Detection Method and Device Based on Multimodal Gas Sensor

Abstract

The present invention relates to the technical field of sensor detection, and discloses a detection method and device based on a multimodal gas sensor. The method includes: performing co-mixing treatment on a metal oxide semiconductor and a conductive material to obtain a bimodal nanocomposite and creating a multimodal gas detection module, and collecting preprocessed multimodal signal data; collecting and analyzing temperature sensor and humidity sensor data to obtain environmental factor compensation parameters; performing adaptive calibration processing on the preprocessed multimodal signal data according to the environmental factor compensation parameters to obtain calibrated multimodal signal data; constructing a multi-input neural network model based on the calibrated multimodal signal data and training it to obtain a gas recognition model; analyzing and judging the real-time collected and calibrated multimodal signals based on the gas recognition model to obtain gas type and concentration information. The present invention realizes high-precision and high-reliability gas detection.

Description

Technical Field

[0001] The present invention relates to the technical field of sensor detection, and in particular, to a detection method and device based on a multimodal gas sensor. Background Art

[0002] With the continuous growth of industrial production and environmental monitoring requirements, gas detection technology plays an increasingly important role in many fields. Traditional single-modal gas sensors have obvious limitations in detection accuracy, anti-interference ability, and identification of complex gas mixtures. This has prompted researchers to continuously explore new multimodal gas sensing technologies to improve the accuracy and reliability of detection. However, current multimodal gas sensors still face many challenges in material design, signal processing, and pattern recognition.

[0003] Existing multimodal gas sensors often adopt simple material composite methods, which are difficult to fully exert the synergistic effect of each component, resulting in limited improvement in sensing performance. At the same time, traditional signal processing methods are difficult to effectively process complex noise and interference in multimodal data, affecting the stability of detection results. In addition, existing gas identification algorithms often perform poorly when dealing with high-dimensional and strongly correlated multimodal data, and it is difficult to accurately distinguish similar gases or low-concentration gases. Summary of the Invention

[0004] The present invention provides a detection method and device based on a multimodal gas sensor, and the present invention realizes high-precision and high-reliability gas detection.

[0005] In a first aspect, the present invention provides a detection method based on a multimodal gas sensor, and the detection method based on a multimodal gas sensor includes:

[0006] Co-mix and process a metal oxide semiconductor and a conductive material to obtain a bimodal nanocomposite, and create a sensor array with a three-dimensional microchannel network based on the bimodal nanocomposite;

[0007] Package and integrate the sensor array to obtain a multimodal gas detection module, and perform synchronous acquisition and digital filtering processing on resistance signals and capacitance signals based on the multimodal gas detection module to obtain preprocessed multimodal signal data;

[0008] Collect and analyze the data of the temperature sensor and humidity sensor in the multimodal gas detection module to obtain environmental factor compensation parameters;

[0009] Perform adaptive calibration processing on the preprocessed multimodal signal data according to the environmental factor compensation parameters to obtain calibrated multimodal signal data;

[0010] Construct a multi-input neural network model based on the calibrated multi-modal signal data and train it to obtain a gas recognition model;

[0011] Analyze and judge the real-time collected and calibrated multi-modal signals based on the gas recognition model to obtain gas type and concentration information.

[0012] In a second aspect, the present invention provides a detection device based on a multi-modal gas sensor. The detection device based on a multi-modal gas sensor includes:

[0013] A creation module for co-mixing a metal oxide semiconductor and a conductive material to obtain a bimodal nanocomposite, and creating a sensor array with a three-dimensional microchannel network based on the bimodal nanocomposite;

[0014] An acquisition module for encapsulating and integrating the sensor array to obtain a multi-modal gas detection module, and synchronously acquiring and digitally filtering resistance signals and capacitance signals based on the multi-modal gas detection module to obtain preprocessed multi-modal signal data;

[0015] An analysis module for acquiring and analyzing the data of the temperature sensor and the humidity sensor in the multi-modal gas detection module to obtain environmental factor compensation parameters;

[0016] A calibration module for adaptively calibrating the preprocessed multi-modal signal data according to the environmental factor compensation parameters to obtain calibrated multi-modal signal data;

[0017] A training module for constructing a multi-input neural network model based on the calibrated multi-modal signal data and training it to obtain a gas recognition model;

[0018] A processing module for analyzing and judging the real-time collected and calibrated multi-modal signals based on the gas recognition model to obtain gas type and concentration information.

[0019] In a third aspect of the present invention, there is provided a computer device, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor invokes the instructions in the memory so that the computer device executes the above-mentioned detection method based on a multi-modal gas sensor.

[0020] In a fourth aspect of the present invention, there is provided a computer-readable storage medium, wherein instructions are stored in the computer-readable storage medium, and when it runs on a computer, it causes the computer to execute the above-mentioned detection method based on a multi-modal gas sensor.

[0021] In the technical solution provided by the present invention, for the design and preparation technology of the dual-modal nanocomposite material, through the synergistic effect of Fe-doped SnO 2 nanoparticles and graphene, the gas-sensing performance and selectivity of the sensing material are significantly improved. The design and manufacture of the three-dimensional microchannel network structure increase the contact area between the gas and the sensitive material, improve the response speed and sensitivity of the sensor, and are conducive to the generation and transmission of multi-modal signals. The application of the temperature and humidity compensation technology effectively reduces the influence of environmental changes on the detection results and improves the stability and reliability of the sensor in complex environments through a dynamic environmental factor compensation model. The adaptive calibration processing technology combines methods such as wavelet packet transform and empirical mode decomposition to achieve precise calibration of multi-modal signals, effectively improving the signal-to-noise ratio and feature expression ability of the signals. The design of the multi-input neural network model integrates advanced algorithms such as multi-scale residual network, bidirectional gated recurrent unit network, and multi-head self-attention mechanism, significantly improving the recognition ability of complex gases and the concentration estimation accuracy. The adoption of the multi-task learning strategy simultaneously optimizes the gas classification and concentration regression tasks, improves the generalization ability and robustness of the model, enabling it to better handle various complex situations in practical applications. The present invention realizes high-precision and high-reliability gas detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0023] Figure 1 It is a schematic diagram of the steps of the detection method based on the multi-modal gas sensor in the embodiment of the present invention;

[0024] Figure 2 It is a schematic diagram of the structure of the detection device based on the multi-modal gas sensor in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] An embodiment of the present invention provides a detection method and device based on a multimodal gas sensor. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and above-mentioned drawings of the present invention are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order different from that shown or described herein. In addition, the term "comprising" or "having" and any variation thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0026] For ease of understanding, the specific process of the embodiment of the present invention will be described below. Please refer to Figure 1 , an embodiment of the detection method based on a multimodal gas sensor in the embodiment of the present invention includes:

[0027] Step S1: Co-mix and process a metal oxide semiconductor and a conductive material to obtain a bimodal nanocomposite material, and create a sensor array with a three-dimensional microchannel network based on the bimodal nanocomposite material;

[0028] It can be understood that the execution subject of the present invention can be a detection device based on a multimodal gas sensor, or a terminal or a server. Specifically, it is not limited here. The embodiment of the present invention takes the server as the execution subject for illustration.

[0029] Specifically, a tin source and an iron source are mixed according to a preset molar ratio to obtain a metal precursor solution, and the solution is placed in a specific reaction vessel. Through hydrothermal reaction treatment, metal ions react under high temperature and high pressure conditions to form Fe-doped SnO 2 nanoparticles. Tin oxide nanoparticles doped with iron elements have excellent gas-sensing characteristics and are one of the basic materials for forming a bimodal nanocomposite material. Graphene is ultrasonically dispersed to obtain a uniform graphene suspension. The previously prepared Fe-doped SnO 2 nanoparticles and the graphene suspension are fully mixed according to a preset mass ratio to form a composite nanomaterial suspension with both good conductivity and gas-sensing performance. The composite nanomaterial suspension is processed by spray drying technology to rapidly evaporate the solvent under the action of hot air to form a uniform granular substance, and then calcination treatment is carried out to remove possible residual organic substances and enhance the crystallinity of the material to obtain a stable bimodal nanocomposite material. The silicon substrate is oxidized to form a uniform SiO 2An insulating layer to improve the electrical stability and isolation effect of the sensor. Based on a substrate with an SiO 2 insulating layer, photolithography and metal deposition processes are carried out. By this method, a designed comb-shaped electrode pattern is formed on the substrate surface, and this electrode structure helps to enhance the electrical response of the sensor. After the preparation of the electrodes, laser direct writing technology is used to process the substrate with the comb-shaped electrodes, and a complex three-dimensional microchannel network structure is formed on the substrate surface. This three-dimensional structure can significantly improve the gas diffusion efficiency and the sensitivity of the sensor. The prepared bimodal nanocomposite is redispersed to obtain a nanomaterial suspension. Through inkjet printing technology, the suspension is precisely deposited onto the substrate structure with the three-dimensional microchannel network to achieve precise positioning and uniform distribution of the materials. The substrate structure deposited with the nanomaterials is heat-treated to remove the possibly remaining solvent and enhance the adhesion and stability of the materials, forming a sensor array with a three-dimensional microchannel network.

[0030] Step S2: Package and integrate the sensor array to obtain a multimodal gas detection module, and based on the multimodal gas detection module, synchronously collect and digitally filter the resistance signal and the capacitance signal to obtain the preprocessed multimodal signal data;

[0031] Specifically, precision machining is performed on the low-temperature co-fired ceramic material to fabricate a sensor housing with preset air inlets and outlets, ensuring uniform gas flow over the sensor array and improving the accuracy of detection. The sensor array is fixed inside the sensor housing, and through precise positioning and bonding processes, a sensor assembly tightly integrated with the sensor housing is formed to ensure its stability in complex environments. For the sensor assembly, ultrasonic bonding technology is used to connect the electrodes of the sensor array to external leads, forming a fully functional sensor unit. At the same time, the temperature sensor and humidity sensor are integrally installed to obtain a multimodal gas detection module. Based on the multimodal gas detection module, a micro air pump and a gas distribution system are designed to ensure uniform gas distribution inside the sensor, improving the stability and accuracy of sensor response. Meanwhile, for the electrical signal processing requirements of the sensor system, the design and installation of a signal conditioning circuit are carried out to construct a signal processing unit with functions of pre-amplification, filtering, and analog-to-digital conversion, which can effectively enhance weak signals and filter out interference. By programming and configuring the microcontroller, a data acquisition and communication control system is formed. Through reasonable software and hardware design, this system realizes the synchronous acquisition of resistance signals and capacitance signals, and through parallel data acquisition settings, ensures the synchronous acquisition and processing of multimodal signals. Based on the synchronous acquisition scheme, the collected original signals are processed by Butterworth low-pass filtering to remove high-frequency noise, reduce interference components in the signals, and retain valid information. For the signals after filtering processing, wavelet transform technology is used for baseline drift removal processing, effectively eliminating baseline drift in the signals and obtaining more stable and accurate signal data. The preprocessed multimodal signal data is obtained.

[0032] Step S3: Collect and analyze the data of the temperature sensor and humidity sensor in the multimodal gas detection module to obtain environmental factor compensation parameters;

[0033] Specifically, data sampling is performed on the temperature sensor and the humidity sensor to obtain the original temperature and humidity time series data, which reflects the dynamic changes of the environmental temperature and humidity over time. Median filtering is performed on the original temperature and humidity time series data to remove the outliers. The filtering process can effectively eliminate the data deviation caused by occasional interference or sensor noise, and obtain smoother and more accurate temperature and humidity data. Based on the temperature and humidity data with outliers removed, fast Fourier transform is performed to obtain the spectral characteristics of the temperature and humidity data. By analyzing the spectral characteristics, the main periodic components and characteristic frequencies in the temperature and humidity changes are identified, and the target pattern of the temperature and humidity changes is extracted. The target pattern reflects the main law of the temperature and humidity changing over time. An autoregressive moving average model is constructed based on the target pattern to predict the future change trend of the temperature and humidity. Kalman filtering is performed on the constructed temperature and humidity change trend prediction model to obtain the optimized temperature and humidity estimated values. Kalman filtering is a linear optimal estimation method, which can effectively reduce the influence of noise in the prediction and improve the reliability of the estimation result. Based on the optimized temperature and humidity estimated values, the influence coefficient of the temperature and humidity on the sensitivity of the gas sensor is calculated to obtain the initial compensation factor, so as to correct the influence of the environmental temperature and humidity changes on the detection accuracy of the gas sensor. Nonlinear least squares fitting is performed on the initial compensation factor to obtain a more accurate temperature and humidity compensation function. Based on the temperature and humidity compensation function, an adaptive fuzzy neural network is constructed to form a dynamic environmental factor compensation model. This compensation model can automatically adjust the compensation parameters according to the real-time temperature and humidity changes, so as to ensure the accuracy of gas detection. Parameter optimization is performed on the dynamic environmental factor compensation model, and through multiple iterations and adjustments, the optimal environmental factor compensation parameters are obtained.

[0034] Step S4: Perform adaptive calibration processing on the preprocessed multi-modal signal data according to the environmental factor compensation parameters to obtain the calibrated multi-modal signal data;

[0035] Specifically, the preprocessed multi-modal signal data is segmented by time window, divided into multiple signal segments of fixed length, ensuring the temporal consistency of the signal during analysis. Based on the environmental factor compensation parameters, calculate the specific impacts of temperature and humidity on each signal segment, obtaining an environmental impact matrix reflecting the degree of influence of environmental changes on the signal. Use singular value decomposition technology to decompose the environmental impact matrix and extract the target impact patterns of environmental factors from it. The target impact patterns can identify the main impact directions and characteristics of environmental factors on the signal. Based on the target impact patterns, construct an environmental noise estimation model, which filters out the interference of environmental noise on the signal, allowing the effective components of the signal to be retained. To extract the useful information in the signal, perform wavelet packet transform on the signal segments and the environmental noise estimation model to obtain multi-scale time-frequency features. The time-frequency features can not only describe the time and frequency changes of the signal but also reveal the correlation between the signal and environmental noise at different scales. By calculating the mutual information between the multi-scale time-frequency features, obtain the correlation index between the signal and environmental noise, quantifying how much information in the signal is interfered by environmental noise. According to the correlation index, perform adaptive threshold denoising on the signal segments, removing the parts with greater interference from environmental noise, obtaining the preliminarily calibrated signal data. Perform empirical mode decomposition on the preliminarily calibrated signal data to separate the intrinsic mode functions of different frequency components. Each intrinsic mode function represents the main change trend of the signal in a certain frequency band and can help identify the essential characteristics of the signal. Based on the intrinsic mode functions, construct a Hilbert-Huang transform spectrum and extract the instantaneous frequency and amplitude of the signal, reflecting the dynamic change of the signal. Perform non-linear mapping and reconstruction on the instantaneous frequency and amplitude to correct the signal distortion caused by environmental factors, obtaining the calibrated multi-modal signal data.

[0036] Step S5: Based on the calibrated multi-modal signal data, construct a multi-input neural network model and train it to obtain a gas recognition model;

[0037] Specifically, time-frequency domain feature extraction is performed on the calibrated multi-modal signal data to obtain the feature sets of the resistance signal and the capacitance signal. Based on the extracted resistance signal feature set, a first multi-scale residual network branch is constructed. This branch consists of three parallel residual blocks, each containing convolutional layers and pooling layers of different scales. Through the multi-scale processing method, it can capture the feature information of different frequency bands and time domains in the resistance signal, and obtain the trained multi-scale resistance feature representation. Similarly, for the capacitance signal feature set, a second multi-scale residual network branch is constructed. The structure of this branch is the same as that of the resistance signal branch, which can effectively extract the multi-scale features of the capacitance signal and obtain the trained multi-scale capacitance feature representation. To optimize and enhance the feature expression ability, the trained multi-scale resistance feature representation and multi-scale capacitance feature representation are subjected to adaptive feature recalibration. By introducing the channel attention mechanism and the spatial attention mechanism, the features are weighted, and the importance of different features is automatically adjusted to obtain the trained recalibrated features, effectively highlighting the features crucial for gas recognition while suppressing redundant information and improving the overall performance of the model. Based on the trained recalibrated features, a bidirectional gated recurrent unit network is constructed to mine the temporal information in the signal. This network consists of two layers, with 128 units in each layer, and uses the tanh activation function. Through the bidirectional processing method, it can capture the context information of the signal in the time series and obtain the trained temporal context features. The multi-head self-attention mechanism is applied to the trained temporal context features. The mechanism contains 8 attention heads, and the dimension of each head is 64. In this way, the model can understand the dependency relationship between features from a global perspective and obtain the trained global dependency features. Based on the trained global dependency features, a pyramid pooling module is constructed. Through three different scale pooling operations of 1x1, 2x2, and 4x4, multi-scale global information is extracted to obtain the trained multi-scale global features. Feature fusion is performed on the global features. Through element-wise addition and 1x1 convolution operations, the multi-scale features are further integrated to obtain the trained fused feature vector. Based on the trained fused feature vector, a multi-task learning head is constructed to simultaneously handle the gas classification and concentration regression tasks. For the gas classification task, the Softmax activation function is used to obtain the probability distribution of gas types; for the concentration regression task, the linear activation function is used to output the estimated value of the gas concentration. Through the multi-task learning method, the model can simultaneously handle classification and regression problems and improve the recognition accuracy of gas types and concentrations. The entire multi-input neural network model is trained end-to-end, using the weighted cross-entropy loss function and the Huber loss function as the optimization objectives. Through the loss function, the importance of the classification and regression tasks can be balanced, and the AdamW optimizer and the cosine annealing learning rate scheduling strategy are used for parameter update, and finally the trained gas recognition model is obtained.

[0038] Step S6: Analyze and judge the real-time collected and calibrated multimodal signals based on the gas recognition model to obtain gas type and concentration information.

[0039] Specifically, synchronously sample and digitize the real-time collected resistance signal and capacitance signal to obtain the original multimodal signal data, which reflects the real-time response of the sensor to the ambient gas. Perform adaptive calibration processing on the original multimodal signal data to eliminate the influence of environmental factors and obtain the calibrated real-time multimodal signal. Extract time-frequency domain features from the calibrated real-time multimodal signal, and respectively extract the real-time resistance feature set and the real-time capacitance feature set from it. Input the real-time resistance feature set into the first multi-scale residual network branch of the gas recognition model, and through the calculation of three parallel residual blocks in this branch, obtain the real-time multi-scale resistance feature representation. Similarly, input the real-time capacitance feature set into the second multi-scale residual network branch of the gas recognition model, and through the calculation of three parallel residual blocks with the same structure, obtain the real-time multi-scale capacitance feature representation. Perform adaptive feature recalibration on the real-time multi-scale resistance feature representation and the real-time multi-scale capacitance feature representation. Through the weighted processing of the channel attention mechanism and the spatial attention mechanism, obtain the recalibrated feature representation, so that important features are highlighted and irrelevant features are suppressed, improving the sensitivity of the model to key information. Input the recalibrated features into the bidirectional gated recurrent unit (GRU) network, and through sequential calculations of two layers, each layer containing 128 units, capture the context relationship of the signal in the time series to obtain the real-time temporal context features. Apply the multi-head self-attention mechanism to the real-time temporal context features. Through the parallel calculation of 8 attention heads, the model can identify the complex dependence relationships between signals and obtain the real-time global dependence features. Input the global dependence features into the pyramid pooling module, and through pooling operations of three different scales of 1x1, 2x2, and 4x4, extract the multi-scale global information of the signal to obtain the real-time multi-scale global features. Perform feature fusion operations on the real-time multi-scale global features, and through element-wise addition and 1x1 convolution operations, integrate information of different scales to obtain the real-time fusion feature vector. Input the real-time fusion feature vector into the multi-task learning head for parallel gas classification and concentration regression calculations. For the gas classification task, use the Softmax activation function to obtain the probability distribution of gas types; for the concentration regression task, use the linear activation function to output the estimated value of the gas concentration. The model can simultaneously provide gas type recognition and concentration estimation, meeting the requirements of multimodal gas detection. Perform threshold judgment and temporal smoothing processing on the real-time gas type probability distribution and concentration estimated value. Through threshold judgment, filter out low-probability or unreliable results, and through temporal smoothing, eliminate outliers caused by instantaneous fluctuations, and finally obtain stable and reliable gas type and concentration information.

[0040] In the embodiments of the present invention, for the design and preparation technology of the dual-modal nanocomposite material, through the synergistic effect of Fe-doped SnO 2 nanoparticles and graphene, the gas-sensing performance and selectivity of the sensing material are significantly improved. The design and manufacture of the three-dimensional microchannel network structure increase the contact area between the gas and the sensitive material, improve the response speed and sensitivity of the sensor, and are conducive to the generation and transmission of multi-modal signals. The application of the temperature and humidity compensation technology effectively reduces the influence of environmental changes on the detection results through a dynamic environmental factor compensation model, and improves the stability and reliability of the sensor in a complex environment. The adaptive calibration processing technology combines methods such as wavelet packet transform and empirical mode decomposition to achieve precise calibration of multi-modal signals, effectively improving the signal-to-noise ratio and feature expression ability. The design of the multi-input neural network model integrates advanced algorithms such as multi-scale residual network, bidirectional gated recurrent unit network, and multi-head self-attention mechanism, significantly improving the recognition ability of complex gases and the concentration estimation accuracy. The adoption of the multi-task learning strategy simultaneously optimizes the gas classification and concentration regression tasks, improves the generalization ability and robustness of the model, enabling it to better handle various complex situations in practical applications. The present invention realizes high-precision and high-reliability gas detection.

[0041] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0042] Mix the tin source and the iron source according to a preset molar ratio to obtain a metal precursor solution, and perform a hydrothermal reaction treatment on the metal precursor solution to obtain Fe-doped SnO 2 nanoparticles;

[0043] Perform ultrasonic dispersion treatment on graphene to obtain a graphene suspension, and mix the Fe-doped SnO 2 nanoparticles and the graphene suspension according to a preset mass ratio to obtain a composite nanomaterial suspension;

[0044] Perform spray drying and calcination treatments on the composite nanomaterial suspension to obtain a dual-modal nanocomposite material, and perform oxidation treatment on the silicon substrate to obtain a substrate with a SiO 2 insulating layer;

[0045] Based on the substrate with a SiO 2 insulating layer, perform photolithography and metal deposition treatments to obtain a substrate with comb-shaped electrodes, and perform laser direct writing treatment on the substrate with comb-shaped electrodes to obtain a substrate structure with a three-dimensional microchannel network;

[0046] Disperse the bimodal nanocomposite to obtain a nanomaterial suspension, and deposit the nanomaterial suspension onto a substrate structure with a three-dimensional microchannel network through inkjet printing technology. Heat-treat the substrate structure deposited with the nanomaterial to obtain a sensor array with a three-dimensional microchannel network.

[0047] Specifically, precisely proportion the tin source (Sn) and the iron source (Fe). Usually, stannous oxide (SnO) and iron(III) chloride (FeCl 3 ) are used as the tin source and the iron source. In actual operation, according to the preset molar ratio, dissolve SnO and FeCl 3 in deionized water respectively to obtain their respective solutions. By mixing these two solutions, a metal precursor solution is formed, which contains Sn²⁺ and Fe³⁺ ions. In a chemical reaction, the precise control of the molar ratio directly affects the doping concentration of Fe in the final product. Assuming the used molar ratio is x:1, then x represents the number of moles of SnO, and 1 represents the number of moles of FeCl 3 . Controlling an appropriate molar ratio can ensure the uniform distribution of Fe in the SnO 2 lattice, and further regulate the electronic structure and gas-sensing performance of the final material. Put the metal precursor solution into an autoclave for hydrothermal reaction treatment. The hydrothermal reaction is usually carried out at a temperature of 180°C to 200°C. Under this condition, Sn²⁺ and Fe³⁺ ions hydrolyze and precipitate in the solution to form Fe-doped SnO 2 nanoparticles. The process of the hydrothermal reaction can be represented by the following chemical reaction equation:

[0048] ;

[0049] wherein, Sn²⁺ and Fe³⁺ are metal ions in the reaction, H 2 O is the reaction medium, and the final product Fe-SnO 2 represents Fe-doped SnO 2 nanoparticles. Due to the doping of Fe, these nanoparticles can significantly improve the gas-sensing performance of the material, especially showing higher sensitivity when detecting specific gases such as nitrogen dioxide (NO 2 ). After obtaining the Fe-doped SnO 2 nanoparticles, it is necessary to further improve the conductivity and mechanical strength of the material. Graphene is used as another part of the composite material. Graphene is famous for its excellent conductivity and flexibility. In actual operation, disperse the graphene powder evenly in the solution through ultrasonic dispersion treatment to form a graphene suspension. Mix the Fe-doped SnO 2 nanoparticles with the graphene suspension according to the preset mass ratio to form a composite nanomaterial suspension. The setting of the mass ratio determines the graphene and Fe-SnO in the final composite material2 ratio. By optimizing this ratio, the conductivity and mechanical properties of the material are adjusted so that it not only maintains good gas-sensing performance but also has excellent conductivity and structural stability. The mixed suspension is processed by spray drying technology to rapidly evaporate the solution, leaving uniform granular substances. These granules are then calcined to remove residual organic matter and enhance the crystallinity and stability of the material, forming a bimodal nanocomposite with good physical and chemical properties. To construct the substrate structure of the gas sensor, the silicon substrate is oxidized. This process generates a uniform SiO 2 insulating layer on the surface of the silicon substrate through an oxidation reaction at high temperature. This insulating layer can not only effectively isolate the current and prevent current leakage but also provide good substrate conditions for subsequent photolithography and metal deposition processes. Photolithography and metal deposition processes are carried out on the substrate with the SiO 2 insulating layer. Through this process, a precise comb-shaped electrode structure is formed on the substrate surface. The comb-shaped electrode can significantly increase the contact area between the electrode and the gas, improving the response speed and sensitivity of the sensor. After the electrode structure is formed, a three-dimensional microchannel network is constructed on the substrate using laser direct writing technology. This network structure enables the gas to contact the sensor surface more fully, improving the efficiency of gas diffusion and the response performance of the sensor. The previously prepared bimodal nanocomposite is redispersed to form a nanomaterial suspension. Through inkjet printing technology, this suspension is precisely deposited onto the substrate structure with a three-dimensional microchannel network. The advantage of inkjet printing technology lies in its high precision and low material consumption, enabling the nanomaterials to be uniformly deposited on complex three-dimensional structures, forming a continuous and uniform active layer. The substrate structure deposited with nanomaterials is heat-treated to enhance the adhesion and stability of the material and remove any possible residual solvents and impurities. The structure after heat treatment forms a sensor array with a three-dimensional microchannel network. This sensor array has extremely high gas sensitivity and selectivity, can accurately detect specific gas components in the environment, and provide timely feedback according to changes in gas concentration.

[0050] In a specific embodiment, the process of performing step S2 may specifically include the following steps:

[0051] Precision machining is performed on the low-temperature co-fired ceramic material to obtain a sensor housing with preset air inlets and outlets, and the sensor array is fixed to obtain a sensor assembly combined with the sensor housing;

[0052] Based on the sensor assembly, ultrasonic bonding is performed to obtain a sensor unit with the electrodes connected to external leads, and a temperature sensor and a humidity sensor are integrally installed to obtain a multimodal gas detection module;

[0053] Design a micro air pump and a gas distribution system based on a multimodal gas detection module to obtain a sensor system with uniform gas distribution, and design and install a signal conditioning circuit for the sensor system to obtain a signal processing unit with pre-amplification, filtering, and analog-to-digital conversion functions;

[0054] Program and configure a microcontroller based on the signal processing unit to obtain a data acquisition and communication control system, and perform parallel data acquisition settings on the data acquisition and communication control system to obtain a synchronous acquisition scheme for resistance signals and capacitance signals;

[0055] Perform Butterworth low-pass filtering on the collected original signals based on the synchronous acquisition scheme to obtain signals with high-frequency noise removed, and perform wavelet transform to remove baseline drift on the signals with high-frequency noise removed to obtain preprocessed multimodal signal data.

[0056] Specifically, perform precision machining on the low-temperature co-fired ceramic material. Using high-precision machining equipment, cut out a sensor housing with a specific geometric shape on the low-temperature co-fired ceramic material according to preset design parameters, and design precise air inlets and outlets on the housing. Fix the pre-prepared sensor array inside the housing. The sensor array usually includes multiple sensing units, and each unit can respond to specific gas components. To ensure the tight combination of the sensor array and the housing, a precision fixing process, such as dispensing or welding, is usually used to firmly install the array at the designated position on the housing. Perform ultrasonic bonding based on the sensor component. Ultrasonic bonding is an efficient welding technology that uses high-frequency ultrasonic vibration energy to cause local heating and melting of the contact surface under pressure to achieve a firm metal connection. During this process, the electrodes on the sensor array are precisely connected to the external leads through ultrasonic bonding technology to form a complete sensor unit. This unit not only includes a gas sensor but also integrates a temperature sensor and a humidity sensor to achieve multimodal detection functions. Design a micro air pump and a gas distribution system to ensure uniform distribution of gas on the sensor surface. The design of the micro air pump needs to consider parameters such as gas flow rate, pressure, and stability, while the gas distribution system needs to ensure that the gas can flow evenly through each sensor unit through reasonable pipeline design and flow control. Through this system, the sensor can obtain a stable and consistent gas sample, avoiding detection errors caused by uneven air flow. Design and install a signal conditioning circuit for the sensor system. The signal conditioning circuit is the core part of the entire detection system, and its main functions include signal pre-amplification, filtering, and analog-to-digital conversion. In the sensor array, the gas sensor usually outputs weak electrical signals, and these signals need to be amplified by a pre-amplifier, and the formula is as follows:

[0057] ;

[0058] where, represents the amplified output voltage, represents the original voltage signal output by the sensor, is the gain of the amplifier, usually a constant greater than 1. The amplified signal will pass through a filter, and a Butterworth low-pass filter is used to remove high-frequency noise and retain the useful components in the signal. The Butterworth filter can effectively remove high-frequency interference without introducing additional phase distortion. The filtered signal is converted into a digital signal through an analog-to-digital converter. After the design and installation of the hardware system are completed, programming and configuration of the microcontroller are carried out based on the signal processing unit to construct a data acquisition and communication control system. The microcontroller is the control core of the system and is responsible for coordinating data acquisition, signal processing, and data transmission of the sensor array. When programming, a parallel data acquisition scheme needs to be set to ensure that resistance signals and capacitance signals can be synchronously acquired and processed. Through reasonable programming and scheduling, the microcontroller can efficiently manage the acquisition process of multi-modal signals and avoid data loss and delay. For the acquired original signal, it is processed through a Butterworth low-pass filter, and the formula is as follows:

[0059] ;

[0060] where, is the transfer function of the filter, is the cut-off frequency, is the order of the filter. The function of this filter is to remove high-frequency noise in the signal and retain the effective information in the low-frequency part to obtain a smoother signal. Wavelet transform is performed on the signal after removing high-frequency noise to remove baseline drift. Wavelet transform is a multi-scale analysis tool that can effectively separate the useful components and baseline drift in the signal to obtain a more accurate signal representation. After wavelet transform processing, the preprocessed multi-modal signal data is obtained.

[0061] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0062] Data sampling is performed on the temperature sensor and the humidity sensor to obtain the original temperature and humidity time series data, and median filtering processing is performed on the original temperature and humidity time series data to obtain the temperature and humidity data with outliers removed;

[0063] Based on the temperature and humidity data with outliers removed, fast Fourier transform is performed to obtain the spectral characteristics of the temperature and humidity data, and the spectral characteristics of the temperature and humidity data are analyzed to obtain the target pattern of temperature and humidity changes;

[0064] Based on the target pattern of temperature and humidity changes, an autoregressive moving average model is constructed to obtain a temperature and humidity change trend prediction model, and Kalman filtering processing is performed on the temperature and humidity change trend prediction model to obtain an optimized temperature and humidity estimated value;

[0065] Calculate the influence coefficient of temperature and humidity on the sensitivity of the gas sensor based on the optimized temperature and humidity estimation values, obtain the initial compensation factor, and perform nonlinear least squares fitting on the initial compensation factor to obtain the temperature and humidity compensation function;

[0066] Construct an adaptive fuzzy neural network based on the temperature and humidity compensation function to obtain a dynamic environmental factor compensation model, and optimize the parameters of the dynamic environmental factor compensation model to obtain environmental factor compensation parameters.

[0067] Specifically, sample data from the temperature sensor and humidity sensor to obtain the original temperature and humidity time series data. Perform median filtering on the original temperature and humidity time series data. Median filtering is a commonly used denoising method. Its principle is to sort the data within a window according to size and then take the median value as the output value of the window. This process can effectively remove sharp noise and outliers while retaining the main features of the signal, obtaining temperature and humidity data with outliers removed. Perform a fast Fourier transform on the temperature and humidity data with outliers removed to obtain the spectral characteristics of the data. The Fourier transform is a mathematical tool that converts a time-domain signal into a frequency-domain signal. Through the fast Fourier transform, the energy distribution of temperature and humidity data at different frequencies can be analyzed to reveal the periodic characteristics of temperature and humidity changes. The spectral characteristics of temperature and humidity data can be expressed by the following formula:

[0068] ;

[0069] where, represents the amplitude of the temperature and humidity data at frequency , represents the temperature and humidity data at the th time point, is the total length of the data. By analyzing the spectral characteristics of the temperature and humidity data, identify the main periods and amplitudes of temperature and humidity changes, obtain the target pattern of temperature and humidity changes, and reflect the change laws of temperature and humidity on different time scales. Based on the target pattern of temperature and humidity changes, construct an autoregressive moving average (ARMA) model to predict the future change trend of temperature and humidity. The ARMA model is a classic statistical model for time series prediction, which can model the linear relationship in the time series by combining the autoregressive (AR) and moving average (MA) components. The expression of the ARMA model is:

[0070] ;

[0071] where, represents the temperature and humidity value at time , is the constant term, and are the autoregressive coefficient and the moving average coefficient respectively, is the white noise error term, and are the orders of autoregression and moving average respectively. By training the ARMA model, a prediction model for temperature and humidity changes is obtained. The output of the ARMA model is processed by Kalman filtering. Kalman filtering is a recursive estimation algorithm that dynamically adjusts the estimation error by weighting the model prediction value and the actual observation value to obtain an optimized temperature and humidity estimation value. The core formula of Kalman filtering is:

[0072] ;

[0073] where, is the state estimation value at the current moment, is the prediction value at the previous moment, is the Kalman gain, is the current observation value, is the observation matrix. The calculation of the Kalman gain ensures the balance of the estimation value between the actual observation value and the model prediction value, so as to obtain a more accurate temperature and humidity estimation in a noisy environment. Based on the optimized temperature and humidity estimation value, the influence coefficient of temperature and humidity on the sensitivity of the gas sensor is calculated to obtain the initial compensation factor. The relationship between the temperature and humidity data and the output of the gas sensor is analyzed, and the influence law of temperature and humidity changes on the response of the gas sensor is fitted through experimental data. The initial compensation factor is used to adjust the output of the gas sensor so that it can maintain a high detection accuracy under different environmental conditions. The initial compensation factor is fitted by non-linear least squares to obtain the temperature and humidity compensation function. Non-linear least squares is a numerical optimization method that finds the best fitting parameters by minimizing the sum of the squares of the errors between the actual value and the fitted value. The temperature and humidity compensation function can be expressed as:

[0074] ;

[0075] where, is the sensitivity of the gas sensor after temperature and humidity compensation, is the temperature, is the humidity, They are fitting parameters. By optimizing these parameters, an accurate compensation model is obtained to make the responses of gas sensors more consistent under different temperature and humidity conditions. Based on the temperature and humidity compensation function, an adaptive fuzzy neural network is constructed to establish a dynamic environmental factor compensation model. The adaptive fuzzy neural network combines the advantages of fuzzy logic and neural networks, can handle nonlinear problems through fuzzy rules, and at the same time uses the learning ability of neural networks to adjust parameters. By training the adaptive fuzzy neural network model, real-time compensation for the responses of gas sensors is achieved, enabling it to maintain high-precision gas detection performance under different environmental conditions. The parameters of the dynamic environmental factor compensation model are optimized to obtain the optimal environmental factor compensation parameters. Through iterative optimization and cross-validation, it is ensured that the compensation model can operate stably under various environmental conditions, and finally effective compensation for the influence of temperature and humidity is achieved, improving the detection accuracy and reliability of gas sensors.

[0076] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0077] The preprocessed multi-modal signal data is segmented by time window to obtain signal segments of a fixed length, and the influence coefficients of temperature and humidity on each signal segment are calculated based on the environmental factor compensation parameters to obtain an environmental impact matrix;

[0078] The environmental impact matrix is subjected to singular value decomposition to obtain the target impact modes of environmental factors, and an adaptive filter is constructed based on the target impact modes of environmental factors to obtain an environmental noise estimation model;

[0079] The signal segments and the environmental noise estimation model are subjected to wavelet packet transform to obtain multi-scale time-frequency features, and the mutual information is calculated based on the multi-scale time-frequency features to obtain the correlation index between the signal and the environmental noise;

[0080] The signal segments are adaptively threshold denoised according to the correlation index to obtain preliminarily calibrated signal data, and the preliminarily calibrated signal data is subjected to empirical mode decomposition to obtain intrinsic mode functions;

[0081] Based on the intrinsic mode functions, a Hilbert-Huang transform spectrum is constructed to obtain the instantaneous frequency and amplitude of the signal, and the instantaneous frequency and amplitude are nonlinearly mapped and reconstructed to obtain calibrated multi-modal signal data.

[0082] Specifically, the preprocessed multi-modal signal data is segmented by time windows. The continuous signal data is divided into signal segments of a fixed length, so as to maintain the temporal consistency of the signal and the comparability of features during the analysis process. For example, assume that the sampling frequency of the multi-modal signal data is 100 Hz, and the time window length of each signal segment is 1 second, then each signal segment will contain 100 data points. Based on the environmental factor compensation parameters, the influence coefficients of temperature and humidity on each signal segment are calculated to quantify the influence of environmental conditions on the signal, and an environmental influence matrix is constructed. The elements of the environmental influence matrix represent the degree of correction of temperature and humidity on the signal features within a specific time window. Assume that the environmental influence matrix is a two-dimensional matrix, where the rows represent different time windows and the columns represent different signal features. Each element in the matrix can be represented by the following formula:

[0083] ;

[0084] Among them, represents the environmental influence value of the th signal feature within the th time window, and are the average temperature and humidity within this time window respectively, and are the corresponding temperature and humidity influence coefficients. By calculating the signal segments of all time windows, a complete environmental influence matrix is constructed. The singular value decomposition is performed on the environmental influence matrix to extract the main influence modes of environmental factors. Singular value decomposition is a matrix decomposition technique that can decompose a matrix into three parts, namely a left singular matrix, a diagonal matrix, and a right singular matrix. Through singular value decomposition, the principal components of the matrix are extracted, that is, the main influence modes of environmental factors on signal features. Assume that the environmental influence matrix is , then the singular value decomposition can be performed in the following way:

[0085] ;

[0086] Among them, and are the left singular matrix and the right singular matrix respectively, is a diagonal matrix, and the elements on the diagonal are singular values. These singular values represent the importance of different modes. By analyzing the magnitudes of the singular values, determine which modes are the main environmental impact modes, and use these modes for the subsequent construction of the adaptive filter. Based on the target impact modes of environmental factors, construct an adaptive filter to generate an environmental noise estimation model. The adaptive filter can adjust its parameters in real time according to the changes in the input signal, thereby effectively removing or suppressing the impact of environmental noise on the signal. The environmental noise estimation model is continuously updated by the adaptive filter to reflect the changing characteristics of the noise under different environmental conditions. Perform wavelet packet transform on the signal segment and the environmental noise estimation model to extract multi-scale time-frequency features. Wavelet packet transform is an enhanced wavelet transform method that can provide a finer frequency band division, thus better capturing the time-frequency features of the signal. Through wavelet packet transform, the signal can be decomposed into time-frequency representations at different scales, which not only reflect the local features of the signal but also reveal the correlation between the signal and the environmental noise. To quantify this correlation, calculate the mutual information based on the multi-scale time-frequency features. Mutual information is a measure of the dependence between two random variables. By calculating the mutual information between the signal and the environmental noise, obtain their correlation index. According to the correlation index, perform adaptive threshold denoising on the signal segment. The adaptive threshold denoising method can automatically select the optimal threshold, remove the part of the signal that is highly correlated with the environmental noise, retain the useful information components, and obtain the preliminarily calibrated signal data. Perform empirical mode decomposition (EMD) on the preliminarily calibrated signal data. EMD is an adaptive decomposition method that can decompose a complex signal into several intrinsic mode functions (IMFs), and these IMFs represent different frequency components of the signal. Each IMF corresponds to an independent oscillation mode, which can reveal the internal structure and dynamic changes of the signal. Construct a Hilbert-Huang transform spectrum based on the IMFs. Hilbert-Huang transform is a time-frequency analysis method. By performing the Hilbert transform on the IMFs, obtain the instantaneous frequency and amplitude of the signal. Assume that a certain IMF is , then its Hilbert transform is:

[0087] ;

[0088] where is the signal after Hilbert transform, and P.V. represents the principal value integral. By calculating the instantaneous frequency and amplitude, generate the Hilbert spectrum of the signal to describe the dynamic change characteristics of the signal. Perform non-linear mapping and reconstruction on the instantaneous frequency and amplitude. Non-linear mapping can convert the complex patterns of the signal into a form that is easier to analyze, and reconstruction integrates these patterns into calibrated multi-modal signal data.

[0089] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0090] Extract time-frequency domain features from the calibrated multi-modal signal data to obtain a resistance signal feature set and a capacitance signal feature set;

[0091] Construct a first multi-scale residual network branch based on the resistance signal feature set. The first multi-scale residual network branch includes 3 parallel residual blocks, and each residual block contains convolutional layers and pooling layers of different scales to obtain a training multi-scale resistance feature representation;

[0092] Construct a second multi-scale residual network branch based on the capacitance signal feature set. The structure of the second multi-scale residual network branch is the same as that of the first multi-scale residual network branch to obtain a training multi-scale capacitance feature representation;

[0093] Perform adaptive feature recalibration on the training multi-scale resistance feature representation and the multi-scale capacitance feature representation, and weight the features through a channel attention mechanism and a spatial attention mechanism to obtain training recalibrated features;

[0094] Construct a bidirectional gated recurrent unit network based on the training recalibrated features. The bidirectional gated recurrent unit network contains 2 layers, with 128 units in each layer, and uses the tanh activation function to obtain training temporal context features;

[0095] Apply a multi-head self-attention mechanism to the training temporal context features. The multi-head self-attention mechanism contains 8 attention heads, and the dimension of each head is 64 to obtain training global dependence features;

[0096] Construct a pyramid pooling module based on the training global dependence features, and extract multi-scale global information through pooling operations of three different scales of 1x1, 2x2, and 4x4 to obtain training multi-scale global features;

[0097] Perform feature fusion on the training multi-scale global features, and obtain a training fusion feature vector through element-wise addition and 1x1 convolution operations;

[0098] Construct a multi-task learning head based on the training fusion feature vector, including a gas classification task and a concentration regression task. The gas classification task uses the Softmax activation function, and the concentration regression task uses the linear activation function to obtain a training gas type probability distribution and a concentration estimate value;

[0099] Perform end-to-end training on the multi-input neural network model, use a weighted cross-entropy loss function and a Huber loss function as optimization objectives, and adopt an AdamW optimizer and a cosine annealing learning rate scheduling strategy for parameter update to obtain a gas recognition model.

[0100] Specifically, time-frequency domain feature extraction is performed on the calibrated multi-modal signal data to separate the useful information in the original signal and represent the signal in two dimensions of time and frequency. For multi-modal signal data, the resistance signal and the capacitance signal are two main input sources. Through time-frequency domain feature extraction, a resistance signal feature set is extracted from the resistance signal, and a capacitance signal feature set is extracted from the capacitance signal. These feature sets contain the change information of the signal at different time points and frequency ranges. A first multi-scale residual network branch is constructed based on the resistance signal feature set. The multi-scale residual network is a deep learning model that can capture different scale information in the signal and is suitable for processing signal data with multi-level features. In the first multi-scale residual network branch, three parallel residual blocks are designed, and each residual block contains convolutional layers and pooling layers of different scales to ensure that various features in the resistance signal can be captured. The role of the convolutional layer is to extract the local features of the signal through convolutional operations, while the pooling layer compresses the features through downsampling operations, reducing the feature dimension and retaining important information. Through multi-scale processing, the network can generate a training multi-scale resistance feature representation, which represents the key information of the resistance signal at different scales. Similarly, a second multi-scale residual network branch is constructed based on the capacitance signal feature set. The structure of this branch is the same as that of the first multi-scale residual network branch, including three parallel residual blocks. The convolutional layers and pooling layers in each residual block are set to different scales to capture the multi-level features of the capacitance signal, and finally a training multi-scale capacitance feature representation is obtained. Adaptive feature recalibration is performed on the training multi-scale resistance feature representation and the multi-scale capacitance feature representation. Adaptive feature recalibration weights the features through the channel attention mechanism and the spatial attention mechanism, highlighting important features and suppressing irrelevant or redundant features. The channel attention mechanism mainly targets the channel dimension of the features and adjusts the importance of the features by learning the weights of different channels; while the spatial attention mechanism weights the spatial dimension of the feature map, enabling the network to pay more attention to key spatial positions. Through these operations, the training recalibrated features are obtained. Based on the training recalibrated features, a bidirectional gated recurrent unit (GRU) network is constructed. The bidirectional GRU network is a recurrent neural network that can capture the context information in time series data. The network contains two layers, each consisting of 128 units, and uses the tanh activation function. The output range of the tanh activation function is [-1, 1], which can effectively process the non-linear relationship in the signal. Through bidirectional processing, the network can consider both the forward and backward information of the signal and obtain the training time series context features, representing the global dependence of the signal in the time series. Based on the training time series context features, the multi-head self-attention mechanism is applied. The multi-head self-attention mechanism contains 8 attention heads, and the dimension of each head is 64. The attention mechanism automatically selects the most important information for the current task by calculating the correlation between different features.Through the design of multiple heads, the network can analyze and process features from multiple perspectives to obtain training global-dependent features. Based on the training global-dependent features, a pyramid pooling module is constructed. Pyramid pooling is a multi-scale pooling technique that extracts multi-scale global information of the signal through three different scale pooling operations of 1x1, 2x2, and 4x4. The pooling operation can further compress the feature dimension while maintaining the spatial structure of the feature map, reducing the computational complexity. Through the pyramid pooling module, training multi-scale global features are obtained, and these features have a stronger ability to express global information. Through element-wise addition and 1x1 convolution operations, feature fusion is performed on the multi-scale global features. Element-wise addition can add the elements of different feature layers one by one, while the 1x1 convolution operation further compresses and integrates the features to obtain a training fusion feature vector. Based on the training fusion feature vector, a multi-task learning head is constructed. The multi-task learning head simultaneously processes the gas classification task and the concentration regression task, where the Softmax activation function is used for the gas classification task. The Softmax function can map the input features to a probability distribution for determining the type of gas; while the linear activation function is used for the concentration regression task to estimate the concentration value of the gas. The multi-task learning method not only improves the utilization efficiency of the model but also enables the model to share features between different tasks, enhancing the overall performance. The entire multi-input neural network model is trained end-to-end. To optimize the model, a weighted cross-entropy loss function and a Huber loss function are used as the optimization objectives. The weighted cross-entropy loss function is used for the classification task, and by assigning different weights to different classes, it addresses the class imbalance problem; the Huber loss function is used for the regression task, and its smooth error curve shows better robustness when dealing with outliers. The parameter update uses the AdamW optimizer, which combines the adaptive learning rate and weight decay mechanism of the Adam optimizer and can effectively prevent overfitting. To improve the training effect of the model, a cosine annealing learning rate scheduling strategy is adopted. By gradually reducing the learning rate, the convergence process of the model is stabilized, and finally, a high-precision gas recognition model is obtained.

[0101] In a specific embodiment, the process of executing step S6 may specifically include the following steps:

[0102] Synchronously sample and digitally process the real-time collected resistance signal and capacitance signal to obtain the original multi-modal signal data, and perform adaptive calibration processing based on the original multi-modal signal data to obtain the calibrated real-time multi-modal signal;

[0103] Extract the time-frequency domain features of the calibrated real-time multi-modal signal to obtain the real-time resistance feature set and the real-time capacitance feature set;

[0104] Input the real-time resistance feature set into the first multi-scale residual network branch of the gas recognition model, and through the calculation of 3 parallel residual blocks, obtain the real-time multi-scale resistance feature representation;

[0105] Input the real-time capacitance feature set into the second multi-scale residual network branch of the gas recognition model, and through the calculation of 3 parallel residual blocks, obtain the real-time multi-scale capacitance feature representation;

[0106] Perform adaptive feature recalibration on the real-time multi-scale resistance feature representation and the real-time multi-scale capacitance feature representation to obtain the real-time recalibrated feature;

[0107] Input the real-time recalibrated feature into the bidirectional gated recurrent unit network, and through the sequential calculation of 2 layers with 128 units, obtain the real-time temporal context feature;

[0108] Apply the multi-head self-attention mechanism to the real-time temporal context feature, and through the parallel calculation of 8 attention heads, obtain the real-time global dependence feature;

[0109] Input the real-time global dependence feature into the pyramid pooling module, and through three different-scale pooling operations, obtain the real-time multi-scale global feature;

[0110] Perform a feature fusion operation on the real-time multi-scale global feature to obtain the real-time fusion feature vector;

[0111] Input the real-time fusion feature vector into the multi-task learning head, and through parallel classification and regression calculations, obtain the real-time gas type probability distribution and concentration estimation value;

[0112] Perform threshold judgment and temporal smoothing on the real-time gas type probability distribution and concentration estimation value to obtain the gas type and concentration information.

[0113] Specifically, the resistance signal and capacitance signal collected in real time are synchronously sampled and digitally processed. Synchronous sampling can avoid errors caused by time differences. Through a high-precision analog-to-digital converter, the analog signal is converted into a digital signal to obtain the original multi-modal signal data. The original multi-modal signal data is subjected to adaptive calibration processing to eliminate the influence of environmental factors such as temperature and humidity on the signal, improving the accuracy of the signal. Through a pre-trained environmental compensation model, the deviation in the signal is adjusted in real time, making the calibrated real-time multi-modal signal more accurately reflect the response of the sensor to the target gas. Time-frequency domain feature extraction is performed on the calibrated real-time multi-modal signal. Through time-frequency analysis methods such as short-time Fourier transform or wavelet transform, the signal is decomposed into different time and frequency intervals to obtain more detailed feature information. A real-time resistance feature set is extracted from the resistance signal, and a real-time capacitance feature set is extracted from the capacitance signal. The real-time resistance feature set is input into the first multi-scale residual network branch of the gas recognition model. This branch consists of three parallel residual blocks, each residual block containing convolutional layers and pooling layers of different scales, aiming to capture multi-level features in the resistance signal. The residual network effectively solves the problem of gradient disappearance in deep networks through skip connections. At the same time, the convolutional operation can extract local features, and the pooling operation is used to compress the feature map, reducing the computational complexity. Through multi-scale processing, the network can generate a real-time multi-scale resistance feature representation, effectively describing the important patterns in the resistance signal. Similarly, the real-time capacitance feature set is input into the second multi-scale residual network branch of the gas recognition model. The structure of this branch is the same as that of the resistance signal branch. Through convolutional and pooling operations of different scales, a real-time multi-scale capacitance feature representation is generated. Adaptive feature recalibration is performed on the real-time multi-scale resistance feature representation and the real-time multi-scale capacitance feature representation. Through the channel attention mechanism and the spatial attention mechanism, the features are weighted to highlight the most important features while suppressing irrelevant or redundant features, obtaining real-time recalibrated features. The channel attention mechanism dynamically adjusts the importance of each channel feature by learning the weights of different channels, while the spatial attention mechanism highlights the key spatial positions according to the response intensity at different positions in the feature map. The real-time recalibrated features are input into a bidirectional gated recurrent unit (GRU) network for processing. The bidirectional GRU network can consider both the forward and backward information of the signal at the same time, enabling the model to capture the global dependence of the time series data. This network consists of two layers, with 128 units in each layer, and uses the tanh activation function to handle the non-linear relationship. The GRU network effectively solves the problem of gradient disappearance in long time series data through the gating mechanism, obtaining real-time temporal context features. The multi-head self-attention mechanism is applied to the real-time temporal context features. The multi-head self-attention mechanism calculates different dependencies in the signal through 8 parallel attention heads and integrates this information. The dimension of each attention head is 64, which can capture the complex interrelationships between features and generate real-time global dependence features.Input the real-time global dependency features into the pyramid pooling module. Pyramid pooling extracts the multi-scale global information of the signal through three different scales of pooling operations: 1x1, 2x2, and 4x4. The multi-scale pooling method can retain the important information in the feature map while compressing the feature dimension and reducing the computational complexity. Through the pyramid pooling module, the generated real-time multi-scale global features have a stronger ability to express global information. Perform feature fusion operations on the real-time multi-scale global features. Feature fusion integrates the information from different scales into a unified feature vector through element-wise addition and 1x1 convolution operations. Element-wise addition can maintain the relative proportion of features at each scale, while 1x1 convolution further compresses the channel dimension of the feature map, reduces redundant information, and obtains the real-time fusion feature vector. Input the real-time fusion feature vector into the multi-task learning head for gas type classification and concentration regression calculation. In the classification task, use the Softmax activation function to output the probability distribution of gas types; in the regression task, use the linear activation function to estimate the gas concentration value. Through parallel computing, the model can simultaneously complete gas identification and concentration estimation, providing comprehensive detection results. Perform threshold judgment and temporal smoothing processing on the real-time obtained gas type probability distribution and concentration estimation values. Threshold judgment is used to filter out low-probability or uncertain results to ensure the stability of the output; temporal smoothing eliminates instantaneous fluctuations through methods such as weighted averaging, making the final output gas type and concentration information more stable and reliable.

[0114] The detection method based on the multi-modal gas sensor in the embodiment of the present invention has been described above. Next, the detection device based on the multi-modal gas sensor in the embodiment of the present invention will be described. Please refer to Figure 2 , an embodiment of the detection device based on the multi-modal gas sensor in the embodiment of the present invention includes:

[0115] A creation module, configured to perform co-blending on the metal oxide semiconductor and the conductive material to obtain a dual-modal nanocomposite, and create a sensor array with a three-dimensional microchannel network based on the dual-modal nanocomposite;

[0116] An acquisition module, configured to perform packaging and integration processing on the sensor array to obtain a multi-modal gas detection module, and perform synchronous acquisition and digital filtering processing on the resistance signal and the capacitance signal based on the multi-modal gas detection module to obtain preprocessed multi-modal signal data;

[0117] An analysis module, configured to collect and analyze the data of the temperature sensor and the humidity sensor in the multi-modal gas detection module to obtain environmental factor compensation parameters;

[0118] A calibration module, configured to perform adaptive calibration processing on the preprocessed multi-modal signal data according to the environmental factor compensation parameters to obtain calibrated multi-modal signal data;

[0119] A training module, configured to construct and train a multi-input neural network model based on the calibrated multi-modal signal data to obtain a gas recognition model;

[0120] A processing module, configured to analyze and judge the real-time collected and calibrated multi-modal signals based on the gas recognition model to obtain gas type and concentration information.

[0121] Through the collaborative cooperation of the above-mentioned various components, the design and preparation technology of the dual-modal nanocomposite material, through the synergistic effect of Fe-doped SnO 2 nanoparticles and graphene, significantly improves the gas-sensing performance and selectivity of the sensing material. The design and manufacture of the three-dimensional microchannel network structure increase the contact area between the gas and the sensitive material, improve the response speed and sensitivity of the sensor, and are conducive to the generation and transmission of multi-modal signals. The application of the temperature and humidity compensation technology effectively reduces the influence of environmental changes on the detection results through the dynamic environmental factor compensation model, and improves the stability and reliability of the sensor in complex environments. The adaptive calibration processing technology combines methods such as wavelet packet transform and empirical mode decomposition to achieve precise calibration of multi-modal signals, effectively improving the signal-to-noise ratio and feature expression ability of the signals. The design of the multi-input neural network model integrates advanced algorithms such as multi-scale residual network, bidirectional gated recurrent unit network, and multi-head self-attention mechanism, significantly improving the recognition ability of complex gases and the concentration estimation accuracy. The adoption of the multi-task learning strategy simultaneously optimizes the gas classification and concentration regression tasks, improves the generalization ability and robustness of the model, enabling it to better handle various complex situations in practical applications. The present invention realizes high-precision and high-reliability gas detection.

[0122] The present invention also provides a computer device, which includes a memory and a processor. Computer-readable instructions are stored in the memory. When the computer-readable instructions are executed by the processor, the processor executes the steps of the detection method based on the multi-modal gas sensor in the above-mentioned embodiments.

[0123] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions run on a computer, the computer executes the steps of the detection method based on the multi-modal gas sensor.

[0124] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described system, system, and unit can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0125] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this 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 for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0126] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of various embodiments of the present invention.

Claims

1. A detection method based on a multimodal gas sensor, characterized in that: The method comprises: A metal oxide semiconductor and a conductive material are co-mixed and synthesized to obtain a bimodal nanocomposite material, and a sensor array with a three-dimensional microchannel network is created based on the bimodal nanocomposite material; specifically, the method comprises: mixing a tin source and an iron source according to a preset molar ratio to obtain a metal precursor solution, and subjecting the metal precursor solution to a hydrothermal reaction to obtain Fe-doped SnO2 nanoparticles; subjecting graphene to an ultrasonic dispersion treatment to obtain a graphene suspension, and mixing the Fe-doped SnO2 nanoparticles and the graphene suspension according to a preset mass ratio to obtain a composite nanomaterial suspension; spray drying and calcining the composite nanomaterial suspension; The invention relates to a method for preparing a nanostructured carbon nanotube film by sintering a silicon substrate and performing a sintering treatment to obtain a bimodal nanocomposite material, and performing an oxidation treatment on a silicon substrate to obtain a substrate having a SiO2 insulating layer; performing photolithography and metal deposition treatment on the substrate having the SiO2 insulating layer to obtain a substrate having a comb-shaped electrode, and performing a laser direct writing treatment on the substrate having the comb-shaped electrode to obtain a substrate structure having a three-dimensional microchannel network; performing a dispersion treatment on the bimodal nanocomposite material to obtain a nanomaterial suspension, and depositing the nanomaterial suspension onto the substrate structure having a three-dimensional microchannel network by inkjet printing technology, and performing a heat treatment on the substrate structure deposited with the nanomaterial to obtain a sensor array having a three-dimensional microchannel network; The sensor array is packaged and integrated to obtain a multimodal gas detection module, and based on the multimodal gas detection module, resistance signals and capacitance signals are synchronously collected and digitally filtered to obtain pre-processed multimodal signal data; Collecting and analyzing the temperature sensor and humidity sensor data in the multi-modal gas detection module to obtain environmental factor compensation parameters; Performing adaptive calibration processing on the preprocessed multimodal signal data according to the environmental factor compensation parameter to obtain calibrated multimodal signal data; Building a multi-input neural network model based on the calibrated multimodal signal data and training it to obtain a gas recognition model; Based on the gas identification model, the multimodal signals collected and calibrated in real time are analyzed and judged to obtain gas type and concentration information.

2. The detection method based on a multimodal gas sensor according to claim 1, characterized in that: The packaging and integration processing of the sensor array is performed to obtain a multi-modal gas detection module, and based on the multi-modal gas detection module, the synchronous acquisition and digital filtering processing of the resistance signal and the capacitance signal are performed to obtain pre-processed multi-modal signal data, including: Precision-processing the low-temperature co-fired ceramic material to obtain a sensor housing with a preset air inlet and air outlet, and fixing the sensor array to obtain a sensor assembly combined with the sensor housing; Ultrasonic bonding is performed based on the sensor assembly to obtain a sensor unit in which electrodes are connected to external leads, and a temperature sensor and a humidity sensor are integrated and installed to obtain a multi-modal gas detection module; Based on the multimodal gas detection module, a micro air pump and a gas distribution system are designed to obtain a sensor system with uniform gas distribution, and a signal conditioning circuit is designed and installed for the sensor system to obtain a signal processing unit with preamplification, filtering and analog-to-digital conversion functions; Programming and configuring a microcontroller based on the signal processing unit to obtain a data acquisition and communication control system, and performing parallel data acquisition settings on the data acquisition and communication control system to obtain a synchronous acquisition scheme for resistance signals and capacitance signals; Based on the synchronous acquisition scheme, the collected original signal is processed by Butterworth low-pass filtering to obtain a signal with high-frequency noise removed, and the signal with high-frequency noise removed is processed by wavelet transformation to remove baseline drift to obtain pre-processed multimodal signal data.

3. The detection method based on a multimodal gas sensor according to claim 1, characterized in that: The collecting and analyzing the temperature sensor and humidity sensor data in the multi-modal gas detection module to obtain the environmental factor compensation parameters includes: Sampling data from the temperature sensor and the humidity sensor to obtain original temperature and humidity time series data, and performing median filtering on the original temperature and humidity time series data to obtain temperature and humidity data with outliers removed; Performing a fast Fourier transform on the temperature and humidity data with outliers removed to obtain frequency spectrum characteristics of the temperature and humidity data, and analyzing the frequency spectrum characteristics of the temperature and humidity data to obtain a target mode of temperature and humidity changes; Based on the target pattern of temperature and humidity changes, an autoregressive moving average model is constructed to obtain a temperature and humidity change trend prediction model, and the temperature and humidity change trend prediction model is subjected to Kalman filtering to obtain optimized temperature and humidity estimation values; Calculating the influence coefficient of temperature and humidity on the sensitivity of the gas sensor based on the optimized temperature and humidity estimation value to obtain an initial compensation factor, and performing nonlinear least squares fitting on the initial compensation factor to obtain a temperature and humidity compensation function; An adaptive fuzzy neural network is constructed based on the temperature and humidity compensation function to obtain a dynamic environmental factor compensation model, and the parameters of the dynamic environmental factor compensation model are optimized to obtain environmental factor compensation parameters.

4. The detection method based on a multimodal gas sensor according to claim 1, characterized in that: The step of performing adaptive calibration processing on the pre-processed multi-modal signal data according to the environmental factor compensation parameter to obtain calibrated multi-modal signal data includes: Performing time window segmentation on the preprocessed multimodal signal data to obtain signal segments of fixed length, and calculating the influence coefficient of temperature and humidity on each signal segment based on the environmental factor compensation parameter to obtain an environmental influence matrix; Performing singular value decomposition on the environmental impact matrix to obtain a target impact pattern of environmental factors, and constructing an adaptive filter based on the target impact pattern of environmental factors to obtain an environmental noise estimation model; Performing wavelet packet transform on the signal segment and the environmental noise estimation model to obtain multi-scale time-frequency features, and calculating mutual information based on the multi-scale time-frequency features to obtain a correlation index between the signal and the environmental noise; Performing adaptive threshold denoising on the signal segments according to the correlation index to obtain preliminary calibrated signal data, and performing empirical mode decomposition on the preliminary calibrated signal data to obtain an eigenmode function; A Hilbert-Huang transform spectrum is constructed based on the eigenmode function to obtain the instantaneous frequency and amplitude of the signal, and the instantaneous frequency and amplitude are nonlinearly mapped and reconstructed to obtain calibrated multimodal signal data.

5. The detection method based on a multimodal gas sensor according to claim 1, characterized in that: The multi-input neural network model is constructed based on the calibrated multi-modal signal data and trained to obtain a gas recognition model, including: Performing time-frequency domain feature extraction on the calibrated multimodal signal data to obtain a resistance signal feature set and a capacitance signal feature set; Constructing a first multi-scale residual network branch based on the resistance signal feature set, wherein the first multi-scale residual network branch includes three parallel residual blocks, each residual block includes convolutional layers and pooling layers of different scales, and obtains a training multi-scale resistance feature representation; constructing a second multi-scale residual network branch based on the capacitance signal feature set, wherein the structure of the second multi-scale residual network branch is the same as that of the first multi-scale residual network branch, and obtaining a training multi-scale capacitance feature representation; Adaptively recalibrating the training multi-scale resistance feature representation and the multi-scale capacitance feature representation, and weighting the features through a channel attention mechanism and a spatial attention mechanism to obtain training recalibrated features; Based on the training recalibration features, a bidirectional gated recurrent unit network is constructed, wherein the bidirectional gated recurrent unit network comprises 2 layers, each layer has 128 units, and a tanh activation function is used to obtain training temporal context features; Applying a multi-head self-attention mechanism to the training temporal context feature, the multi-head self-attention mechanism comprising 8 attention heads, each with a dimension of 64, to obtain a training global dependency feature; A pyramid pooling module is constructed based on the training global dependency features, and multi-scale global information is extracted through pooling operations of three different scales: 1×1, 2×2, and 4×4, to obtain training multi-scale global features; Performing feature fusion on the training multi-scale global features, and obtaining a training fusion feature vector through element-by-element addition and 1x1 convolution operation; Based on the training fusion feature vector, a multi-task learning head is constructed, including a gas classification task and a concentration regression task, wherein the gas classification task uses a Softmax activation function, and the concentration regression task uses a linear activation function, to obtain a probability distribution of training gas types and a concentration estimation value; The multi-input neural network model is trained end-to-end, a weighted cross entropy loss function and a Huber loss function are used as optimization targets, and an AdamW optimizer and a cosine annealing learning rate scheduling strategy are used to update parameters to obtain a gas recognition model.

6. The detection method based on a multimodal gas sensor according to claim 1, characterized in that: The multimodal signals collected and calibrated in real time are analyzed and judged based on the gas identification model to obtain gas type and concentration information, including: The resistance signal and the capacitance signal collected in real time are synchronously sampled and digitally processed to obtain original multimodal signal data, and adaptively calibrated based on the original multimodal signal data to obtain a calibrated real-time multimodal signal; Performing time-frequency domain feature extraction on the calibrated real-time multimodal signal to obtain a real-time resistance feature set and a real-time capacitance feature set; Inputting the real-time resistance feature set into the first multi-scale residual network branch of the gas recognition model, and calculating through three parallel residual blocks to obtain a real-time multi-scale resistance feature representation; Inputting the real-time capacitance feature set into the second multi-scale residual network branch of the gas recognition model, and calculating through three parallel residual blocks to obtain a real-time multi-scale capacitance feature representation; Performing adaptive feature recalibration on the real-time multi-scale resistance feature representation and the real-time multi-scale capacitance feature representation to obtain real-time recalibrated features; Input the real-time recalibration feature into a bidirectional gated recurrent unit network, and obtain the real-time temporal context feature through sequential calculation of 2 layers of 128 units; Applying a multi-head self-attention mechanism to the real-time temporal context features, and obtaining real-time global dependency features through parallel calculation of 8 attention heads; The real-time global dependency features are input into a pyramid pooling module, and real-time multi-scale global features are obtained through pooling operations at three different scales; Performing a feature fusion operation on the real-time multi-scale global feature to obtain a real-time fused feature vector; The real-time fused feature vector is input into a multi-task learning head, and real-time gas type probability distribution and concentration estimation are obtained through parallel classification and regression calculation; Threshold determination and time series smoothing are performed on the real-time gas type probability distribution and concentration estimation value to obtain gas type and concentration information.

7. A detection device based on a multimodal gas sensor, characterized in that: Used to perform the detection method based on a multimodal gas sensor as described in any one of claims 1 to 6, the detection device based on the multimodal gas sensor comprises: A creation module is used to perform a co-compounding process on a metal oxide semiconductor and a conductive material to obtain a bimodal nanocomposite material, and to create a sensor array having a three-dimensional microchannel network based on the bimodal nanocomposite material; An acquisition module is used to package and integrate the sensor array to obtain a multi-modal gas detection module, and to synchronously acquire and digitally filter the resistance signal and the capacitance signal based on the multi-modal gas detection module to obtain pre-processed multi-modal signal data; An analysis module, used to collect and analyze data from the temperature sensor and humidity sensor in the multi-modal gas detection module to obtain environmental factor compensation parameters; A calibration module, configured to perform adaptive calibration processing on the pre-processed multimodal signal data according to the environmental factor compensation parameter to obtain calibrated multimodal signal data; A training module, used to construct a multi-input neural network model based on the calibrated multimodal signal data and perform training to obtain a gas recognition model; The processing module is used to analyze and judge the multi-modal signals collected and calibrated in real time based on the gas recognition model to obtain gas type and concentration information.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the detection method based on a multimodal gas sensor according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the processor is enabled to execute the detection method based on a multi-modal gas sensor as claimed in any one of claims 1 to 6.

Images