An electronic nose device based on microwave resonator array and gas analysis method

Abstract

The application discloses a kind of electronic nose devices and gas analysis methods based on microwave resonator array, the device includes upper layer sensing array layer, and lower layer feed layer below upper layer sensing array layer;Lower layer feed layer includes ground plane, lower layer dielectric substrate is arranged on ground plane, and microwave coupling feed line is arranged on lower layer dielectric substrate;Upper layer sensing array layer includes upper layer dielectric substrate, and resonator array is arranged on upper layer dielectric substrate, and resonator array includes several resonator units, and several resonator units are coated with different gas-sensitive material coating, for detecting multiple gas.The device utilizes the different characteristics of different gas-sensitive materials for different gas adsorption capacity, realizes the selective detection and multimode combined analysis of gas, can simultaneously realize multiple gas detection, and effectively resist the influence of environmental temperature and humidity on gas detection accuracy, has wide application prospect in gas monitoring and sensor device integration field.

Description

Technical Field

[0001] This invention relates to an electronic nose device and a gas analysis method based on a microwave resonator array, belonging to the field of microwave sensor technology. Background Technology

[0002] Microwave gas sensing technology, with its non-invasiveness, high penetration capability, and sensitive response to the dielectric properties of the medium, has become an important research direction in fields such as environmental monitoring, industrial safety protection, and food safety monitoring.

[0003] However, existing technologies have significant limitations: traditional single-layer microwave sensors, due to the integral integration of the sensing unit and the feeding network, suffer from fixed device functions and limited lifespan. Once the sensitive material fails or the detection target changes, the entire device must be scrapped. On the other hand, mainstream single-resonant cavity structures can only coat the sensitive area with one type of sensitive material. When facing complex mixed gas environments, multiple device replacements, deployments of multiple devices, or serial detection are required, resulting in low efficiency, increased costs, and difficulty in capturing the interaction effects between gases. Furthermore, electrochemical sensors relying on surface redox reactions, which use materials such as metal oxide semiconductors as sensitive materials, typically require high-temperature activation above 200°C. This not only brings high energy consumption and thermal control challenges but also easily leads to material structure degradation, restricting the long-term stability and applicable scenarios of the equipment. Summary of the Invention

[0004] To address the aforementioned issues, this invention provides an electronic nose device and gas analysis method based on a microwave resonator array. The electronic nose device employs a dual-layer, detachable microwave sensor array architecture, utilizing a physically separated modular design. This allows the upper sensor array layer to be flexibly replaced for different detection scenarios, significantly reducing maintenance costs and overcoming the limitation of single-use devices. On the surface of the upper sensor array layer, nine independently designed interdigital resonator units are arranged in a 3×3 matrix. Each resonator unit optimizes the excitation of a highly localized electromagnetic field distribution through unique geometric parameters, forming a spectrally separated resonant mode. The surface of each resonator unit can be functionally coated with a gas-sensitive material coating specific to a particular gas. The change in dielectric constant caused by gas molecule adsorption perturbs the resonant frequency and amplitude, enabling parallel real-time detection of nine gases. This electronic nose device achieves high sensitivity at the ppm level at room temperature, completely avoiding the need for high-temperature operation and significantly improving the device's energy efficiency and safety.

[0005] The technical solution of the present invention is as follows:

[0006] This invention provides an electronic nose device based on a microwave resonator array, comprising an upper sensing array layer and a lower feed layer located below the upper sensing array layer; the lower feed layer includes a ground plane, a lower dielectric substrate disposed on the ground plane, and a microwave coupling feed line disposed on the lower dielectric substrate; the upper sensing array layer includes an upper dielectric substrate and a resonator array disposed on the upper dielectric substrate, the resonator array including a plurality of resonator units, the plurality of resonator units being coated with different gas-sensitive material coatings for detecting multi-component gases.

[0007] In one embodiment of the present invention, the equivalent circuit of each resonator unit of the resonator array is represented as an RLC series circuit, and its resonant frequency is:

[0008]

[0009] Where L sensor C is the equivalent inductance of the resonator. D It represents the capacitance change after the gas-sensitive material combines with gas molecules, and C is the equivalent capacitance of the resonator.

[0010] In one embodiment of the present invention, the resonator array is coated with different types of gas-sensitive material coatings, including P-type semiconductors, N-type semiconductors, polymers, and high-dielectric ceramic materials.

[0011] Each of the gas-sensitive material coatings is sensitive to one or more gases. When a gas is adsorbed onto the gas-sensitive material coating, the change in the material's dielectric properties is determined using the Debye model.

[0012]

[0013] where ω=2πf, τ=1 / (2πf R ), f R It is the relaxation frequency, ε′ ∞ It is the real part of the resonant frequency f→∞, ε′ s The real part is f→0. The higher the operating frequency, the greater the change in dielectric constant. The capacitance change C of the resonator caused by gas adsorption is... D The larger the value, the more significant the change in resonant frequency and S21 parameter.

[0014] In one embodiment of the present invention, the upper sensing array layer and the lower feed layer are detachably connected by screws.

[0015] In one embodiment of the present invention, the resonator array, microwave coupling feed line, and ground plane are made of conductive materials.

[0016] In one embodiment of the present invention, the upper sensing array layer and the lower feeding layer are both prepared on a polytetrafluoroethylene dielectric substrate by a wet etching process.

[0017] In one embodiment of the present invention, the resonator array includes nine independently designed interdigital resonator units arranged in a 3×3 matrix, which can generate resonant modes at different frequencies depending on the size of the resonators.

[0018] In one embodiment of the present invention, the exposed surface of the resonator array is coated with nine gas-sensitive materials, including three P-type semiconductors, three N-type semiconductors, a polymer, a two-dimensional material, and a ceramic material, to selectively adsorb target gas molecules. The N-type semiconductors include SnO2, ZnO, and TiO2. The oxygen vacancies on the anatase phase surface of TiO2 enhance NO2 capture capability. The dual N-type semiconductors of SnO2 and ZnO cross-validate the NO2 signal, with the high sensitivity of SnO2 complementing the fast recovery characteristics of ZnO. The P-type semiconductors include NiO, Co3O4, and CuO. NiO's hole-dominated conductivity mechanism exhibits intrinsic selectivity for reducing gases. NH3 molecules react with pre-adsorbed oxygen, releasing electrons to neutralize holes. The spinel structure of Co2O4 provides multiple oxidation states, enhancing its catalytic oxidation activity for NH3 and improving sensitivity compared to NiO. The P-type characteristics of CuO assist in NH3 detection, and its surface Cu... + The active sites preferentially adsorb ammonia molecules; the two-dimensional material selected is Ti3C2T. x As a result, the surface terminating groups form a directional hydrogen bond network with the ethanol hydroxyl groups, leading to a significant change in the imaginary part of the dielectric constant. Its layered structure provides a dielectric constant >800m. 2 / g specific surface area enables highly sensitive ethanol detection; polyaniline, as a polymer material, undergoes conformational changes in the aniline unit molecular chain induced by acetone carbonyl groups, altering the polaron migration path and enabling specific differentiation between ketone and alcohol interference; barium titanate, a strong dielectric ceramic material, exhibits ferroelectric domain flipping response polar molecules, with a real part of the dielectric constant changing by up to 40%, broadening its temperature and humidity adaptability and eliminating the influence of environmental humidity on the accuracy of electronic nose devices.

[0019] This invention also provides a gas analysis method based on a microwave resonator array, employing the aforementioned electronic nose device based on a microwave resonator array. This gas analysis method is based on a graph neural network model and specifically includes the following steps:

[0020] Step 1: Introduce the gas to be tested into the resonator array and collect the S21 parameters of each resonator unit, including the resonant frequency shift Δf, amplitude attenuation Δ|S21|, and phase change. And calculate the amount of change of the above characteristics;

[0021] Step 2: Treat each resonator unit as an independent graph node to construct a graph structure. Node features include Δf, Δ|S21|, Dynamic edges are generated based on electromagnetic coupling characteristics: frequency coupling edges are established when the resonant frequency difference is <50MHz, and response correlation edges are established when the correlation coefficient of response change is >0.7.

[0022] Step 3: Analyze mixed gas information using a hierarchical attention mechanism, a gas-level attention mechanism, and a concentration-level attention mechanism. Calculate gas-specific weights based on a pre-defined learnable lookup table for each gas. Generate concentration-sensitive weights using a lightweight multilayer perceptron. Use multi-head attention to filter resonator nodes and finally focus on key feature regions of different gas responses.

[0023] Step 4: Parallel output of three collaborative tasks: one is gas presence detection, which uses a sigmoid classifier to output the probability of gas presence; another is gas type classification, which concatenates node-level features and uses a Softmax classifier to give probabilities of different combinations; and the last is concentration probability modeling, which outputs a 95% confidence interval based on the concentration expectation.

[0024] The present invention also provides the application of an electronic nose device based on a microwave resonator array as described above in multi-gas specific identification.

[0025] The beneficial effects of this invention are as follows:

[0026] (1) The present invention provides an electronic nose device based on a microwave resonator array. By designing a double-layer detachable structure of an upper sensing array layer and a lower feeding layer, the lower feeding layer can be reused. Furthermore, the upper sensing array layer can be freely replaced according to actual usage requirements for different gas detection environments, requiring only recoating or attaching a customized gas-sensitive material coating. The electronic nose device adopts an arrayed resonator approach, which, compared to traditional single-resonator detection, enables the mixed detection of multiple gases. By utilizing the changes in different resonance modes under different gases and concentrations, specific detection and quantitative analysis of multiple gases can be achieved.

[0027] (2) This invention can directly functionalize nine different gas-sensitive materials, including three types of P-type semiconductors, three types of N-type semiconductors, and one each of polymer, two-dimensional material, and strong dielectric ceramic. Through the specific interaction between the material properties and the target gas, the selective capture capability of complex gas mixtures is significantly improved. The multimode electromagnetic field excited by the nine independent resonator units can synchronously sense the changes in multiple microwave parameters caused by gas adsorption, constructing an information density far exceeding that of single-mode sensors. This multimode joint analysis mechanism effectively overcomes the misjudgment problem caused by cross-sensitivity in traditional sensors.

[0028] (3) This invention, based on a graph neural network model, achieves a paradigm shift in gas detection from single-point response analysis to systematic relational reasoning. It constructs a dynamic graph model using nine resonator units, modeling each resonator unit as a graph node, with its response characteristics serving as node attributes. The electromagnetic coupling effect and spatial correlation between units are then constructed as dynamic edge weights. Through hierarchical graph convolution and attention mechanisms, the network can adaptively learn the microwave parameter changes caused by different gas combinations, first completing the qualitative classification of multi-component gases, and then accurately quantifying the concentration of each gas through regression branches. This collaborative innovation of hardware and algorithms not only breaks through the traditional sensor's "single-target" detection bottleneck but also lays the technological foundation for building a high-throughput, low-power intelligent gas sensing platform. The model architecture significantly improves the accuracy of identifying four-component mixed gases (NH3 / NO2 / ethanol / acetone) compared to traditional methods, completely solving the false alarm problem caused by cross-sensitivity.

[0029] (4) This invention utilizes the different adsorption capacities of various gas-sensitive materials for different gases to achieve selective gas detection and multi-mode joint analysis. Because the dielectric properties of the near-field resonator change after the gas-sensitive material adsorbs gas molecules, the corresponding resonant modes also change. The electronic nose device utilizes this multi-mode variation characteristic to simultaneously detect multiple gases and effectively resist the influence of ambient temperature and humidity on gas detection accuracy. This electronic nose device has a simple manufacturing process, high sensitivity, and is suitable for industrial environments and food quality monitoring, showing broad application prospects in the fields of gas monitoring and sensor integration. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0031] Figure 1 A schematic diagram of the structure of the electronic nose device based on a microwave resonator array provided by the present invention;

[0032] Figure 2 A schematic diagram of the structure of nine resonator units of different sizes provided by the present invention;

[0033] Figure 3 Surface electric field distribution diagrams of nine resonator units of different sizes provided for this invention;

[0034] Figure 4 Simulation S21 parameter diagram of the microwave resonator array provided by the present invention;

[0035] Figure 5 S21 response curves of the microwave resonator array provided by the present invention under different gas concentrations;

[0036] Figure 6 This is a fitted curve of the gas concentration predicted by the neural network and the actual target gas concentration used in this invention;

[0037] Figure 7 The structural block diagram of the graph neural network model provided by this invention.

[0038] In the figure: 1. Upper sensor array layer; 2. Lower feed layer; 3. Ground plane; 4. Lower dielectric substrate; 5. Microwave coupling feed line; 6. Upper dielectric substrate; 7. Resonator array; 8. Gas-sensitive material coating. Detailed Implementation

[0039] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The terms "first position" and "second position" refer to two different positions.

[0041] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections or detachable connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal connections between two components. Those skilled in the art can understand the specific meaning of these terms in this invention based on the specific circumstances.

[0042] Example 1

[0043] like Figure 1 , Figure 2 and Figure 3As shown, this embodiment provides an electronic nose device based on a microwave resonator array for gas category detection and concentration identification. The electronic nose device includes an upper sensing array layer 1 and a lower feed layer 2 located below the upper sensing array layer 1. The lower feed layer 2 includes a ground plane 3, a lower dielectric substrate 4 disposed on the ground plane 3, and a microwave coupling feed line 5 disposed on the lower dielectric substrate 4. The upper sensing array layer 1 includes an upper dielectric substrate 6 and a resonator array 7 disposed on the upper dielectric substrate 6. The resonator array 7 includes a plurality of resonator units, and the plurality of resonator units are coated with different gas-sensitive material coatings 8 for detecting multi-component gases.

[0044] Optionally, the resonator array 7, microwave coupling feed line 5, and ground plane 3 are made of conductive materials, such as copper.

[0045] Optionally, the upper sensing array layer 1 and the lower power supply layer 2 are detachably connected by screws.

[0046] Optionally, the upper sensing array layer 1 and the lower feeding layer 2 are both prepared on a polytetrafluoroethylene dielectric substrate (Teflon@0.54mm, Cu@30um) by a wet etching process, and then positioning screw holes with a diameter of 4mm are drilled at the top corners of the lower dielectric substrate 4 and the upper dielectric substrate 6 using a drilling tool.

[0047] Optionally, the resonator array 7 includes nine independently designed interdigitated resonator units arranged in a 3×3 matrix, which can generate resonant modes at different frequencies depending on the size of the resonators. The number of resonator units can be adjusted according to actual usage requirements, such as 2×2 or 4×4. This embodiment selects a 3×3 array size for detecting quaternary gas mixtures.

[0048] The equivalent circuit of each resonator unit in the resonator array 7 can be represented as an RLC series circuit, with the following resonant frequency:

[0049]

[0050] Where L sensor C is the equivalent inductance of the resonator. D It represents the capacitance change after the gas-sensitive material combines with gas molecules, and C is the equivalent capacitance of the resonator.

[0051] The exposed copper on the resonator array 7 is coated with different types of gas-sensitive material coatings 8, including P-type semiconductors, N-type semiconductors, polymers, and high-dielectric ceramic materials. Different gas-sensitive materials exhibit different response characteristics to different gases, thus forming detection units with varying gas sensitivities together with the microwave resonator units. The response differences between these units provide data support for identifying gas types and predicting concentrations in complex gas atmospheres, thereby enabling multi-mode joint analysis using variations in nine different resonant modes.

[0052] In this embodiment, nine different gas-sensitive materials are coated on top of the microwave resonator array, including three P-type semiconductors, three N-type semiconductors, one polymer, one two-dimensional material, and one ceramic material. In actual use, the gas-sensitive material coating 8 can be replaced as needed, i.e., the upper sensing array layer 1 can be replaced to reuse the lower feed layer 2.

[0053] Furthermore, each of the gas-sensitive material coatings 8 is sensitive to one or more gases. When a gas is adsorbed onto the gas-sensitive material coating 8, the change in the material's dielectric properties is determined using the Debye model:

[0054]

[0055] where ω=2πf, τ=1 / (2πf R ), f R It is the relaxation frequency, ε′ ∞ It is the real part of the resonant frequency f→∞, ε′ s The real part is f→0. The higher the operating frequency, the greater the change in dielectric constant. The capacitance change C of the resonator caused by gas adsorption is... D The larger the value, the more significant the change in resonant frequency and S21 parameter.

[0056] Different gas-sensitive materials exhibit significant differences in their response characteristics to different gases. This difference manifests in the specific reactions of materials coated on microwave resonator units (such as P-type semiconductors, N-type semiconductors, polymers, or strongly dielectric ceramics) upon gas exposure. Each resonator unit thus constitutes a detection unit with unique gas sensitivity. The response differences between units are mainly manifested in variations of multiple radio frequency derived parameters, including resonant frequency shift, amplitude attenuation, phase angle changes, and quality factor (Q factor) fluctuations. These differences in derived parameters stem from electromagnetic perturbations caused by the interaction between gas molecules and the gas-sensitive material, providing a rich data matrix for gas identification in complex gas atmospheres. By analyzing the patterns of these parameters (e.g., using machine learning algorithms to handle frequency-amplitude correlations), the system can achieve highly selective gas identification and concentration prediction, thereby supporting multi-mode joint analysis and accurate detection, demonstrating olfactory perception capabilities surpassing those of humans. Therefore, the electronic nose device based on a microwave resonator array provided by this invention can be applied to multi-gas specific identification.

[0057] Example 2

[0058] This embodiment provides an electronic nose device based on a microwave resonator array for gas category detection and concentration identification. The electronic nose device includes an upper sensing array layer 1 and a lower feed layer 2. The upper sensing array layer 1 includes a 0.54 mm thick polytetrafluoroethylene (PTFE) dielectric substrate 6, an interdigitated resonator array 7 formed by copper etching, a gas-sensitive material coating 8 on the surface of the resonator array 7, and four corner positioning screw holes. The lower feed layer 2 includes a ground plane 3, a 0.54 mm thick PTFE dielectric substrate 4, an S-shaped microwave coupling feed line 5, and positioning screw holes corresponding to the corner positioning screw holes of the upper sensing array layer 1. The dielectric substrate is fabricated on a copper-clad substrate (Cu@30 μm) using a wet etching process, and mechanical alignment is achieved using 4 mm diameter positioning screw holes. The lower feed layer 2 is connected to a vector network analyzer via a coaxial interface, responsible for exciting the resonator array 7 and acquiring the S21 transmission parameters in real time.

[0059] As attached Figure 2 As shown, the interdigital resonator array 7 consists of nine independent copper resonator units arranged in a 3×3 matrix. Each resonator unit is suspended above the S-shaped microwave coupling feed line 5 via a bottom coupling region. By differentiating the resonator dimensions and changing the unit electrical length, the array generates nine spectrally separated resonant peaks within the 2.5–6 GHz frequency band, such as... Figure 4 As shown, resonator elements 1 to 9 correspond to 2.77 GHz, 3.11 GHz, 3.44 GHz, 3.68 GHz, 4.00 GHz, 4.36 GHz, 4.73 GHz, 5.13 GHz, and 5.41 GHz, respectively. The surface electric field of each resonator element is highly localized as follows: Figure 4 As shown, a gas-sensitive region is formed. By coating the exposed copper surface of the resonator array 7 with nine gas-sensitive materials, including three P-type semiconductors, three N-type semiconductors, one polymer, one two-dimensional material, and one ceramic material, target gas molecules can be selectively adsorbed.

[0060] In this embodiment, a 3×3 resonator array 7 is preferred to balance the detection requirements of quaternary mixed gases with the spectral resource usage. However, the resonator unit topology can be expanded to a 2×2 or 4×4 configuration depending on the actual application. Its gas detection mechanism is based on: gas molecule adsorption causing a change in the dielectric constant of the gas-sensitive material → perturbing the electromagnetic field distribution of the corresponding resonator unit → leading to a shift in the resonant frequency, amplitude attenuation, and phase jump of the S21 parameter → and collaborative analysis and inversion of the gas type and concentration through multiple resonant modes. This design verifies the technical feasibility of a microwave resonator array-based electronic nose device with a dual-layer detachable architecture in achieving high reusability and parallel detection of multiple gases.

[0061] Example 3

[0062] This embodiment provides a method for selecting gas-sensitive materials and analyzing gases in an electronic nose device based on a microwave resonator array. The advantage of this microwave resonator array-based electronic nose device lies in its multiple independent resonant modes, allowing for the analysis of a wider range of parameters, including resonant frequency shift, amplitude attenuation, and phase change. Utilizing the multi-parameter advantage of microwaves, multi-parameter inversion of gas type and concentration can be achieved.

[0063] This embodiment details the gas-sensitive material configuration strategy and gas analysis mechanism of an electronic nose device based on a microwave resonator array. The core advantage of this electronic nose device lies in its multimodal electromagnetic response (resonant frequency shift Δf, amplitude attenuation Δ|S21|, phase change) excited by nine independent resonator units. A three-dimensional parameter space is constructed to achieve holographic analysis of gas fingerprints. Addressing the urgent need for monitoring industrial toxic gas leaks (such as NO2) and food safety (such as ethanol / acetone residues), this embodiment selects ethanol, acetone, NH3, and NO2 as target gases.

[0064] Gas-sensitive materials can be further categorized into N-type semiconductors, P-type semiconductors, polymers, two-dimensional materials, ceramic materials, etc., depending on their type. The following are the material selections used in this embodiment:

[0065] N-type semiconductors include tin dioxide (SnO2), zinc oxide (ZnO), and titanium dioxide (TiO2). Among them, SnO2 is the most commonly used gas-sensitive material, as it is sensitive to a variety of oxidizing and reducing gases, such as ethanol, NO2, and CO. ZnO, on the other hand, is sensitive to gases such as ethanol and propanol. Based on the material's sensitivity to different gases, SnO2, ZnO, and TiO2 can be selected as N-type semiconductor gas-sensitive materials.

[0066] Furthermore, the oxygen vacancies on the surface of the anatase phase of TiO2 enhance the NO2 capture capability, and the cross-validation of the NO2 signal by the dual N-type semiconductors of SnO2 and ZnO complements the high sensitivity of SnO2 and the fast recovery characteristics of ZnO.

[0067] P-type semiconductors include nickel oxide (NiO), cobalt oxide (Co3O4), and copper oxide (CuO). Among them, Co3O4 is sensitive to gases such as VOCs and ethanol, NiO is sensitive to reducing gases, and CuO is relatively sensitive to gases such as CO and NO2. Therefore, these three materials were selected as gas-sensitive materials for P-type semiconductors.

[0068] Furthermore, the hole-dominated conductivity mechanism of NiO exhibits intrinsic selectivity for reducing gases. NH3 molecules react with pre-adsorbed oxygen (4NH3 + 5O2 → 4NO + 6H2O), releasing electrons to neutralize holes; the spinel structure of Co3O4 provides multiple oxidation states (Co... 2+ / Co 3+ This enhances the catalytic oxidation activity of NH3, with improved sensitivity compared to NiO. The p-type characteristics of CuO also aid in NH3 detection, and its surface Cu... + The active site preferentially adsorbs ammonia molecules.

[0069] For two-dimensional materials, MXene (Ti3C2T) was selected. x As a gas-sensitive material, the surface terminating groups (-OH / -F) form a directional hydrogen bond network with the ethanol hydroxyl groups, resulting in a significant change in the imaginary part (ε) of the dielectric constant. Its layered structure provides >800m 2 / g specific surface area enables highly sensitive ethanol detection.

[0070] As a polymer material, polyaniline (PANI) undergoes conformational changes in the aniline unit molecular chain induced by the carbonyl group (C=O) of acetone, altering the polaron migration pathway and enabling it to specifically distinguish between interference from ketones and alcohols.

[0071] Barium titanate (BaTiO3), a high-dielectric ceramic material, exhibits ferroelectric domain-flipping response in polar molecules, with a dielectric constant real part (ε') varying by up to 40%, thus broadening its temperature and humidity adaptability (-20℃ to 85℃). It can be used to eliminate the influence of environmental humidity on the accuracy of electronic nose devices.

[0072] When the mixed gas contacts the resonator array, the differentiated responses of the nine materials are manifested in three aspects: physical adsorption, chemical adsorption, and nodal perturbation: MXene / PANI captures ethanol / acetone through van der Waals forces / hydrogen bonds; NiO / Co3O4 catalyzes the oxidation of NH3, and TiO2 / SnO2 bonds NO2; BaTiO3 ferroelectrically responds to changes in the gas dipole moment. Each material may respond to the remaining gases; therefore, a neural network is used to jointly analyze the Δf, Δ|S21|, and Δ|S21| of the nine resonator units. It can specifically detect four gases: ethanol, acetone, NH3, and NO2.

[0073] like Figure 5 As shown, the S21 parameter changes caused by the introduction of 10-500 ppm acetone gas are illustrated, and the S21 response generated by the nine-resonator can be observed. Because they are coated with different gas-sensitive materials, their S21 changes differ when different concentrations of acetone are introduced. Based on this, multiple sets of data were tested, and the S21 parameters were input into a neural network for classification and regression. The final fitted curves of the predicted and target concentrations are shown below. Figure 6 As shown, R = 0.99, demonstrating extremely high detection accuracy. This indicates that the proposed electronic nose device based on a microwave resonator array is suitable for gas discrimination and detection, providing a new monitoring solution for industrial and food safety fields.

[0074] The above embodiments are merely preferred embodiments for fully illustrating the present invention. The 3×3 resonator array size is only for illustrative purposes; the resonator array size can be adjusted according to actual needs, such as 2×2, 2×3, 4×4, etc. The operating modes of the resonator arrays described in the above embodiments are not limited to a single mode; multiple operating modes can be performed simultaneously. For example, resonant mode 1 and resonant mode 2 can be performed simultaneously to achieve single-gas detection; resonant modes 1 to 6 can be performed simultaneously to achieve the differentiation and concentration detection of binary gases; and combined resonant modes 1 to 9 can achieve the specific identification and specific concentration detection of quaternary gases, effectively eliminating the influence of ambient temperature and humidity on detection accuracy.

[0075] Furthermore, depending on the gas to be detected, this electronic nose device can be combined with different sensitive materials, including but not limited to the gas-sensitive materials mentioned above. Combining the advantages of a detachable upper sensing array layer and microwave detection, the electronic nose can be reconstructed at room temperature. Specific material selection for the gas-sensitive materials includes:

[0076] In N-type semiconductors, indium oxide, tungsten oxide, iron oxide, etc., are sensitive to gases such as hydrogen, ozone, and oxygen. They are often doped with noble metals such as Pt, Pd, and Au or other oxides to improve selectivity and sensitivity.

[0077] Chromium trioxide and lanthanum oxide in P-type semiconductors are also sensitive to gases such as ethanol, NO2, and acetone. However, due to selectivity issues, they can be modified and optimized through methods such as doping, composite formation, and nanostructure regulation.

[0078] For conductive polymers, gas-sensitive materials that can be used include polypyrrole (sensitive to NO2, NH3, H2S, etc., with good film-forming properties), polythiophene and its derivatives (sensitive to VOCs, NH3, NO2, etc., with good performance tunability), and polystyrene sulfonate, etc.

[0079] Carbon-based materials include those with high specific surface area, electrical conductivity, and chemical stability, such as graphene, carbon nanotubes, and carbon black.

[0080] Metal-organic frameworks (MOFs) are a new type of porous crystalline material that are self-assembled from metal ions / clusters and organic ligands. Typical examples include the ZIF series, MIL series, and UiO series. They have ultra-high specific surface area, highly tunable pore size and chemical environment, strong adsorption capacity for specific gas molecules, and huge selectivity potential.

[0081] Other materials include, but are not limited to, the types mentioned above. In general, the choice of gas-sensitive material depends on the target gas, the application scenario (such as whether high-temperature operation is required, power consumption limitations, cost requirements, stability requirements, and whether flexibility is required), and the required performance indicators (sensitivity, selectivity, response speed, etc.). The selection of gas-sensitive materials can be flexibly adjusted according to actual usage needs and the size of the resonator array. Finally, classification and regression are performed using neural networks.

[0082] In summary, this invention provides an electronic nose device based on a microwave resonator array. This electronic nose device employs a dual-layer, detachable microwave sensor array architecture, i.e., a physically separated modular design. The lower feed layer integrates a ground plane, a lower dielectric substrate, and microwave coupling feed lines, while the upper sensor array layer is pluggable. A tight fit between the upper sensor array layer and the lower feed layer is achieved through fixing holes and screws on both layers. This design allows the upper sensor array layer to be flexibly replaced for different detection scenarios, significantly reducing maintenance costs and overcoming the limitation of single-use devices. On the surface of the upper sensor array layer, nine independently designed interdigital resonator units are arranged in a 3×3 matrix. Each resonator unit optimizes the excitation of a highly localized electromagnetic field distribution through unique geometric parameters, forming a spectrally separated resonant mode. The surface of each resonator unit can be functionally coated with a gas-sensitive material coating for specific gases (such as NH3 and VOCs). The change in dielectric constant caused by gas molecule adsorption perturbs the resonant frequency and amplitude, enabling parallel real-time detection of nine gases. Crucially, this electronic nose device can achieve a high sensitivity response at the ppm level at room temperature, completely avoiding the need for high-temperature operation and significantly improving the equipment's energy efficiency and safety.

[0083] like Figure 7 As shown, to fully analyze the response signals such as frequency shift, amplitude attenuation, and phase change generated by the sensor array, and to solve the problem of cross-interference in mixed gases, this invention provides a gas analysis method based on a microwave resonator array. It employs an electronic nose device based on a microwave resonator array. This gas analysis method is based on a graph neural network (GNN) model, where each resonator unit is modeled as a graph node, its response characteristics are used as node attributes, and the electromagnetic coupling effect and spatial correlation between units are constructed as dynamic edge weights. Through hierarchical graph convolution and attention mechanisms, the network can adaptively learn the microwave parameter changes caused by different gas combinations, first completing the qualitative classification of multi-component gases, and then accurately quantifying the concentration of each gas through regression branches. The gas analysis method specifically includes the following steps:

[0084] Step 1: Introduce the gas to be tested into the resonator array 7 and collect the S21 parameters of each resonator unit, including the resonant frequency shift Δf, amplitude attenuation Δ|S21|, and phase change. (Additionally, parameters such as quality factors and Smith charts can be used), and the changes in the above characteristics can be calculated;

[0085] Step 2: Treat each resonator unit as an independent graph node to construct a graph structure. Node features include Δf, Δ|S21|, Based on electromagnetic coupling characteristics, dynamic edges are generated: frequency coupling edges are established when the resonant frequency difference is <50MHz, and response correlation edges are established when the response change correlation coefficient is >0.7;

[0086] Step 3: Analyze mixed gas information using a hierarchical attention mechanism, a gas-level attention mechanism, and a concentration-level attention mechanism. Calculate gas-specific weights based on a pre-defined learnable lookup table for each gas. Generate concentration-sensitive weights using a lightweight multilayer perceptron. Use multi-head attention to filter key resonator nodes. Finally, focus on key feature regions of different gas responses.

[0087] Step 4: Parallel output of three collaborative tasks: one is gas presence detection, which uses a sigmoid classifier to output the probability of gas presence; another is gas type classification, which concatenates node-level features and uses a Sofrmax classifier to give probabilities for different combinations; and the last is concentration probability modeling, which outputs a 95% confidence interval based on the concentration expectation.

[0088] This collaborative innovation of hardware and algorithms not only breaks through the traditional sensor's "single-target" detection bottleneck, but also provides a methodological foundation for building a high-throughput, low-power intelligent gas sensing platform.

[0089] This invention utilizes the varying adsorption capacities of different gas-sensitive materials for different gases to achieve selective gas detection and multi-mode joint analysis. Because the near-field dielectric properties of the resonator change after the gas-sensitive material adsorbs gas molecules, the corresponding resonant modes also change. The electronic nose device leverages this multi-mode variation to simultaneously detect multiple gases and effectively resists the influence of ambient temperature and humidity on gas detection accuracy. This electronic nose device has a simple fabrication process, high sensitivity, and is suitable for industrial environments and food quality monitoring, showing broad application prospects in gas monitoring and sensor integration.

[0090] This document uses specific embodiments to illustrate the principles and implementation methods of the present invention. The descriptions of these embodiments are merely for the purpose of helping to understand the method and core ideas of the present invention. It should be noted that those skilled in the art can make various improvements and modifications to the present invention without departing from its principles, and these improvements and modifications also fall within the scope of protection of the claims of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. An electronic nose device based on a microwave resonator array, characterized in that, The system includes an upper sensing array layer (1) and a lower feed layer (2) located below the upper sensing array layer (1). The lower feed layer (2) includes a ground plane (3), a lower dielectric substrate (4) disposed on the ground plane (3), and a microwave coupling feed line (5) disposed on the lower dielectric substrate (4). The upper sensing array layer (1) includes an upper dielectric substrate (6) and a resonator array (7) disposed on the upper dielectric substrate (6). The resonator array (7) includes a plurality of resonator units, and the plurality of resonator units are coated with different gas-sensitive material coatings (8) for detecting multi-component gases. The upper sensing array layer (1) and the lower feed layer (2) are detachably connected by screws. The resonator array (7) includes nine independently designed interdigital resonator units arranged in a 3×3 matrix, which can generate resonant modes at different frequencies depending on the size of the resonators. The exposed surface of the resonator array (7) is coated with nine gas-sensitive materials, including three P-type semiconductors, three N-type semiconductors, a polymer, a two-dimensional material and a ceramic material, to selectively adsorb target gas molecules. The equivalent circuit of each resonator unit in the resonator array (7) is represented as an RLC series circuit, and its resonant frequency is: in It is the equivalent inductance of the resonator. It refers to the change in capacitance after the gas-sensitive material binds to gas molecules. It is the equivalent capacitance of the resonator; Each of the gas-sensitive material coatings (8) is sensitive to one or more gases, and the change in the dielectric properties of the material when a gas is adsorbed onto the gas-sensitive material coating (8) is determined by the Debye model: （1） （2） in , , It is the relaxation frequency. It is the resonant frequency. f The real part as it approaches infinity for f The real part at →0 indicates that the higher the operating frequency, the greater the change in dielectric constant, and the greater the change in capacitance of the resonator caused by gas adsorption. The larger the value, the more significant the change in resonant frequency and S21 parameter.

2. The electronic nose device based on a microwave resonator array according to claim 1, characterized in that, The resonator array (7), microwave coupling feed line (5), and ground plane (3) are made of conductive materials.

3. The electronic nose device based on a microwave resonator array according to claim 1, characterized in that, The upper sensing array layer (1) and the lower feeding layer (2) are both prepared on a polytetrafluoroethylene dielectric substrate by wet etching process.

4. The electronic nose device based on a microwave resonator array according to claim 1, characterized in that, The N-type semiconductors include SnO2, ZnO, and TiO2. The oxygen vacancies on the anatase phase surface of TiO2 enhance NO2 capture capability. The dual N-type semiconductors of SnO2 and ZnO cross-validate the NO2 signal, with SnO2's high sensitivity and ZnO's fast recovery characteristics complementing each other. The P-type semiconductors include NiO, Co3O4, and CuO. NiO's hole-dominated conductivity mechanism exhibits intrinsic selectivity for reducing gases; NH3 molecules react with pre-adsorbed oxygen, releasing electrons to neutralize holes. Co3O4's spinel structure provides multiple oxidation states, enhancing its catalytic oxidation activity for NH3 and improving sensitivity compared to NiO. CuO's P-type characteristics assist in NH3 detection, and its surface Cu... + The active sites preferentially adsorb ammonia molecules; the two-dimensional material selected is Ti3C2T. x The surface-terminating groups form a directional hydrogen bond network with the ethanol hydroxyl groups, resulting in a significant change in the imaginary part of the dielectric constant. Its layered structure provides a specific surface area of ​​>800 m² / g, enabling highly sensitive ethanol detection. As a polymer material, polyaniline exhibits conformational changes in the aniline unit molecular chain induced by acetone carbonyl groups, altering the polaron migration path and enabling specific differentiation between ketone and alcohol interference. The strong dielectric ceramic material barium titanate, with its ferroelectric domain flipping response polar molecules, shows a 40% change in the real part of the dielectric constant, broadening its temperature and humidity adaptability and eliminating the influence of environmental humidity on the accuracy of electronic nose devices.

5. A gas analysis method based on a microwave resonator array, characterized in that, An electronic nose device based on a microwave resonator array, as described in any one of claims 1-4, is used. This gas analysis method is based on a graph neural network model and specifically includes the following steps: Step 1: Pass the gas to be tested into the resonator array (7), collect the S21 parameters of each resonator unit, including the resonant frequency shift Δf, amplitude attenuation Δ|S21|, and phase change Δφ, and calculate the changes of the above characteristics; Step 2: Treat each resonator unit as an independent graph node to construct a graph structure. The node features include Δf, Δ|S21|, and Δφ. Generate dynamic edges based on electromagnetic coupling characteristics: establish frequency coupling edges when the resonant frequency difference is <50MHz, and establish response correlation edges when the response change correlation coefficient is >0.

7. Step 3: Analyze mixed gas information using a hierarchical attention mechanism, a gas-level attention mechanism, and a concentration-level attention mechanism. Calculate gas-specific weights based on a pre-defined learnable lookup table for each gas. Generate concentration-sensitive weights using a lightweight multilayer perceptron. Use multi-head attention to filter resonator nodes and finally focus on key feature regions of different gas responses. Step 4: Parallel output of three collaborative tasks: one is gas presence detection, which uses a sigmoid classifier to output the probability of gas presence; another is gas type classification, which concatenates node-level features and uses a Sofrmax classifier to give probabilities for different combinations; and the last is concentration probability modeling, which outputs a 95% confidence interval based on the concentration expectation.

6. The application of an electronic nose device based on a microwave resonator array as described in any one of claims 1-4 in multi-gas specific identification.

Images