Multi-component gas detection method and system

Abstract

The invention relates to the technical field of gas detection, and provides a multi-component gas detection method and system.The method comprises the steps that an MEMS-MOS gas sensor array is designed to be composed of a plurality of miniature gas sensing units; each temperature control loop is used for controlling each micro gas sensing unit to maintain the working temperature aiming at the target gas, so that each micro gas sensing unit generates differentiated and complementary dynamic resistance response signals in a mixed gas environment; collecting the resistance change of each micro gas sensing unit in the gas exposure process in real time, and generating a multi-channel digital time sequence data flow; reconstructing the multi-channel digital time sequence data stream into a two-dimensional matrix of channel number * time step; and inputting the two-dimensional matrix into a trained lightweight convolutional neural network model, and outputting gas types and corresponding concentration values contained in the mixed gas. According to the invention, rapid and accurate detection of multi-component gas can be realized in a complex environment, and the sensor has the characteristics of high sensitivity, low power consumption, suitability for portable deployment and the like.

Description

Technical Field

[0001] This invention relates to the field of gas detection technology, and in particular to a method and system for detecting multi-component gases. Background Technology

[0002] With the rapid development of environmental monitoring, industrial safety, medical diagnosis, and smart homes, higher demands are being placed on gas detection technology, especially the increasingly urgent need for simultaneous, rapid, and accurate identification of multiple gas components in complex environments. Traditional gas detection methods mainly rely on single gas sensors or simple sensor combinations, typically employing principles such as electrochemistry, infrared absorption, catalytic combustion, or metal oxide semiconductors (MOS). Among these, MOS-based gas sensors are widely used due to their simple structure, low cost, and high sensitivity.

[0003] However, MOS sensors generally suffer from poor selectivity—they respond to multiple gases, making it difficult to distinguish specific gas types. This is especially problematic in mixed gas environments, where cross-interference can easily occur, leading to misjudgments or inaccurate quantification. To improve selectivity, existing technologies often employ methods such as heat modulation, doping modification, or sensor array construction. While heat modulation gives a single sensor more response characteristics, it significantly increases the challenge of long-term stability. Doping modification can improve the selectivity of a sensor for a specific target gas, but it still cannot avoid interference from other components in the mixed gas. Electronic noses composed of sensor arrays can output combined gas-sensitive response signals, which are more advantageous for identifying mixed gases. However, most of these sensors are manufactured using ceramic processes, resulting in large size, high power consumption, and difficulty in mass integration. Furthermore, most systems only use shallow machine learning models (such as PCA, LDA, and SVM), which have limited ability to resolve high-dimensional nonlinear response signals, making it difficult to meet the requirements for real-time and accurate multi-component gas identification. Summary of the Invention

[0004] The purpose of this invention is to solve at least one technical problem in the background art and to provide a method and system for detecting multi-component gases.

[0005] To achieve the above objectives, the present invention provides a method for detecting multi-component gases, comprising: Design a MEMS-MOS gas sensor array, which consists of multiple miniature gas sensing units; Each micro gas sensing unit is equipped with an independent temperature control loop. Each temperature control loop controls the micro gas sensing unit to maintain the working temperature for the target gas, so that each micro gas sensing unit generates differentiated and complementary dynamic resistance response signals in the mixed gas environment. The resistance changes of each micro gas sensing unit are collected in real time when it comes into contact with multi-component gas. The current / voltage signals are obtained through constant voltage or constant current excitation and then sent to the analog-to-digital converter after being amplified and filtered by a low-noise analog front end to generate a multi-channel digital time-series data stream. The multi-channel digital time-series data stream is reconstructed into a two-dimensional matrix of channel number × time step; The two-dimensional matrix is ​​input into the trained lightweight convolutional neural network model, which outputs the types of gases contained in the gas mixture and their corresponding concentrations.

[0006] According to one aspect of the present invention, each of the micro gas sensing units is fabricated on a silicon substrate using standard MEMS surface micromachining processes, and includes a suspended thermal isolation film structure, a micro heater integrated within the suspended thermal isolation film structure, a temperature sensor, and a metal oxide semiconductor sensitive material layer coated above the heating area. Each of the aforementioned micro gas sensing units is independent of the others, and each of the aforementioned micro gas sensing units is integrated into the same chip or package to form an array layout; The number of the micro gas sensing units is 2. N , where N is the number of target gas types contained in the gas mixture.

[0007] According to one aspect of the present invention, the sensitive material of the metal oxide semiconductor sensitive material layer is selected from one or more of SnO2, WO3, ZnO, and In2O3, and then the surface functionalization modification by doping with noble metals, heterojunction, or core-shell structure is used to adjust the selective response characteristics to different gases.

[0008] According to one aspect of the invention, the temperature control circuit is composed of an external temperature control circuit that cooperates with the micro heater.

[0009] According to one aspect of the present invention, the reconstruction of the multi-channel digital time-series data stream into a two-dimensional matrix of channel number × time step is as follows: Signal preprocessing operations are performed on multi-channel digital time-series data streams, including baseline drift correction, normalization, time alignment, and feature map construction, thereby reconstructing the multi-channel digital time-series data streams into a two-dimensional matrix response map in the form of channel number × time step.

[0010] According to one aspect of the present invention, the lightweight convolutional neural network model includes multiple convolutional layers, multiple pooling layers, and multiple fully connected output layers, with each convolutional layer followed by a pooling layer; Each convolutional layer is followed by a ReLU activation function, and the pooling layer is used to reduce the dimensionality of the feature map activated by the ReLU activation function. Each fully connected output layer uses a linear function. When training the lightweight convolutional neural network model, mean squared error is used as the loss function, the Adam optimizer is used for parameter updates, and the learning rate is set to 0.01.

[0011] To achieve the above objectives, the present invention also provides a multi-component gas detection system, comprising: MEMS-MOS gas sensor array module, design of MEMS-MOS gas sensor array, the MEMS-MOS gas sensor array is composed of multiple miniature gas sensing units; The temperature control module equips each micro gas sensing unit with an independent temperature control loop. Each temperature control loop controls each micro gas sensing unit to maintain the working temperature for the target gas, enabling each micro gas sensing unit to generate differentiated and complementary dynamic resistance response signals in the mixed gas environment. The signal acquisition module collects the resistance changes of each micro gas sensing unit in real time when it comes into contact with multi-component gas. It acquires current / voltage signals through constant voltage or constant current excitation, and then sends them to the analog-to-digital converter after amplification and filtering by a low-noise analog front end to generate a multi-channel digital time-series data stream. The data processing module reconstructs the multi-channel digital time-series data stream into a two-dimensional matrix of channel number × time step; The intelligent recognition module inputs a two-dimensional matrix into a trained lightweight convolutional neural network model and outputs the types of gases contained in the gas mixture and their corresponding concentrations.

[0012] To achieve the above objectives, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the multi-component gas detection method as described above.

[0013] To achieve the above objectives, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the multi-component gas detection method as described above.

[0014] According to the present invention, the present invention utilizes a MEMS-MOS gas sensor array to generate cross-complementary gas-sensitive signals for a mixed gas, and performs pattern recognition and analysis on the gas-sensitive signals through a convolutional neural network to simultaneously achieve qualitative and quantitative detection of multiple gas components in the mixed gas.

[0015] This invention utilizes MEMS technology and the selection of cross-sensitive gas-sensitive materials to construct a miniaturized MOS sensor array. This not only significantly reduces system size and manufacturing costs but also achieves independent and precise temperature control for each sensing unit through on-chip integrated microheaters combined with a dedicated temperature-controlled ASIC chip. This ensures that each unit operates at its optimal operating temperature for its specific gas, maximizing response variability and selectivity, and forming a more discriminative high-dimensional response fingerprint. Furthermore, a convolutional neural network (CNN) is introduced into the gas signal processing flow. The time-series response data output from the sensor array is reconstructed into an image-like input network. Leveraging the powerful spatial-temporal feature extraction capabilities of CNN, end-to-end simultaneous analysis of multiple components in a mixed gas is achieved. This invention effectively overcomes the limitations of traditional algorithms in terms of reliance on manual feature engineering and weak generalization ability by employing step-by-step identification and concentration regression. The entire system features high sensitivity, rapid response, and low power consumption, along with excellent portability and edge deployment capabilities, making it suitable for various scenarios such as wearable devices, industrial site monitoring, environmental emergency response, and medical breath diagnostics. Thanks to the compatibility of MEMS and CMOS processes, the system can also achieve large-scale mass production, further reducing unit costs and enhancing commercial viability. In summary, this invention offers significant advantages over existing technologies in terms of detection performance, system integration, energy consumption control, environmental adaptability, and economic efficiency, providing an efficient, reliable, and scalable technical path for intelligent sensing of multi-component gases. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a multi-component gas detection method according to an embodiment of the present invention is shown. Figure 2 This is a planar layout diagram of the MEMS-MOS gas sensor array in Example 1; Figure 3 This is a graph showing the error between the predicted and actual benzene gas concentrations in Example 1. Figure 4 This is a graph showing the error between the predicted and actual formaldehyde concentrations in Example 1. Figure 5 This is a graph showing the error between the predicted and actual ammonia gas concentrations in Example 1. Detailed Implementation

[0017] The invention will now be discussed with reference to exemplary embodiments. It should be understood that the described embodiments are merely intended to enable those skilled in the art to better understand and thus implement the invention, and are not intended to imply any limitation on the scope of the invention.

[0018] As used herein, the term "comprising" and its variations are to be interpreted as open-ended terms meaning "including but not limited to". The term "based on" is to be interpreted as "at least partially based on". The terms "one embodiment" and "an embodiment" are to be interpreted as "at least one embodiment".

[0019] Figure 1 A flowchart illustrating a multi-component gas detection method according to an embodiment of the present invention is shown. Figure 1 As shown, in this embodiment, the multi-component gas detection method includes: Design a MEMS-MOS gas sensor array, which consists of multiple miniature gas sensing units; Each micro gas sensing unit is equipped with an independent temperature control loop. Each temperature control loop controls the micro gas sensing unit to maintain the working temperature for the target gas, so that each micro gas sensing unit generates differentiated and complementary dynamic resistance response signals in the mixed gas environment. The resistance changes of each micro gas sensing unit are collected in real time when it comes into contact with multi-component gas. The current / voltage signals are obtained through constant voltage or constant current excitation and then sent to the analog-to-digital converter (ADC) after being amplified and filtered by a low-noise analog front end to generate a high-resolution (≥12-bit) multi-channel digital timing data stream. The multi-channel digital time-series data stream is reconstructed into a two-dimensional matrix of channel number × time step; The two-dimensional matrix is ​​input into a lightweight convolutional neural network model trained offline. The lightweight convolutional neural network model performs end-to-end inference on the input two-dimensional matrix and directly outputs the types of gases contained in the mixed gas (qualitative identification) and their corresponding concentration values ​​(quantitative estimation).

[0020] Furthermore, according to one embodiment of the present invention, each micro gas sensing unit is manufactured on a silicon substrate using standard MEMS surface micromachining technology, and includes a suspended thermal isolation film structure, a micro heater integrated within the suspended thermal isolation film structure, a temperature sensor, and a metal oxide semiconductor sensitive material layer coated above the heating area. Each micro gas sensing unit is physically independent, and they are integrated into the same chip or package to form a compact array layout. The number of miniature gas sensing units is 2 N , where N is the number of target gas types contained in the gas mixture.

[0021] Furthermore, according to one embodiment of the present invention, the sensitive material of the metal oxide semiconductor sensitive material layer is selected from one or more common metal oxides such as SnO2, WO3, ZnO, and In2O3, and then the selective response characteristics to different gases are adjusted by surface functionalization modification through doping with noble metals (such as Pt, Pd, Au), heterojunction, and core-shell structure.

[0022] Furthermore, according to one embodiment of the present invention, the temperature control loop is composed of an external temperature control circuit that cooperates with the micro heater, which can stably maintain the corresponding micro gas sensing unit at its optimal operating temperature for the target gas response (typically between 200°C and 400°C) according to a preset strategy, thereby generating differentiated and complementary dynamic resistance response signals in a mixed gas environment.

[0023] Furthermore, according to one embodiment of the present invention, the multi-channel digital time-series data stream is reconstructed into a two-dimensional matrix of channel number × time step as follows: Signal preprocessing operations are performed on multi-channel digital time-series data streams, including baseline drift correction, normalization, time alignment, and feature map construction, thereby reconstructing the multi-channel digital time-series data streams into a two-dimensional matrix response map in the form of channel number × time step.

[0024] Furthermore, according to one embodiment of the present invention, the lightweight convolutional neural network model includes multiple convolutional layers, multiple pooling layers, and multiple fully connected output layers, with each convolutional layer followed by a pooling layer; Each convolutional layer is followed by a ReLU activation function, each pooling layer is used to reduce the dimensionality of the feature maps activated by the ReLU activation function, and each fully connected output layer uses a linear function. When training the lightweight convolutional neural network model, mean squared error is used as the loss function, the Adam optimizer is used for parameter updates, and the learning rate is set to 0.01.

[0025] According to the above-described scheme of the present invention, this invention provides a miniature intelligent detection method for intelligent identification of multi-component gases by combining a MEMS-based metal-oxide-semiconductor (MOS) gas sensor array with a convolutional neural network (CNN) algorithm. The development of MEMS technology enables the miniaturization and low-power consumption of MOS sensors, and supports the integration of temperature control units, thereby allowing different sensing units to operate at their respective optimal operating temperatures and improving response specificity. At the same time, artificial intelligence algorithms, especially convolutional neural networks (CNNs), have demonstrated powerful capabilities in image and signal pattern recognition, and can also greatly enhance the combined signal processing capabilities of the gas sensor array output, thereby strengthening the olfactory system's qualitative and quantitative identification capabilities of mixed gases.

[0026] This invention can screen out gas-sensitive materials with cross-complementarity based on the composition of a mixed gas. If the mixed gas contains N kinds of gases, then 2N kinds of gas-sensitive materials with cross-complementarity need to be screened to construct a MOS sensor array. By combining the response differences of each sensor at different optimal operating temperatures, a high-dimensional feature fingerprint is formed.

[0027] This invention utilizes MEMS technology to highly integrate the sensor and heater, significantly reducing system size and power consumption.

[0028] This invention optimizes the temperature control strategy by configuring an independent temperature control loop for each sensor unit in the array, that is, equipping the sensor array with a dedicated temperature control ASIC chip, so that it operates at the temperature point most sensitive to the target gas, thereby enhancing the response specificity.

[0029] This invention introduces a convolutional neural network (CNN) to perform end-to-end learning of the time-series response signal output by the sensor array, automatically extracting spatial-temporal features, achieving highly robust multi-gas synchronous identification and concentration estimation, and improving the system's ability to perform qualitative and quantitative analysis of mixed gases.

[0030] This invention enables rapid and accurate detection of multi-component gases in complex environments, while also featuring high sensitivity, good selectivity, low power consumption, low cost, small size, and suitability for portable deployment.

[0031] Furthermore, to achieve the above objectives, the present invention also provides a multi-component gas detection system, comprising: MEMS-MOS gas sensor array module, design of MEMS-MOS gas sensor array, the MEMS-MOS gas sensor array is composed of multiple miniature gas sensing units; The temperature control module equips each micro gas sensing unit with an independent temperature control loop. Each temperature control loop controls each micro gas sensing unit to maintain the working temperature for the target gas, enabling each micro gas sensing unit to generate differentiated and complementary dynamic resistance response signals in the mixed gas environment. The signal acquisition module collects the resistance changes of each micro gas sensing unit in real time when it comes into contact with multi-component gas. It acquires current / voltage signals through constant voltage or constant current excitation, and then sends them to the analog-to-digital converter after amplification and filtering by a low-noise analog front end to generate a multi-channel digital time-series data stream. The data processing module reconstructs the multi-channel digital time-series data stream into a two-dimensional matrix of channel number × time step; The intelligent recognition module inputs a two-dimensional matrix into a trained lightweight convolutional neural network model and outputs the types of gases contained in the gas mixture and their corresponding concentrations.

[0032] The multi-component gas detection system according to the present invention can realize the above-mentioned multi-component gas detection method. The specific process steps are as described above and will not be repeated here.

[0033] Furthermore, to achieve the above objectives, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the multi-component gas detection method as described above.

[0034] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the multi-component gas detection method as described above.

[0035] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely one preferred embodiment of the invention and are only used to explain the invention. They do not limit the scope of protection of the invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0036] Example 1 This embodiment applies a typical application scenario of gas safety monitoring during the conductive adhesive curing process in electronic manufacturing. In this scenario, the conductive adhesive releases various harmful volatile organic compounds during heat curing, mainly including three target gases: benzene (C6H6), formaldehyde (HCHO), and ammonia (NH3) (i.e., N=3). To achieve high-precision simultaneous identification and concentration estimation of these three components, such as... Figure 2 As shown, this embodiment constructs a MEMS-MOS gas sensor array containing 8 independent micro gas sensing units based on the 2N optimization principle. Each micro gas sensing unit adopts different metal oxide semiconductor materials and noble metal doping strategies to form a response fingerprint with cross-sensitivity but complementary discrimination capabilities.

[0037] Specifically, the eight gas-sensitive materials are Pt-SnO2, Pd-SnO2, Au-WO3, Pd-ZnO, Pt-MoO3, Cu-MoO3, Pt-In2O3, and NiO-In2O3. These eight materials are coated onto eight independent suspended thermally insulating film structures, forming a single-chip integrated sensor array. Each miniature gas sensing unit is equipped with a miniature heater and a temperature sensor, and its independent temperature setting is achieved through a dedicated temperature control ASIC chip. Preliminary experimental calibration shows that the optimal operating temperatures for each miniature gas sensing unit are: 250°C (Pt-SnO2), 275°C (Pd-SnO2), 250°C (Au-WO3), 300°C (Pd-ZnO), 225°C (Pt-MoO3), 250°C (Cu-MoO3), 200°C (Pt-In2O3), and 225°C (NiO-In2O3).

[0038] The training of the intelligent recognition module is based on a sample set constructed from a standard gas mixture of known concentrations. The construction process is as follows: multiple sets of mixed gases with different preset concentrations of ammonia, formaldehyde, and benzene are generated through a flow control device; subsequently, the MEMS-MOS gas sensor array is sequentially exposed to these standard gas samples according to a fixed procedure, which includes introducing clean air to reset the sensor, introducing standard gas samples to excite the response, and introducing clean air again to prepare for the next measurement; the response data of the sensor array in each measurement is fully recorded, and the response data is correlated with its corresponding standard gas concentration information to form a sample dataset for training the convolutional neural network model.

[0039] The signal acquisition module continuously acquires the sensor array response signal at a sampling rate of 5Hz, obtaining raw time-series data of 8 channels × 12000 points. The data processing module first performs baseline correction and normalization, and then extracts key dynamic features for each channel: including the maximum slope of the rising edge (reflecting the response speed), the ratio of the steady-state response amplitude to the baseline (reflecting sensitivity), and the maximum slope of the falling edge (reflecting desorption characteristics). These features are then concatenated with the raw time-series segments to form an enhanced input matrix. To further reduce dimensionality and retain key information, the matrix is ​​compressed using sym4 wavelet transform and a 1D convolutional autoencoder, outputting a feature map (two-dimensional matrix) with a dimension of 8×64.

[0040] The feature map was fed into a lightweight convolutional neural network model deployed on an ARM Cortex-M7 microcontroller. The lightweight convolutional neural network model consists of three convolutional layers, three pooling layers, and three fully connected output layers stacked sequentially, with each convolutional layer followed by a pooling layer. Small convolutional kernels are used to reduce the number of trainable parameters and improve training efficiency, while each pooling layer halves the spatial dimension of the feature map. The final pooling output has 32 channels, is 8 in size, flattened into a 256-dimensional vector, and fed into the fully connected layer. Dropout is set to 0.2 in the fully connected layer to prevent overfitting. A linear function is used in the output layer to support regression. The model uses mean squared error (MSE) as the loss function, the Adam optimizer for parameter updates, a learning rate of 0.01, and is trained for 2000 epochs. The model's output vector is the concentration values ​​of the three target gases, enabling simultaneous quantitative analysis of the concentrations of multiple components in a gas mixture. The model was trained and tested on the aforementioned sample set (sample dataset), as shown below. Figure 3 , Figure 4 and Figure 5 As shown, the results indicate that the average relative error in estimating the concentrations of benzene, formaldehyde, and ammonia is less than 10%. The entire system consumes approximately 200 mW in detection mode and less than 20 mW in standby mode. Its size is no larger than 15 mm × 15 mm × 8 mm, and it can be powered via USB or battery. This embodiment fully verifies the feasibility and practicality of this invention in achieving highly selective, low-power, and miniaturized intelligent detection of multi-component gases in complex industrial environments.

[0041] Those skilled in the art will recognize that the modules and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented using electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0042] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described apparatus and equipment can be referred to the corresponding process in the foregoing method implementation, and will not be repeated here.

[0043] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0044] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the objectives of the embodiments of the present invention, depending on actual needs.

[0045] In addition, the functional modules in the embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0046] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the sending / receiving methods of various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0047] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in this application.

[0048] It should be understood that the sequence number of each step in the invention and its embodiments does not absolutely imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

Claims

1. A method for detecting multi-component gases, characterized in that, include: Design a MEMS-MOS gas sensor array, which consists of multiple miniature gas sensing units; Each micro gas sensing unit is equipped with an independent temperature control loop. Each temperature control loop controls the micro gas sensing unit to maintain the working temperature for the target gas, so that each micro gas sensing unit generates differentiated and complementary dynamic resistance response signals in the mixed gas environment. The resistance changes of each micro gas sensing unit are collected in real time when it comes into contact with multi-component gas. The current / voltage signals are obtained through constant voltage or constant current excitation and then sent to the analog-to-digital converter after being amplified and filtered by a low-noise analog front end to generate a multi-channel digital time-series data stream. The multi-channel digital time-series data stream is reconstructed into a two-dimensional matrix of channel number × time step; The two-dimensional matrix is ​​input into the trained lightweight convolutional neural network model, which outputs the types of gases contained in the gas mixture and their corresponding concentrations.

2. The multi-component gas detection method according to claim 1, characterized in that, Each of the aforementioned micro gas sensing units is fabricated on a silicon substrate using standard MEMS surface micromachining technology, and includes a suspended thermal isolation film structure, a micro heater integrated within the suspended thermal isolation film structure, a temperature sensor, and a metal oxide semiconductor sensitive material layer coated above the heating area. Each of the aforementioned micro gas sensing units is independent of the others, and each of the aforementioned micro gas sensing units is integrated into the same chip or package to form an array layout; The number of the micro gas sensing units is 2. N , where N is the number of target gas types contained in the gas mixture.

3. The multi-component gas detection method according to claim 2, characterized in that, The sensitive material of the metal oxide semiconductor sensitive material layer is selected from one or more of SnO2, WO3, ZnO, and In2O3. Then, the selective response characteristics to different gases are adjusted by surface functionalization modification methods such as doping with noble metals, heterojunction, and core-shell structure.

4. The multi-component gas detection method according to claim 2, characterized in that, The temperature control circuit consists of an external temperature control circuit that works in conjunction with the micro heater.

5. The multi-component gas detection method according to claim 1, characterized in that, The reconstruction of the multi-channel digital time-series data stream into a two-dimensional matrix of channel number × time step is as follows: Signal preprocessing operations are performed on multi-channel digital time-series data streams, including baseline drift correction, normalization, time alignment, and feature map construction, thereby reconstructing the multi-channel digital time-series data streams into a two-dimensional matrix response map in the form of channel number × time step.

6. The method for detecting multi-component gases according to any one of claims 1-5, characterized in that, The lightweight convolutional neural network model includes multiple convolutional layers, multiple pooling layers, and multiple fully connected output layers, with each convolutional layer followed by a pooling layer. Each convolutional layer is followed by a ReLU activation function, and the pooling layer is used to reduce the dimensionality of the feature map activated by the ReLU activation function. Each fully connected output layer uses a linear function. When training the lightweight convolutional neural network model, mean squared error is used as the loss function, the Adam optimizer is used for parameter updates, and the learning rate is set to 0.

01.

7. A multi-component gas detection system, characterized in that, include: MEMS-MOS gas sensor array module, design of MEMS-MOS gas sensor array, the MEMS-MOS gas sensor array is composed of multiple miniature gas sensing units; The temperature control module equips each micro gas sensing unit with an independent temperature control loop. Each temperature control loop controls each micro gas sensing unit to maintain the working temperature for the target gas, enabling each micro gas sensing unit to generate differentiated and complementary dynamic resistance response signals in the mixed gas environment. The signal acquisition module collects the resistance changes of each micro gas sensing unit in real time when it comes into contact with multi-component gas. It acquires current / voltage signals through constant voltage or constant current excitation, and then sends them to the analog-to-digital converter after amplification and filtering by a low-noise analog front end to generate a multi-channel digital time-series data stream. The data processing module reconstructs the multi-channel digital time-series data stream into a two-dimensional matrix of channel number × time step; The intelligent recognition module inputs a two-dimensional matrix into a trained lightweight convolutional neural network model and outputs the types of gases contained in the gas mixture and their corresponding concentrations.

8. An electronic device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the multi-component gas detection method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the multi-component gas detection method as described in any one of claims 1-6.

Images