fully meeting the stringent requirements of high-precision detection scenarios such as industrial production and medical health, and filling the precision gap of portable detection equipment in complex atmospheres. Attached Figure Description

[0015] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a flowchart illustrating the implementation steps of a complex atmosphere gas detection method based on GC-MS in one embodiment of the present invention. Figure 2 This is a schematic diagram of the electronic nose in one embodiment of the present invention. Detailed Implementation

[0016] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0017] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or sets.

[0018] To keep the drawings concise, only the parts relevant to the invention are shown schematically in each figure, and they do not represent the actual structure of the product. Furthermore, for ease of understanding, in some figures, components with the same structure or function are shown only schematically, or only one is labeled. In this document, "a" not only means "only one," but can also mean "more than one."

[0019] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the specific implementation methods of the present invention will be described below with reference to the accompanying drawings. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings and other implementation methods can be obtained based on these drawings without any creative effort.

[0021] One embodiment of the present invention provides a method for detecting gases in complex atmospheres based on GC-MS, such as... Figure 1 As shown, the detection method includes: Step 01: Use GC-MS to detect the ambient gases containing the target gas, obtain the main gas components, and identify other gases besides the target gas as interfering gases. The target gas is determined based on specific detection needs and could be formaldehyde in a newly renovated home, hydrogen in a lithium battery thermal runaway environment, or ethylene released when fruit ripens.

[0022] Understandably, relying on the high-precision qualitative and quantitative capabilities of GC-MS, it is possible to accurately distinguish target gases from various complex interfering gases in the test environment, clarify the gas composition and concentration distribution, and solve the problem of mixed gas types and difficult-to-define components in complex atmospheres. This provides a reliable reference for gas composition in subsequent sensor selection and model building. It should be noted that in some specific detection scenarios, when the target gas is unclear, it is necessary to first determine the target gas by consulting relevant literature and GC-MS detection. For example, when determining fruit ripeness, it is necessary to first identify the gas affecting fruit ripeness (i.e., the target gas). This can be determined by consulting relevant literature and GC-MS detection, identifying ethylene gas as the target gas. Similarly, when determining whether a lithium battery has experienced thermal runaway, it is necessary to first identify the characteristic gas produced by lithium battery thermal runaway (i.e., the target gas). This can be determined by consulting relevant literature and GC-MS detection, identifying hydrogen gas produced by lithium battery thermal runaway as the target gas. The target gas can be one gas or multiple gases.

[0023] The specific steps for step 01 are as follows: Step 011: After identifying the target gas, use an inert gas sampling bag or stainless steel sampling container to conduct multiple sampling points in a representative area of ​​the desired testing environment, with the sampling volume controlled between 50 mL and 500 mL. For example, if the target gas is determined to be ethylene, then collect the gas released when fruit ripens.

[0024] Step 012: Start the GC-MS equipment to detect the sample, obtain the total ion chromatogram of the gas sample and the mass spectrum corresponding to each chromatographic peak, and then determine the specific gas types and the content of each type of gas in the environmental gas to be detected.

[0025] Step 013: Screen out the main gas components in the ambient gas to be detected, and treat other main gas components besides the target gas as interfering gases. For example, screen out gases with a percentage content of less than 1%, and treat the remaining gases as the main component gases.

[0026] It is understandable that if the content of a certain gas is very small, its interference with the detection of the target gas by the gas sensor is also very small and can be ignored. In order to reduce the workload, the gas with a small content is screened out.

[0027] Step 02 involves configuring multiple sets of mixed standard gases with different target gas concentrations and different interfering gas concentration gradients. These mixed standard gases are used to screen multiple gas-sensitive sensor arrays, selecting the array that produces the most differentiated response to changes in target gas and interfering gas concentrations as the detection array. Each sensor in the detection array must produce a differentiated response to the target gas. Specifically, each gas-sensitive sensor array comprises multiple gas-sensitive sensors arranged in an array, and each sensor must produce a differentiated response to the target gas; that is, the resistance, voltage, and current changes in amplitude / rate / sensitivity of each sensor are different. It should be noted that the number and arrangement of gas-sensitive sensors are consistent across all gas-sensitive sensor arrays.

[0028] Understandably, by using a multi-concentration gradient mixture of standard gases to select the best sensor array, a detection array with high sensitivity to the target gas and strong resistance to interfering gases is chosen. Furthermore, each sensor within the array can produce a differentiated response to the target gas, effectively avoiding signal superposition and signal offset issues caused by interfering gases in complex atmospheres, thus enhancing gas identification and discrimination capabilities at the hardware level. Moreover, by having multiple sensors capable of producing differentiated responses to the target gas collaboratively acquire response signals, multi-dimensional information about the characteristic gas can be comprehensively captured, reducing the impact of environmental interference (such as temperature, humidity, and background gases) on a single sensor.

[0029] Preferably, the gas sensor array is a heterogeneous gas sensor array. A heterogeneous gas sensor array is a composite detection system that integrates multiple gas sensor units of different types, principles, or sensitive materials onto the same substrate. It consists of at least three gas sensors with different sensing mechanisms; for example, a heterogeneous gas sensor array may include MOX metal oxide sensors, electrochemical sensors, and catalytic combustion sensors. Heterogeneous gas sensor arrays can further enhance "multi-dimensional sensing capabilities," simultaneously capturing the combined response of the target gas and interfering gases in a mixed standard gas.

[0030] The specific steps for step 02 are as follows: Step 021: Set multiple concentration gradients for the target gas and each interfering gas according to the actual operating concentration range, and configure multiple sets of mixed standard gases with "different target gas concentrations + different interfering gas concentration gradients" to cover various concentration combination scenarios that may occur in actual applications and ensure the comprehensiveness of sample data; use high-purity nitrogen as the dilution background gas, and adjust the flow ratio of the target gas, interfering gas and dilution gas according to the orthogonal ratio method, and fully mix them in the mixing chamber to obtain multiple sets of mixed standard gases with different target gas concentrations and different interfering gas concentration gradients.

[0031] Step 022: Select gas sensors with different sensitive materials (such as metal oxides, electrochemicals, semiconductors, MOFs-based) and different detection principles to form 10 to 20 candidate arrays. Each array contains 6 to 12 sensors (ensuring that the response mechanisms of the sensors in the array to the gas are different).

[0032] Step 023: In random order, each group of mixed standard gases is introduced into the test platform, with each group being stably supplied for an extended period (ensuring the sensors reach response equilibrium). Simultaneously, the response signals (such as resistance change rate, voltage output value, and current change) of each sensor in each candidate array are collected. After each gas test is completed, clean air is introduced to purge the sensor array before proceeding to the next gas test to avoid cross-contamination. Then, the average value of the differential response coefficients of all sensors in each candidate array is calculated, and the array with the largest average differential response coefficient is selected as the detection array.

[0033] Step 03: Correlate the response signals output by the detection array under each mixed standard gas environment with the concentration of the corresponding target gas to form sample data.

[0034] Specifically, the process of correlating the response signal output by the detection array under various mixed standard gases with the corresponding concentration of the target gas to form sample data includes: Step 031: Use a detection array to detect the mixed standard gas and output a continuous sequence of electrical signals.

[0035] It should be noted that during the detection process, the detection array continuously outputs electrical signals that change over time in a mixed standard gas environment. These electrical signals are the dynamic response signals of the sensor under the action of the gas, rather than single trigger signals. Each electrical signal fully reflects the sensor's response, stabilization, and recovery process.

[0036] Step 032: Extract feature parameters from the electrical signal sequence output by the detection array, and combine the feature parameters according to the order of each gas sensor in the detection array to form a multi-dimensional feature vector; wherein, the feature parameters include response amplitude, response time and curve integral value.

[0037] Understandably, extracting feature parameters from the continuous response signals of the detection array and combining them sequentially to form a multi-dimensional feature vector can fully capture the response differences between different gas components and gas sensors, significantly improving the accuracy of characterizing the target gas content. Through multi-dimensional feature extraction, the original electrical signal sequence is transformed into "structured, highly recognizable" feature parameters, uncovering the underlying gas component response patterns and addressing the insufficient accuracy of directly modeling from the original signal. The multi-dimensional feature vector is formed by combining the gas sensors in the detection array in sequence, preserving the array's "spatial distribution information." This allows the feature vector to reflect both the response characteristics of individual gas sensors and the collaborative sensing advantages of the array, providing a core basis for subsequent models to distinguish between target and interfering gases.

[0038] It should be noted that the response amplitude refers to the difference between the maximum response value of the sensor when it reaches a steady state under the action of the target gas and the baseline value; the response time refers to the time required for the sensor signal to change from the baseline value to reach the preset proportion of the maximum response amplitude; the curve integral value refers to the area value obtained by integrating the sensor response curve over time within the preset detection time window, which is used to characterize the overall response degree of the sensor to gas exposure.

[0039] When the detection array contains N gas sensors, the feature parameters extracted by each gas sensor are combined in the order of the sensors to construct a unified multidimensional feature vector, which has the following form: X=[A1, T] res 1 I1, ... A N T res N I N ] Each set of parameters in the multidimensional feature vector corresponds to the dynamic response characteristics of a gas sensor; A i T represents the response amplitude of the i-th sensor; res i I represents the response time of the i-th sensor; i This represents the curve integral value of the i-th sensor within the preset detection time window.

[0040] Step 033: Correlate the multidimensional feature vector with the corresponding concentration of the target gas to form sample data.

[0041] Understandably, the multi-dimensional feature vectors corresponding to a set of mixed standard gases obtained in step 032 are mapped one-to-one with the concentrations of the target gas in that set of mixed standard gases, forming a standardized, repeatable, and scalable sample dataset. This provides a high-quality training and testing foundation for subsequent machine learning, pattern recognition, and other algorithm models, which is beneficial for improving the accuracy and generalization ability of the concentration detection model. Through a complete process of signal extraction, feature combination, and correlation with target gas concentration, the original sensor signals are transformed into sample data that can be directly used for intelligent discrimination, realizing a digital mapping from gas signals to target gas concentrations, providing key support for the rapid detection of target gas concentrations. The multi-sensor collaborative response + multi-dimensional feature fusion approach can reduce the impact of single sensor failure or drift, improving the robustness of the entire detection system and its adaptability to practical application scenarios. Furthermore, the dataset includes scenario coverage with different target gas concentrations and different interfering gas concentrations, enabling the subsequently trained model to have anti-interference and generalization capabilities, avoiding the problem of narrow applicability caused by modeling under a single working condition.

[0042] Step 04: Based on the sample data and the support vector machine algorithm, a concentration detection model is trained to reflect the relationship between the response signal output by the detection array and the concentration of the target gas.

[0043] Specifically, the execution of a concentration detection model trained based on sample data and a support vector machine algorithm, which reflects the relationship between the response signal output by the detection array and the concentration of the target gas, includes: Using the multidimensional feature vectors corresponding to each group of mixed standard gases as input samples and the concentration of the corresponding target gas as the output label, a support vector machine (SVM) algorithm is employed to construct an optimal classification hyperplane in the feature space. This ensures maximum margin separation of samples with different target gas concentrations in the high-dimensional feature space, thus obtaining the concentration detection model. The concentration detection model trained by SVM can adapt to complex situations with different concentration ranges and different interference ratios, exhibits strong generalization ability, and eliminates the need for repeated calibration for specific scenarios, reducing maintenance costs and operational complexity in practical applications.

[0044] To improve model training efficiency and enhance generalization ability, principal component analysis (PCA) is used to reduce the dimensionality of multidimensional feature vectors before model construction. PCA projects high-dimensional features into a low-dimensional orthogonal space, preserving the main variance information while removing redundant information. In this process, the principal components with the largest eigenvalues ​​are selected to construct the low-dimensional feature space, thus retaining the most informative features from the original data. The specific dimensionality reduction steps include data standardization, covariance matrix calculation, eigenvalue decomposition, and principal component selection, the corresponding mathematical expressions of which are shown in the following formulas:

[0045]

[0046]

[0047]

[0048] Where x represents the original input data; Represents the original data matrix; and These represent the sample mean and standard deviation, respectively. This represents the standardized data matrix; Represent the covariance matrix; Represents eigenvalues; This represents the corresponding feature vector; Indicates from the previous The projection matrix composed of the principal component eigenvectors; This represents the feature matrix after dimensionality reduction; Indicates the number of samples.

[0049] The dimensionality-reduced data is then fed into a Support Vector Classifier (SVC) for training. The principle is to find a hyperplane that separates the two classes of samples, maximizing the margin between them. The sample point closest to the hyperplane is the support vector. This can be transformed into a mathematical problem: minimizing the magnitude of the weight vector.

[0050] Make:

[0051] in, The hyperplane normal vector; For bias terms; Number the sample; For sample labels.

[0052] Thus, SVC can solve linearly separable classification problems. For linearly inseparable classification problems, a suitable kernel function can be used to map the data into a high-dimensional space to make it linearly separable; slack variables can be introduced to enhance the generalization ability of the model.

[0053]

[0054] in, This is the penalty coefficient; These are slack variables.

[0055] The model performance was evaluated using 100 repeated 5-fold cross-validation to ensure the reliability of the evaluation results.

[0056] Understandably, the Support Vector Machine (SVM) algorithm constructs an optimal classification hyperplane in the feature space to achieve maximum margin separation of samples with different target gas concentrations. This effectively handles the classification problem of high-dimensional feature data, significantly reduces the risk of misjudgment caused by sample overlap, and greatly improves the model's accuracy in judging target gas concentrations. SVM uses kernel functions to map nonlinearly separable multidimensional feature vectors in low-dimensional space to high-dimensional feature space, accurately capturing the complex nonlinear relationship between the gas sensor's response features and the target gas concentration. Compared to traditional linear modeling methods, this more closely reflects the real correlation between actual gas response and target gas concentration, further improving the accuracy of the model's representation. Furthermore, for multidimensional feature vectors formed by the fusion of detection arrays, the SVM algorithm can build an efficient classification model without relying on a large number of samples, reducing dependence on the size of the sample data. Simultaneously, its optimal classification hyperplane is determined by only a few support vectors, resulting in low computational complexity, high modeling efficiency, and strong resistance to noisy data, ensuring the stability and computational efficiency of model training.

[0057] Step 05: Place the detection array in the gas environment to be tested, collect the real-time response signal and input it into the concentration detection model to obtain the detection concentration of the target gas in the gas environment to be tested.

[0058] Specifically, the detection array is placed in the environment of the gas to be tested, allowing it to come into contact with the gas. The array then outputs a continuous sequence of electrical signals. The multi-dimensional feature vectors acquired in real-time are input into a pre-trained concentration detection model. The model outputs the corresponding target gas concentration based on the distribution of the feature vectors in the feature space. Then, depending on the specific application scenario, the purpose of the detected target gas concentration is determined. For example, the concentration of ethylene (target gas) can be used to determine the ripeness of fruit, the concentration of hydrogen (target gas) can be used to determine if a lithium battery is experiencing thermal runaway, or the concentration of formaldehyde (target gas) can be used to determine if formaldehyde levels exceed the standard.

[0059] It should be noted that the identification of interfering gases using GC-MS in step 01, the selection of the detection array in step 02, the formation of sample data in step 03, and the training of the concentration detection model in step 04 were all completed in the laboratory, while the detection of the target gas in step 05 was completed in the actual test environment. This detection method can be used to detect formaldehyde in indoor air pollutants, with formaldehyde as the target gas; it can also be used to detect characteristic combustible gases released in the early stages of lithium battery thermal runaway, providing early warning of lithium battery thermal runaway, with hydrogen, methane, or carbon monoxide as the target gas; and it can also be used to detect gases released during fruit ripening, determining fruit ripeness, with ethylene as the target gas. Of course, this detection method can also detect other gases, which will not be elaborated upon here.

[0060] In summary, the detection method in this embodiment directly addresses the core pain points of traditional portable detectors in complex environments, namely "severe cross-interference and large detection deviation." By providing a true and reliable reference for gas composition through GC-MS, combined with the multi-dimensional feature capture capability of the gas-sensitive sensor array, and then establishing a precise mapping model through the support vector machine algorithm, the influence of similar interfering gases is eliminated in principle. This makes the accuracy and stability of the detection results far exceed those of traditional single-sensor solutions, fully meeting the stringent requirements of high-precision detection scenarios such as industrial production and medical health, and filling the precision gap of portable detection equipment in complex atmospheres.

[0061] In this embodiment, the detection method further includes: Record historical detection data generated under normal detection conditions to form a standard database; compare the real-time feature parameters acquired by the current detection array with the historical feature parameters in the standard database; when the feature parameters output by a certain gas sensor in the detection array continuously deviate from its historical normal distribution range, the gas sensor is determined to be an abnormal sensor; delete the feature parameters corresponding to the abnormal sensor and perform data compensation for the missing feature parameters.

[0062] Specifically, the process of deleting the feature parameters corresponding to the abnormal sensors and performing data compensation for the missing feature parameters includes: The feature parameters corresponding to the abnormal sensors are removed from the multidimensional feature vector. Based on the feature parameters generated by the other gas sensors in the detection array, the feature compensation for the removal position of the multidimensional feature vector is performed by interpolation estimation.

[0063] To address potential anomalies such as aging, contamination, or failure of individual sensors during long-term operation of the detection array, this embodiment constructs an anomaly identification and signal compensation mechanism based on historical data and multi-sensor consistency analysis. The specific steps include: First, abnormal sensor identification: A standard database is formed based on historically stable detection data. During real-time detection, the consistency of currently acquired sensor response amplitude, response time, and recovery time with their historical statistical ranges is compared and analyzed. When the output characteristic parameters of a sensor continuously deviate from their historical distribution range, the sensor is determined to be an abnormal sensor. The recovery time refers to the time required for the sensor signal to recover from its maximum response value to a preset proportion close to the baseline value after the target gas is removed.

[0064] Secondly, the elimination of abnormal sensors: when the sensor output is completely ineffective or the noise is severe, the characteristic parameters corresponding to the sensor are directly removed from the multidimensional feature vector.

[0065] Finally, the compensation mechanism for abnormal signals: After removing the feature parameters corresponding to abnormal sensors, interpolation estimation or model prediction methods are used to compensate for the removed feature parameters based on the output signals of the remaining normal sensors. This forms a complete and effective multidimensional feature vector, ensuring the consistency of the input dimensions of the concentration detection model and the stability of the detection results. For example, when the response amplitude of a sensor in the detection array is abnormal, the response amplitude can be estimated and compensated based on the response characteristics of other sensors under the same detection conditions using multiple regression or similarity matching models.

[0066] Understandably, establishing a standard database under normal detection conditions provides an objective and quantifiable reference for determining the sensor's operational status, avoiding the subjectivity and lag of relying solely on human experience to judge whether a sensor is abnormal. Continuous monitoring and interval comparison of the characteristic parameters of a single gas sensor can accurately locate individual abnormal sensors (rather than the entire device), enabling "early detection and early determination" of sensor faults. This avoids the problem of overall distortion of multi-dimensional feature vectors due to the failure of a single sensor (such as decreased sensitivity or response shift), which in turn leads to incorrect target gas concentration detection results, thus ensuring the reliability of the detection.

[0067] In this embodiment, the detection method further includes: The detection array is placed in a standard target gas environment of known concentration, and the baseline signal and characteristic parameters generated by the detection array are collected, and the corresponding calibration coefficients are generated accordingly.

[0068] During the testing process, calibration coefficients are used to perform baseline correction, sensitivity compensation, or scale normalization on the real-time acquired sensor signals in order to eliminate the influence of environmental changes and sensor drift on the test results.

[0069] In the specific implementation process, during the calibration phase, the detection array is exposed to a standard target gas environment of known concentration, and baseline signals and response characteristic parameters of each sensor are collected, and corresponding calibration coefficients are generated accordingly. In the subsequent detection process, the system uses the calibration coefficients to perform baseline correction, sensitivity compensation or scale normalization on the real-time sensor signals to eliminate the influence of environmental changes and sensor drift on the detection results, thereby ensuring the detection accuracy and consistency during long-term use.

[0070] Among them, baseline correction: eliminates baseline drift caused by changes in ambient temperature and humidity, ensuring that the "zero starting point" of the sensor is consistent under different environments; sensitivity compensation: compensates for the sensitivity decay of the sensor during use (such as response passivation caused by long-term exposure to gas), and restores its response accuracy to the target gas; scale normalization: unifies the output signals of different sensors to the same scale range, avoids the imbalance of feature vector weights caused by differences in sensor range and response amplitude, and ensures the standardization of model input data.

[0071] As is understandable, this embodiment eliminates systematic errors caused by environmental changes and sensor drift at the source, making detection data from different times, environments, and usage stages comparable, significantly improving the accuracy and consistency of target gas concentration detection results, eliminating the need for frequent shutdowns to replace sensors, ensuring that the detection system can maintain stable detection accuracy during long-term continuous operation, and extending the effective service life of the sensor.

[0072] Another embodiment of the present invention provides an electronic nose, such as Figure 2 As shown, the electronic nose includes: A gas-sensitive sensor array is used to detect the gas to be measured and output a response signal. Each sensor in the gas-sensitive sensor array can produce a different response to the target gas.

[0073] The signal acquisition module is used to acquire and preprocess the response signal to complete any missing data. Specifically, gas samples of different concentrations are treated as independent data points, with a sampling frequency of 2 Hz. The response phase lasts for 60 s, and the recovery phase lasts for 180 s. Therefore, the total acquisition time for each measurement cycle is 240 s. To address the potential data loss during long-term data acquisition, the signal acquisition module employs a linear interpolation method to complete the dataset, ensuring the integrity of the response signal.

[0074] The storage unit is used to store the processor's executable instructions and the concentration detection model.

[0075] The processor is electrically connected to the gas sensor array, the signal acquisition module, and the memory, respectively. When the processor executes the computer program, it acquires the response signal output by the gas sensor array and inputs the response signal into the concentration detection model to output the concentration detection result of the target gas, so that the electronic nose can realize the complex atmosphere gas detection method based on GC-MS as described in the above embodiment.

[0076] The communication module can be coupled to the processor to enable data interaction between the electronic nose and external devices, including but not limited to computers, mobile terminals, cloud servers or industrial control equipment, to complete functions such as transmission of detection results, model updates and remote monitoring.

[0077] The concentration detection model in this embodiment was constructed as follows: First, gas chromatography-mass spectrometry (GC-MS) was used to perform qualitative and quantitative analysis of volatile components in a complex atmosphere; then, a sensor array consisting of multiple gas sensors was constructed, and response data of the sensor array under different concentration conditions were collected; finally, a machine learning algorithm was used to establish the mapping relationship between the sensor response signal and the target gas concentration, thus forming the concentration detection model.

[0078] It should be noted that, in the method embodiments of the present invention, the construction process of the concentration detection model can be completed in a laboratory environment and pre-stored in the memory of the electronic nose device. In practical applications, the electronic nose only needs to perform the concentration prediction process, that is, quickly output the target gas concentration based on the real-time sensor response signal, thereby achieving online detection.

[0079] The electronic nose of this invention utilizes GC-MS to guide sensor array design and combines machine learning to establish a concentration detection model. Compared to traditional methods that rely on experience to select sensors, it offers advantages such as high detection accuracy, strong anti-interference capability, fast response speed, and ease of miniaturization and integration. It can be widely applied in fields such as fruit and vegetable maturity monitoring, food quality evaluation, environmental gas detection, and smart agriculture. Furthermore, unlike traditional methods that rely on experience, struggle to quantify complex on-site testing environments, and limit the application of large-scale equipment to miniaturized rapid testing, this electronic nose features a compact structure, small size, and better portability, meeting the demands of the miniaturized rapid testing market.

[0080] A communication module can be used to communicate with other devices or apparatuses, such as sending or receiving data and / or signals. A communication module may have at least one communication module for communication. A communication module may include any interface necessary for communicating with other devices. Exemplarily, a communication module may be a transceiver, circuit, bus, module, or other type of communication module.

[0081] The processor may include, but is not limited to, one or more of the following: a general-purpose computer, a special-purpose computer, a microcontroller, a digital signal processor (DSP), or a controller-based multi-core controller architecture. The device may have multiple processors, such as application-specific integrated circuit (ASIC) chips, which are time-dependent on a clock synchronized with the main processor.

[0082] The memory may include one or more non-volatile memories and one or more volatile memories. Examples of non-volatile memories include, but are not limited to, at least one of the following: read-only memory (ROM), electrically programmable read-only memory (EPROM), flash memory, hard disk, compact disc (CD), digital video disc (DVD), or other magnetic and / or optical storage. Examples of volatile memories include, but are not limited to, at least one of the following: random access memory (RAM), or other volatile memories that do not persist during the duration of a power outage.

[0083] A computer program consists of computer-executable instructions that are executed by an associated processor. Programs can be stored in ROM. A processor can perform any appropriate action and processing by loading the program into RAM.

[0084] Possible implementations of this application can be achieved through a program, enabling the communication device to execute any of the processes discussed in the foregoing embodiments. Possible implementations of this application can also be achieved through hardware or a combination of software and hardware.

[0085] In some implementations, the program may be tangibly contained in a computer-readable storage medium, which may include in a device (such as in memory) or other storage device accessible by the device. The program may be loaded from the computer-readable storage medium into RAM for execution. The computer-readable storage medium may include any type of tangible non-volatile memory, such as ROM, EPROM, flash memory, hard disk, CD, DVD, etc.

[0086] The above description is merely a specific embodiment of the present invention. Under the teachings of the present invention, those skilled in the art can make other improvements or modifications based on the above embodiments. Those skilled in the art should understand that the above specific description is only to better explain the purpose of the present invention, and the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for detecting gases in complex atmospheres based on GC-MS, characterized in that, The detection method includes: GC-MS is used to detect the ambient gas containing the target gas, obtain the main gas components, and other gases other than the target gas in the main gas components are regarded as interfering gases. Multiple sets of mixed standard gases with different target gas concentrations and different interfering gas concentration gradients are configured. Multiple gas-sensitive sensor arrays are screened using the mixed standard gases. The gas-sensitive sensor array that produces the most differentiated response to changes in the target gas concentration and the interfering gas concentration is selected as the detection array. Each sensor in the gas-sensitive sensor array can produce differentiated responses to the target gas. The response signal output by the detection array under each of the mixed standard gases is correlated with the concentration of the corresponding target gas to form sample data; Based on the sample data and the support vector machine algorithm, a concentration detection model that can reflect the relationship between the response signal output by the detection array and the concentration of the target gas is trained. The detection array is placed in the gas environment to be tested, real-time response signals are collected and input into the concentration detection model to obtain the detection concentration of the target gas in the gas environment to be tested.

2. The method for detecting complex atmosphere gases based on GC-MS according to claim 1, characterized in that, The step of correlating the response signal output by the detection array under each of the mixed standard gases with the concentration of the corresponding target gas to form sample data includes: The mixed standard gas is detected using the detection array, and a continuous sequence of electrical signals is output. Feature parameters are extracted from the electrical signal sequence output by the detection array, and the feature parameters are combined according to the order of each gas sensor in the detection array to form a multidimensional feature vector; The multidimensional feature vector is correlated with the corresponding concentration of the target gas to form sample data.

3. The method for detecting complex atmosphere gases based on GC-MS according to claim 2, characterized in that, The execution of training a concentration detection model based on the sample data and a support vector machine algorithm to reflect the relationship between the response signal output by the detection array and the concentration of the target gas includes: The multidimensional feature vectors corresponding to the mixed standard gases in each group are used as input samples, and the concentrations of the corresponding target gases are used as output labels. The optimal classification hyperplane is constructed in the feature space using the support vector machine algorithm, so that samples with different target gas concentrations can achieve maximum separation in the high-dimensional feature space, thereby obtaining the concentration detection model.

4. The method for detecting complex atmosphere gases based on GC-MS according to claim 2, characterized in that, The characteristic parameters include response amplitude, response time, and curve integral value.

5. The method for detecting complex atmosphere gases based on GC-MS according to claim 2, characterized in that, The detection method further includes: Record historical testing data generated under normal testing conditions to form a standard database; The real-time feature parameters acquired by the current detection array are compared with the historical feature parameters in the standard database. When the feature parameters output by a gas sensor in the detection array continuously deviate from its historical normal distribution range, the gas sensor is determined to be an abnormal sensor. Delete the feature parameters corresponding to the abnormal sensor and perform data compensation for the missing feature parameters.

6. The method for detecting complex atmosphere gases based on GC-MS according to claim 5, characterized in that, The process of deleting the feature parameters corresponding to the abnormal sensor and compensating for the missing feature parameters includes: The feature parameters corresponding to the abnormal sensor are removed from the multidimensional feature vector. Based on the feature parameters generated by the other gas sensors in the detection array, an interpolation estimation method is used to perform feature compensation on the removal position of the multidimensional feature vector.

7. The method for detecting complex atmosphere gases based on GC-MS according to claim 1, characterized in that, The detection method further includes: The detection array is placed in a standard target gas environment of known concentration, and the baseline signal and characteristic parameters generated by the detection array are collected, and the corresponding calibration coefficients are generated accordingly. During the detection process, the calibration coefficients are used to perform baseline correction, sensitivity compensation, or scale normalization on the real-time acquired sensor signals to eliminate the influence of environmental changes and sensor drift on the detection results.

8. An electronic nose, characterized in that, The electronic nose includes: The gas-sensitive sensor array is used to detect the gas to be measured and output a response signal. Each sensor in the gas-sensitive sensor array can produce a differentiated response to the target gas. The signal acquisition module is used to acquire the response signal and perform preprocessing. The storage device is used to store the executable instructions of the processor and the concentration detection model; The processor is electrically connected to the gas sensor array, the signal acquisition module, and the memory, respectively, and is used to acquire the response signal output by the gas sensor array and input the response signal into the concentration detection model to output the concentration detection result of the target gas, so that the electronic nose realizes the complex atmosphere gas detection method based on GC-MS as described in any one of claims 1 to 7.

Images