Gas identification method and gas intelligent identification method based on gas sensor array

Abstract

The application provides a gas identification method and a gas intelligent identification method based on a gas sensor array, and the gas identification method comprises the following steps: providing a test cavity provided with an air inlet and an air outlet, N heat-conducting gas sensors being arranged in the test cavity at intervals, the working temperatures of the N heat-conducting gas sensors being T1, T2,..., TN respectively, N>=2, inputting a reference gas into the test cavity through the air inlet, performing baseline calibration on the N heat-conducting gas sensors, inputting a to-be-detected gas into the test cavity through the air inlet, obtaining characteristic parameters of the to-be-detected gas according to the collection of output voltages of the N heat-conducting gas sensors, comparing the characteristic parameters with characteristic parameter values in a database, and obtaining the gas type of the to-be-detected gas. N The gas identification method can realize the identification of multiple gases, avoid the interference of gas concentration changes on gas identification, and improve the accuracy of gas identification.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor integrated circuit manufacturing technology, and relates to a gas identification method and a gas intelligent identification method based on a gas sensor array. Background Technology

[0002] Selective detection and identification of gases is one of the core requirements in the field of gas sensing technology, with significant application value in various fields such as industrial production, environmental monitoring, medical and health care, and safety protection. Currently, various gas sensor technologies have been developed in this field, including electrochemical, metal-oxide-semiconductor, field-effect transistor, catalytic combustion, and thermal conductivity sensors. Each technology, with its unique working principle and performance characteristics, is suitable for different gas detection scenarios. Among them, the thermal conductivity sensor (TCS), as a classic gas detection device, works based on the difference in thermal conductivity between the target gas and a reference gas (usually air). Due to its advantages such as simple structure, low cost, fast response speed, and responsiveness to various gases including hydrogen and carbon dioxide, it has been widely used in the concentration detection and leak monitoring of the aforementioned gases.

[0003] However, existing thermal conductivity sensor technology has significant core limitations: the response signals lack specificity, making it unable to independently distinguish and identify gas types. Specifically, on the one hand, different gases with similar thermal conductivity will produce similar response signals when acting on a thermal conductivity sensor, causing the sensor to be unable to effectively distinguish between these gases; on the other hand, even two gases with significantly different thermal conductivity may cause the sensor to exhibit similar response characteristics under different concentration conditions, further exacerbating the problem of insufficient accuracy in gas identification.

[0004] To improve the aforementioned selectivity defects, selective filters are often added to the sensor probe, catalytic coatings are applied to the surface of the sensitive element, or thermal conductivity sensors are used in conjunction with gas separation systems. However, these methods are all auxiliary improvement measures that can only weaken the influence of interfering gases to a certain extent. They cannot fundamentally overcome the inherent limitation of the non-specific response signal of thermal conductivity sensors, and it is still difficult to realize the independent application of thermal conductivity sensors in gas identification.

[0005] Therefore, how to provide a gas identification method and a gas intelligent identification method based on a gas sensor array to achieve accurate identification of multiple gases has become an important problem that needs to be solved by those skilled in the art.

[0006] It should be noted that the above introduction to the technical background is only for the purpose of providing a clear and complete explanation of the technical solutions of this application and facilitating understanding by those skilled in the art. It should not be assumed that these technical solutions are known to those skilled in the art simply because they have been described in the background section of this application. Summary of the Invention

[0007] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide a gas identification method and a gas intelligent identification method based on a gas sensor array, so as to solve the problem that the response signal of thermal conductivity sensor technology lacks specificity and the gas identification accuracy is low in the prior art.

[0008] To achieve the above and other related objectives, the present invention provides a gas identification method, characterized by comprising the following steps:

[0009] Step 1: Provide a test chamber with an air inlet and an air outlet. N thermal conductivity gas sensors are installed within the test chamber at intervals. The operating temperatures of the N thermal conductivity gas sensors are set as T1...T... N N≥2;

[0010] Step 2: Input reference gas into the test chamber through the air inlet. The reference gas flows through N thermal conductivity gas sensors and exits the test chamber through the air outlet to perform baseline calibration on the N thermal conductivity gas sensors.

[0011] Step 3: Input the gas to be tested into the test chamber through the air inlet, and the gas to be tested flows through N thermal conductivity gas sensors and flows out of the test chamber through the air outlet;

[0012] Step 4: Based on the acquisition of the output voltages of the N thermal conductivity gas sensors, obtain the characteristic parameters of the gas to be tested, compare the characteristic parameters with the characteristic parameter values ​​in the database, and obtain the gas type of the gas to be tested.

[0013] Optionally, the N thermal conductivity sensors are arranged in an array to form a thermal conductivity sensor array.

[0014] Optionally, step 4 includes the following steps:

[0015] Two of the N thermal conductivity gas sensors are selected as preliminary gas classification sensors. When the thermal conductivity of the gas to be tested is greater than that of the reference gas, the output signal of the preliminary gas classification sensor is set to V > 0, and otherwise the output signal is V < 0.

[0016] The initial type of the gas to be tested is determined by the output signals of the two preliminary gas classification sensors;

[0017] After the gas to be tested flows through N of the thermal conductivity gas sensors, the output signals of the two preliminary gas classification sensors and the output voltages measured by the N thermal conductivity gas sensors are acquired.

[0018] Based on the output signals of the two preliminary gas classification sensors, the preliminary type of the gas to be tested is determined;

[0019] The characteristic parameter K is obtained by selecting the ratio of any two output voltages among the N output voltages measured by the thermal conductivity gas sensors.

[0020] The characteristic parameter K is compared with the characteristic parameters in the database of the corresponding preliminary type to obtain the gas type of the gas to be tested.

[0021] Optionally, setting the preliminary type of the gas to be tested using the output signals of the two preliminary gas classification sensors includes the following steps:

[0022] The output signals of the two preliminary gas sorting sensors are denoted as V1 and V2;

[0023] When V1 > 0 and V2 > 0, the preliminary type of the gas to be tested is determined to be a Class I gas;

[0024] When V1 < 0 and V2 > 0, the preliminary type of the gas to be tested is determined to be a Class II gas;

[0025] When V1 < 0 and V2 < 0, the preliminary type of the gas to be tested is determined to be a Class III gas.

[0026] Optionally, when performing the step of selecting the ratio of any two output voltages among the N output voltages measured by the thermal conductivity gas sensors, the two output voltages correspond to the highest and lowest operating temperatures of the thermal conductivity gas sensors, respectively.

[0027] Optionally, step 4 includes the following steps:

[0028] When the thermal conductivity of the gas to be measured is greater than that of the reference gas, the output voltage V of the thermal conductivity gas sensor is set to be greater than 0; when the thermal conductivity of the gas to be measured is greater than that of the reference gas, the output voltage V of the thermal conductivity gas sensor is set to be less than 0; when the thermal conductivity of the gas to be measured is equal to that of the reference gas, the output voltage V of the thermal conductivity gas sensor is set to be 0.

[0029] Among the N thermal conductivity gas sensors, the two thermal conductivity gas sensors with the lowest and highest operating temperatures are designated as preliminary gas classification sensors.

[0030] After the gas to be tested flows through N of the thermal conductivity gas sensors, N output voltages are obtained from the N thermal conductivity gas sensors.

[0031] The combined output voltage values ​​of the two preliminary gas classification sensors are read from the N output voltages. When the positive and negative signs of the output voltages of the two preliminary gas classification sensors are different and neither is zero, the operating temperature of the thermal conductivity gas sensor with output voltage V=0 among the N output voltages is obtained, and the characteristic parameter T is obtained.

[0032] The gas type of the gas to be tested is obtained by comparing the feature parameter T with the feature parameters in the database.

[0033] Optionally, the following steps are also included:

[0034] Select the output voltage of any one of the thermal conductivity gas sensors;

[0035] The concentration calculation formula is determined based on the type of gas to be measured and the operating temperature of the selected thermal conductivity gas sensor.

[0036] Substituting the output voltage into the concentration calculation formula, the gas concentration of the gas to be measured is obtained.

[0037] Optionally, the following steps are also included:

[0038] Based on the type of gas to be measured and the operating temperatures of the N thermal conductivity gas sensors, determine N concentration calculation formulas;

[0039] The output voltage V of the N thermal conductivity gas sensors T1 ···V TN Substitute these values ​​into their respective concentration calculation formulas to obtain N gas concentration values ​​X. T1 ···X TN ;

[0040] X of the N gas concentration values T1 ···X TN The gas concentration of the gas to be measured is obtained by averaging.

[0041] This invention also provides a gas intelligent identification method based on a gas sensor array, comprising the following steps:

[0042] Step 1: Signal Acquisition: Provide a sensor array with N thermal conductivity sensors. The operating temperatures of the N thermal conductivity sensors are evenly distributed. A reference gas is introduced to perform baseline calibration on the N thermal conductivity sensors. The gas to be tested is introduced into the sensor array through a gas delivery device. Sensor response curves at different temperatures are acquired, where N≥2.

[0043] Step 2: Data preprocessing: The acquired signals are preprocessed, including data augmentation and normalization, feature extraction, and data sample formation.

[0044] Step 3: Prediction and Recognition: Input the data samples into multiple classification models, train multiple classification models using a grid search method, select the classification model with the highest recognition rate among multiple classification models to identify the gas types, obtain the gas classification results, and finally generate high-precision recognition results.

[0045] Optionally, extracting feature values ​​in step 2 includes the following steps:

[0046] Extract the inflection point information of the collected response curves, wherein the inflection point information includes the response value of each response curve and the ratio of any two response values;

[0047] The gas types are tagged, and a data structure is constructed that includes the tags and the inflection point information.

[0048] As described above, the gas identification method of the present invention includes the steps of: providing a test chamber with an inlet and an outlet, wherein N thermal conductivity gas sensors are arranged at intervals within the test chamber, and the operating temperatures of the N thermal conductivity gas sensors are set to T1...T1...T2...T3...T4...T5...T6...T7...T7...T8...T9...T1 ... N The method involves several steps: N ≥ 2; a reference gas is input into the test chamber through the inlet, flows through N thermal conductivity gas sensors, and exits the test chamber through the outlet to perform baseline calibration on the N thermal conductivity gas sensors; the gas to be tested is input into the test chamber through the inlet, flows through N thermal conductivity gas sensors, and exits the test chamber through the outlet; based on the collected output voltages of the N thermal conductivity gas sensors, characteristic parameters of the gas to be tested are obtained, and these characteristic parameters are compared with the characteristic parameter values ​​in the database to determine the gas type. This gas identification method can identify multiple gases, avoids interference from changes in gas concentration, and improves the accuracy of gas identification. Attached Figure Description

[0049] Figure 1 The diagram shows the steps of the gas identification method of the present invention.

[0050] Figure 2 The diagram shown is a structural schematic of a test chamber in the gas identification method of the present invention.

[0051] Figure 3 The image shown is a schematic diagram illustrating the setting of a preliminary type in the gas identification method of the present invention.

[0052] Figure 4 The graph shows the thermal conductivity-temperature of different gases.

[0053] Figure 5 The graph shows the K-concentration curves for three different gases.

[0054] Figure 6 The diagram shows the steps of the intelligent gas identification method based on a gas sensor array according to the present invention.

[0055] Figure 7 The image shown is a schematic diagram illustrating the extraction of inflection point information of the response curve in the gas intelligent identification method based on a gas sensor array according to the present invention.

[0056] Figure 8 This is a schematic diagram of a PCA dimensionality reduction classification result based on the KNN algorithm.

[0057] Figure 9 Displayed as Figure 8 A diagram showing the corresponding recognition accuracy.

[0058] Explanation of reference numerals in the attached figures

[0059] 1 Test chamber 11 air intake 12 air outlet 13 Thermal conductivity gas sensor 14 Thermal conductivity sensor array Detailed Implementation

[0060] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0061] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, whole, step, or component, but does not exclude the presence or addition of one or more other features, wholes, steps, or components.

[0062] Features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, combined with features in other embodiments, or substituted for features in other embodiments.

[0063] In the detailed description of embodiments of the present invention, for ease of explanation, the schematic diagrams illustrating the device structure may be partially enlarged without adhering to the general scale, and the schematic diagrams are merely examples and should not limit the scope of protection of the present invention. Furthermore, in actual manufacturing, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0064] For ease of description, spatial relation terms such as “below,” “under,” “lower than,” “below,” “above,” and “upper” may be used herein to describe the relationship between one element or feature shown in the accompanying drawings and other elements or features. It will be understood that these spatial relation terms are intended to include directions other than those depicted in the drawings for devices in use or operation. Furthermore, when a layer is referred to as being “between” two layers, it may be the only layer between the two layers, or there may be one or more layers in between.

[0065] In the context of this application, the structure described above the first feature may include embodiments in which the first and second features are in direct contact, or embodiments in which additional features are formed between the first and second features, such that the first and second features may not be in direct contact.

[0066] It should be noted that the illustrations provided in this embodiment are only schematic representations of the basic concept of the present invention. Therefore, the illustrations only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0067] Example 1

[0068] This embodiment provides a gas identification method. Please refer to [link / reference]. Figure 1 The flowchart shown is a step diagram of the gas identification method of the present invention, which includes at least the following steps:

[0069] Step 1: Please refer to Figure 2 A test chamber 1 is provided, which has an air inlet 11 and an air outlet 12. N thermal conductivity gas sensors 13 are arranged at intervals within the test chamber 1. The operating temperatures of the N thermal conductivity gas sensors 13 are set to T1...T... N , N≥2.

[0070] For example, please refer to [link / reference]. Figure 2 N thermal conductivity sensors 13 are arranged in an array to form a thermal conductivity sensor array 14. In this embodiment, N=4. Figure 2 The arrows in the diagram indicate the direction of gas flow.

[0071] As an example, the thermal conductivity gas sensor 13 includes a sensitive unit chip and a reference unit chip. Specifically, the thermal conductivity sensor operates under different conditions (e.g., different operating temperatures). The test chamber 1 has an electrical adapter to connect the chip to an external circuit. The gas is identified by analyzing the response curves under different conditions.

[0072] As an example, the thermal conductivity gas sensor 13 is a sensor fabricated using MEMS technology. Sensors fabricated using MEMS technology have a small structure, low cost, and low power consumption, making them suitable for miniaturized scenarios such as wearable devices.

[0073] As an example, the operating temperature range of the N thermal conductivity gas sensors 13 is 50°C to 500°C, for example, the operating temperature of the thermal conductivity gas sensors 13 is set to 100°C, 150°C, 200°C or 250°C.

[0074] Step 2: Input reference gas into the test chamber 1 through the air inlet 11. The reference gas flows through N thermal conductivity gas sensors 13 and exits the test chamber 1 through the air outlet 12 to perform baseline calibration on the N thermal conductivity gas sensors 13.

[0075] Specifically, baseline calibration refers to adjusting the external circuit to make the output voltage of the N thermal conductivity gas sensors 13 zero under the atmosphere of the reference gas, so as to ensure the consistency of the environment in which the reference resistance and the sensitive resistance of the N sensors are located. In this embodiment, the reference gas is nitrogen. In other embodiments, other suitable gases may be used as the reference gas.

[0076] Step 3: The gas to be tested is input into the test chamber 1 through the air inlet 11. The gas to be tested flows through N thermal conductivity gas sensors 13 and flows out of the test chamber 1 through the air outlet 12.

[0077] As an example, the types of gases to be tested include, but are not limited to, methane, ethane, propane, isobutane, and hydrogen.

[0078] Step 4: Based on the acquisition of the output voltage of the N thermal conductivity gas sensors 13, obtain the characteristic parameters of the gas to be tested, compare the characteristic parameters with the characteristic parameter values ​​in the database, and obtain the gas type of the gas to be tested.

[0079] As an example, step 4 includes the following steps:

[0080] Step 41: Select two thermal conductivity gas sensors 13 from the N thermal conductivity gas sensors 13 as preliminary gas classification sensors. When the thermal conductivity of the gas to be measured is greater than that of the reference gas, set the output signal V > 0 of the preliminary gas classification sensor, and otherwise set the output signal V < 0.

[0081] As an example, when performing the step of selecting the ratio of any two output voltages among the N output voltages measured by the thermal conductivity gas sensors 13, the two output voltages correspond to the highest and lowest operating temperatures of the thermal conductivity gas sensors 13, respectively.

[0082] Step 42: The preliminary type of the gas to be tested is set by the output signals of the two preliminary gas classification sensors.

[0083] For details, please see Figure 3 The preliminary type of the gas to be tested is determined by the output signals of the two preliminary gas classification sensors, comprising the following steps:

[0084] (1) The output signals of the two preliminary gas sorting sensors are denoted as V1 and V2;

[0085] (2) When V1 > 0 and V2 > 0, the preliminary type of the gas to be tested is determined to be a Class I gas;

[0086] (3) When V1 < 0 and V2 > 0, the preliminary type of the gas to be tested is determined to be a second type of gas;

[0087] (4) When V1 < 0 and V2 < 0, the preliminary type of the gas to be tested is determined to be a Class III gas.

[0088] To better illustrate the process of setting the preliminary type of the gas to be tested using the output signals of the two preliminary gas classification sensors, preliminary gas classification was performed on some sample gases. The preliminary gas classification results are shown in Table 1 below:

[0089] Table 1 Preliminary gas classification results of sample gases

[0090]

[0091] The operating temperatures of the two preliminary gas sorting sensors are set to 100°C and 200°C, respectively.

[0092] Step 43: After the gas to be tested flows through N of the thermal conductivity gas sensors 13, acquire the output signals of the two preliminary gas classification sensors and the output voltages measured by the N thermal conductivity gas sensors 13.

[0093] Step 44: Based on the output signals of the two preliminary gas classification sensors, determine the preliminary type of the gas to be tested.

[0094] Step 45: Select the ratio of any two output voltages among the N output voltages measured by the thermal conductivity gas sensors 13 to obtain the characteristic parameter K.

[0095] Specifically, based on the principle of thermal conductivity sensors, the output voltage V out With the target gas concentration X TG and the difference in thermal conductivity (λ) TG -λ REF They are directly proportional: V out ∝XTG •(λ TG -λ REF The output voltage of two thermal conductivity sensors with operating temperatures of T1 and T2 is set to V. T1 and V T2 The formula for calculating the value of K is as follows:

[0096]

[0097] Wherein, (λ) TG，T2 -λ REF，T2 ) represents the difference in thermal conductivity between the target gas and the reference gas at the operating temperature T2, (λ) TG，T1 -λ REF，T1 The value of K represents the difference in thermal conductivity between the target gas and the reference gas at the operating temperature T1. As can be seen from the formula above, the K value is only related to the temperature characteristics of the gas's thermal conductivity and is independent of the gas concentration. Using the K value as a characteristic parameter will not be affected by changes in gas concentration, thus avoiding interference from gas concentration variations in gas identification and enabling accurate identification of the type of gas to be tested.

[0098] For details, please see Figure 4 The graph displays the thermal conductivity-temperature dot plots for different gases. It shows one gas from each of the three categories (Type I, Type II, and Type III), demonstrating that different gases have different K values. Type I gases are represented by H₂ or CH₄, Type II by C₃H₈, and Type III by CO₂. The reference gas is N₂.

[0099] For details, please see Figure 5 The graph shows the K-concentration curves for three different gases. It can be seen that the K value for each gas is a stable straight line at different concentrations, meaning that the K value is not affected by changes in concentration. Using the K value as a characteristic parameter to determine the type of gas can be more accurate.

[0100] Step 46: Compare the feature parameter K with the feature parameters in the database of the corresponding preliminary type to obtain the gas type of the gas to be tested.

[0101] Specifically, under the same initial type, each feature parameter in the database corresponds to a type of gas. It should be noted that the database is obtained by measuring the K values ​​of multiple known gas types under different test conditions (e.g., different operating temperatures) according to the steps described above, thus constructing the database.

[0102] Furthermore, in steps 41 to 46 above, determining the preliminary type of the gas to be tested before measuring the characteristic parameter K can further improve the gas identification accuracy. Specifically, gases in different preliminary types may have the same characteristic parameter K. If the preliminary type is not determined, misjudgment is likely to occur, and the gas type of the gas to be tested cannot be correctly identified. Therefore, by first distinguishing the preliminary type, the characteristic parameter K of the gas to be tested can be directly compared with the characteristic parameters in the database that determine the preliminary type (e.g., the first type of gas). This can avoid misjudgment and accurately identify gases with the same characteristic parameter K but in different preliminary types, thereby further improving the accuracy of gas type identification.

[0103] As an example, step 4 includes the following steps:

[0104] Step 41': When the thermal conductivity of the gas to be tested is greater than that of the reference gas, the output voltage V of the thermal conductivity gas sensor 13 is set to be greater than 0; when the thermal conductivity of the gas to be tested is greater than that of the reference gas, the output voltage V of the thermal conductivity gas sensor 13 is set to be less than 0; when the thermal conductivity of the gas to be tested is equal to that of the reference gas, the output voltage V of the thermal conductivity gas sensor 13 is set to be 0.

[0105] Step 42': Among the N thermal conductivity gas sensors 13, the two thermal conductivity gas sensors 13 with the lowest and highest operating temperatures are designated as preliminary gas classification sensors.

[0106] Step 43': After the gas to be tested flows through N of the thermal conductivity gas sensors 13, obtain N output voltages measured by the N thermal conductivity gas sensors 13.

[0107] Step 44': Read the combined output voltage values ​​of the two preliminary gas classification sensors from the N output voltages. When the positive and negative signs of the output voltages of the two preliminary gas classification sensors are different and neither is zero, obtain the operating temperature of the thermal conductivity gas sensor 13 with output voltage V=0 among the N output voltages, and obtain the characteristic parameter T.

[0108] Step 45': Compare the characteristic parameter T with the characteristic parameters in the database to obtain the gas type of the gas to be tested. It should be noted that the database is obtained by measuring the corresponding T values ​​of multiple known gas types under different test conditions (e.g., different operating temperatures) according to the above steps, thus constructing the database.

[0109] Specifically, by setting the operating temperature of the thermally conductive gas sensor 13 with an output voltage V=0 as the characteristic parameter T, the type of gas can be directly determined by the output voltage of the thermally conductive gas sensor 13. This is because each gas corresponds to a different temperature value when the output voltage V=0. Based on the above, the N intervals of the thermally conductive gas sensors 13 in the test chamber 1 have N operating temperatures, and the thermally conductive gas sensors 13 at the N operating temperatures will correspond to N types of gas when the output voltage V=0. In other words, the N types of gas can be identified by the characteristic parameter T corresponding to the output voltage V=0 (i.e., the operating temperature of the thermally conductive gas sensor 13 with the output voltage V=0), without having to calculate the ratio of any two output voltages (i.e., saving the step 45 of calculating the characteristic parameter K mentioned above). For example, please refer to Table 1. The characteristic parameter of ethane is T=100℃, while the characteristic parameter of isobutane is T=200℃. When the output voltage V of the thermal conductivity gas sensor 13 with an operating temperature of 100℃ in the test chamber 1 is 0, the type of gas to be tested can be determined to be ethane. When the output voltage V of the thermal conductivity gas sensor 13 with an operating temperature of 200℃ in the test chamber 1 is 0, the type of gas to be tested can be determined to be isobutane.

[0110] As an example, step 4 is followed by the following steps:

[0111] (1) Select any one of the output voltages of the thermal conductivity gas sensor 13;

[0112] (2) Determine the concentration calculation formula based on the type of gas to be measured and the operating temperature of the selected thermal conductivity gas sensor 13;

[0113] (3) Substitute the output voltage into the concentration calculation formula to obtain the gas concentration of the gas to be measured.

[0114] Specifically, the concentration calculation formula is V=F(X,T), where V is the output voltage, X is the gas concentration, and T is the operating temperature of any of the thermal conductivity gas sensors 13.

[0115] As an example, step 4 is followed by the following steps:

[0116] (1) Based on the type of gas to be measured and the operating temperatures of the N thermal conductivity gas sensors 13, determine N concentration calculation formulas;

[0117] (2) Convert the output voltage V of N thermal conductivity gas sensors 13 T1 ···V TN Substitute these values ​​into their respective concentration calculation formulas to obtain N gas concentration values ​​X. T1 ···X TN ;

[0118] (3) Take the N gas concentration values ​​X T1 ···X TN The gas concentration of the gas to be tested is obtained by averaging. The gas concentration of the gas to be tested can be determined more accurately by averaging.

[0119] The gas identification method in this embodiment uses characteristic parameters independent of gas concentration to determine the type of gas to be tested. It identifies the gas based on the difference in the rate of change of gas thermal conductivity with temperature, which can realize the identification of multiple gases, avoid the interference of gas concentration changes on gas identification, and improve the accuracy of gas identification.

[0120] Example 2

[0121] This embodiment provides a gas intelligent identification method based on a gas sensor array. Please refer to [link to relevant documentation]. Figure 6 The flowchart shown is a step-by-step diagram of the gas intelligent identification method based on a gas sensor array according to the present invention, including the following steps:

[0122] Step 1: Signal Acquisition: A gas sensor array with N thermal conductivity sensors is provided. The operating temperatures of the N thermal conductivity sensors are distributed equally. A reference gas is introduced to perform baseline calibration on the N thermal conductivity sensors. The gas to be tested is introduced into the gas sensor array through a gas delivery device. Sensor response curves at different temperatures are acquired, where N≥2. In this embodiment, nitrogen is used as the reference gas. In other embodiments, other suitable gases can also be used as the reference gas.

[0123] As an example, the N thermal conductivity sensors use thermal conductivity chips that integrate both sensitive and reference units as thermal conductivity sensors, which can monitor the change in thermal conductivity of ambient gas relative to a reference gas (e.g., nitrogen) in real time.

[0124] As an example, the types of gases to be tested include, but are not limited to, hydrogen, methane, ethane, propane, butane, carbon dioxide, Freon, and other gases.

[0125] As an example, each of the thermal conductivity sensors in the gas sensor array has an independent operating temperature adjustment module, which can produce a non-specific response to all gas molecules, but the thermal conductivity of different gases changes at different rates with temperature, and the gas is identified based on this difference.

[0126] Step 2: Data preprocessing: The acquired signals are preprocessed, including data augmentation and normalization, feature extraction, and data sample formation.

[0127] As an example, in step 2, the data augmentation algorithm uses a Gaussian perturbation algorithm, where the normalization formula is as follows:

[0128]

[0129] In the formula, x represents the original data, u represents the mean of the original dataset, and σ represents the standard deviation of the original dataset.

[0130] As an example, step 2, extracting feature values, includes the following steps:

[0131] (1) Please refer to Figure 7 Extract the inflection point information of the collected response curves. The inflection point information includes the response values ​​of each response curve (i.e., output voltage a and output voltage b) and the ratio of any two response values ​​(i.e., the response ratio c mentioned later).

[0132] (2) Tag the gas types and construct a data structure for the tags and the inflection point information.

[0133] Step 3: Prediction and Recognition: Input the data samples into multiple classification models, train multiple classification models using a grid search method, select the classification model with the highest recognition rate among multiple classification models to identify the gas types, obtain the gas classification results, and finally generate high-precision recognition results.

[0134] Specifically, in step 3, the process of selecting the classification model with the highest recognition rate among multiple classification models based on the performance evaluation results is as follows: This involves calculating the accuracy, R... 2 The performance of the classification model is evaluated using metrics such as F1 score, training time, and overfitting, and the best classification model (i.e., the one with the highest recognition rate) is selected. In this embodiment, the coefficient of determination R is used. 2 As a model evaluation metric, Ri 2 The calculation formula is as follows:

[0135]

[0136] in, These are actual observed values. This is the average of the actual observed values. This represents the model's predicted output value, where n is the number of test points.

[0137] In other embodiments, the coefficient of determination F1 can also be used as the model evaluation index. The formula for calculating F1 is as follows:

[0138]

[0139] Precision is the accuracy rate, and Recall is the recall rate.

[0140] As an example, the classification model includes Support KNN, Support Vector Machine, Decision Tree, Random Forest, Logistic Regression, and Multilayer Perceptron. In this embodiment, Support KNN is used.

[0141] Please see Figure 8 The figure shows a schematic diagram of PCA dimensionality reduction classification results based on the KNN algorithm. This figure is based on measurements from a gas sensor array consisting of two thermal conductivity sensors with different operating temperatures (where the operating temperatures of the two thermal conductivity sensors are T1 and T2, respectively). Figure 8 The horizontal and vertical coordinates (PC1 and PC2) in the graph are calculated from the output voltage a of the gas at T1, the output voltage b at T2, and the response ratio c (c=a / b) measured by the sensor array. When the gas to be tested passes through the sensor array, the sensor array will output the output voltage a1 at T1, the output voltage b1 at T2, and the response ratio c1 of the gas to be tested. From this, a specific PC1 value and PC2 value can be obtained. The two values ​​determine a data point in the graph. The location of the data point of the gas to be tested is determined by the three known gas samples closest to that location, which vote equally to determine which gas the gas to be tested belongs to.

[0142] Specifically, the relationships between PC1 and PC2 and the output voltages a and b, as well as the response ratio c, are as follows:

[0143] PC1 = 0.689a + 0.640b + 0.339c

[0144] PC2 = -0.081a - 0.396b + 0.951c

[0145] The steps for calculating the coefficients in the above formula are as follows: Input known gas data samples into the model. Each data sample includes four items: [gas type, a, b, c], for a total of 800 data samples. The model code calculates the coefficients based on these known data samples, thus constructing the model. Figure 8 The dimensionality reduction region diagram shown should be noted as follows: Figure 8 The area diagram showing six gases is just an example; in other embodiments, more gases can be in the same layout.

[0146] The calculation process is based on standard PCA dimensionality reduction, a mature method, and the main steps include the following:

[0147] (1) Data standardization: First, standardize each feature to convert the data to a mean of 0 and a variance of 1;

[0148] (2) Calculate the covariance matrix: The goal is to find the linear relationship between all features. We have 800 samples and 3 features;

[0149] (3) Calculate the eigenvalues ​​and eigenvectors of the covariance matrix: find the eigenvalues ​​λ and the corresponding eigenvectors v of the covariance matrix Σ;

[0150] (4) Selecting principal components: This determines how many dimensions to retain;

[0151] (5) Construct the projection matrix: Arrange the selected first k eigenvectors to form an n×k matrix W;

[0152] (6) Data dimensionality reduction: Multiply the original data X with the projection matrix W to obtain the dimensionality-reduced data Y and the dimensionality-reduced region map.

[0153] Please see Figure 9 Displayed as Figure 8 The diagram shows the corresponding recognition accuracy. The numbers in the squares represent the number of predictions. As can be seen from the diagram, the recognition accuracy of the six gas types based on the KNN algorithm is close to 100%, which indicates that the gas identification method based on the classification model in this embodiment has high recognition accuracy.

[0154] This embodiment combines a gas intelligent identification method based on a classification model with a gas sensor array. This method automatically extracts and classifies the features of the response curve, significantly improving the accuracy and intelligence of gas identification.

[0155] In summary, the gas identification method of the present invention includes the following steps: providing a test chamber with an inlet and an outlet, wherein N thermal conductivity gas sensors are arranged at intervals within the test chamber, and the operating temperatures of the N thermal conductivity gas sensors are set to T1...T1...T2...T3...T4...T5...T6...T7...T7...T8...T9...T1 ... N The method involves inputting a reference gas into the test chamber through the inlet, allowing it to flow through N thermal conductivity gas sensors and exit the chamber through the outlet, thus performing baseline calibration on the N sensors. The method also inputs the gas to be tested into the test chamber through the inlet, allowing it to flow through N thermal conductivity gas sensors and exit through the outlet. Based on the voltage outputs of the N thermal conductivity gas sensors, characteristic parameters of the gas to be tested are obtained. These parameters are then compared with values ​​in a database to determine the gas type. This gas identification method enables the identification of multiple gases, avoids interference from changes in gas concentration, and improves the accuracy of gas identification. Therefore, this invention effectively overcomes the shortcomings of existing technologies and has high industrial application value.

[0156] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A gas identification method, characterized in that, Includes the following steps: Step 1: Provide a test chamber with an air inlet and an air outlet. N thermal conductivity gas sensors are installed within the test chamber at intervals. The operating temperatures of the N thermal conductivity gas sensors are set as T1...T... N N≥2; Step 2: Input reference gas into the test chamber through the air inlet. The reference gas flows through N thermal conductivity gas sensors and exits the test chamber through the air outlet to perform baseline calibration on the N thermal conductivity gas sensors. Step 3: Input the gas to be tested into the test chamber through the air inlet, and the gas to be tested flows through N thermal conductivity gas sensors and flows out of the test chamber through the air outlet; Step 4: Based on the collected output voltage data from the N thermal conductivity gas sensors, obtain the characteristic parameters of the gas to be tested. Compare these characteristic parameters with the characteristic parameter values ​​in the database to determine the gas type of the gas to be tested. This includes the following steps: When the thermal conductivity of the gas to be tested is greater than that of the reference gas, the output voltage V of the thermal conductivity gas sensor is set to be greater than 0; when the thermal conductivity of the gas to be tested is less than that of the reference gas, the output voltage V of the thermal conductivity gas sensor is set to be less than 0; when the thermal conductivity of the gas to be tested is equal to that of the reference gas, the output voltage V of the thermal conductivity gas sensor is set to be 0. Among the N thermal conductivity gas sensors, the two thermal conductivity gas sensors with the lowest and highest operating temperatures are designated as preliminary gas classification sensors. After the gas to be tested flows through N of the thermal conductivity gas sensors, N output voltages are obtained from the N thermal conductivity gas sensors. The combined output voltage values ​​of the two preliminary gas classification sensors are read from the N output voltages. When the positive and negative signs of the output voltages of the two preliminary gas classification sensors are different and neither is zero, the operating temperature of the thermal conductivity gas sensor with output voltage V=0 among the N output voltages is obtained, and the characteristic parameter T is obtained. The gas type of the gas to be tested is obtained by comparing the feature parameter T with the feature parameters in the database.

2. The gas identification method according to claim 1, characterized in that: N of the aforementioned thermal conductivity gas sensors are arranged in an array to form a thermal conductivity sensor array.

3. The gas identification method according to claim 1, characterized in that, It also includes the following steps: Select the output voltage of any one of the thermal conductivity gas sensors; The concentration calculation formula is determined based on the type of gas to be measured and the operating temperature of the selected thermal conductivity gas sensor. Substituting the output voltage into the concentration calculation formula, the gas concentration of the gas to be measured is obtained.

4. The gas identification method according to claim 1, characterized in that, It also includes the following steps: Based on the type of gas to be measured and the operating temperatures of the N thermal conductivity gas sensors, determine N concentration calculation formulas; The output voltage V of the N thermal conductivity gas sensors T1 ···V TN Substitute these values ​​into their respective concentration calculation formulas to obtain N gas concentration values ​​X. T1 ···X TN ; X of the N gas concentration values T1 ···X TN The gas concentration of the gas to be tested is obtained by averaging.

Images