Multi-channel electrochemical detection electrode, electrochemical sensor and detection method

Abstract

The invention relates to the technical field of electrochemical sensors, and provides a multi-channel electrochemical detection electrode, an electrochemical sensor and a detection method.The multi-channel electrochemical detection electrode comprises an electrode substrate and a plurality of working electrodes arranged on the electrode substrate; the surfaces of the plurality of working electrodes are respectively modified with different types of metal nano-catalysts to detect different gas components, and the surfaces of the working electrodes are covered with PDMS hydrophobic layers to form a hydrophilic / hydrophobic channel separation structure. The surfaces of a plurality of working electrodes are modified with different types of metal nano-catalysts and covered with PDMS hydrophobic layers to form a hydrophilic / hydrophobic channel separation structure, and the multi-channel electrochemical detection electrode separates gas molecules through hydrophilic / hydrophobic channels, so that the cross reaction can be inhibited, and the cross sensitivity can be reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrochemical sensor, in particular to a multi-channel electrochemical detection electrode, an electrochemical sensor and a detection method. BACKGROUND

[0002] An electrochemical sensor is an instrument for measuring various chemical components. The electrochemical sensor works by reacting with the measured gas and generating an electric signal proportional to the gas concentration. A typical electrochemical sensor is composed of a sensing electrode (or working electrode) and a counter electrode, and is separated by a thin electrolytic layer.

[0003] Electrochemical gas sensors are widely used in industrial safety, environmental monitoring and other fields due to their high sensitivity, selectivity and real-time monitoring capability. Traditional electrochemical sensors usually use single-channel working electrode design to realize the detection of a single gas (such as H2, CO, NOx) through oxidation-reduction reaction. The core detection principle is based on Faraday's law or Nernst equation, which quantifies the target gas concentration through current or potential change. For example, Pt electrode is used for hydrogen H2 detection, and WO3 film is used for NOx monitoring. Such design performs well in a single gas environment. However, traditional electrochemical sensor electrodes mostly use homogeneous catalytic layers (such as Pt, Pd, etc. single metal), which are difficult to achieve specific recognition of multi-component gas. For example, the oxidation-reduction potential of H2 and CO on the Pt electrode is similar, which easily leads to cross interference. SUMMARY

[0004] Therefore, in order to reduce the cross-sensitivity of the electrochemical sensor electrode detection, the present application provides a multi-channel electrochemical detection electrode, an electrochemical sensor and a detection method, and the specific technical solutions are as follows:

[0005] A multi-channel electrochemical detection electrode, comprising an electrode substrate and a plurality of working electrodes arranged on the electrode substrate, the surfaces of the plurality of working electrodes are respectively modified with different kinds of metal nano catalysts to detect different gas components, and the surfaces of the working electrodes are covered with a PDMS hydrophobic layer to form a hydrophilic / hydrophobic channel separation structure.

[0006] The multi-channel electrochemical detection electrode forms a multi-channel detection electrode array by arranging a plurality of working electrodes on the electrode substrate. By modifying different kinds of metal nano catalysts on the surfaces of the plurality of working electrodes and covering the PDMS hydrophobic layer to form a hydrophilic / hydrophobic channel separation structure, the multi-channel electrochemical detection electrode separates gas molecules through hydrophilic / hydrophobic channels, which can inhibit cross-reaction and reduce cross-sensitivity.

[0007] Preferably, the electrode substrate is made of 3D printed porous graphene material, and the different kinds of metal nano-catalysts include Pt / Pd bimetallic catalysts and WO3 / ZnO heterojunction catalysts.

[0008] A multi-channel electrochemical sensor comprising the multi-channel electrochemical detection electrode.

[0009] Preferably, the multi-channel electrochemical sensor further comprises a concentration calculation unit, which is configured to:

[0010] obtain a cross-sensitivity correction factor according to the concentration information of the interference gas and the cross-sensitivity coefficient;

[0011] obtain an environmental interference factor according to the environmental temperature and humidity information;

[0012] and obtain the initial molar concentration value of the target gas according to the cross-sensitivity correction factor, the environmental interference factor, the target gas basic information, the working electrode basic information, and the detection response current of the working electrode.

[0013] A multi-channel electrochemical sensor detection method applied to the multi-channel electrochemical sensor, comprising the following steps:

[0014] for a target gas, obtaining the detection response current of the working electrode of the channel corresponding to the target gas;

[0015] obtaining the initial molar concentration value of the target gas according to the detection response current.

[0016] Preferably, the specific method for obtaining the initial molar concentration value of the target gas according to the detection response current comprises the following steps:

[0017] obtaining a cross-sensitivity correction factor according to the concentration information of the interference gas and the cross-sensitivity coefficient;

[0018] obtaining an environmental interference factor according to the environmental temperature and humidity information;

[0019] and obtaining the initial molar concentration value of the target gas according to the cross-sensitivity correction factor, the environmental interference factor, the target gas basic information, the working electrode basic information, and the detection response current.

[0020] Preferably, the multi-channel electrochemical sensor detection method further comprises the following steps:

[0021] constructing a CNN-LSTM dual-channel network;

[0022] performing convolution operation on the current signals of the multi-channel working electrodes through the CNN branch to extract spatial features;

[0023] The current signal of the multi-channel working electrode is modeled by the LSTM branch to extract the time sequence feature.

[0024] The spatial feature and the time sequence feature are fused to obtain a predicted concentration value of the kth gas.

[0025] Preferably, the specific method for constructing the CNN-LSTM dual-channel network comprises the following steps:

[0026] A loss function is constructed.

[0027] The CNN-LSTM dual-channel network is constructed according to the loss function. BRIEF DESCRIPTION OF DRAWINGS

[0028] The present application can be further understood from the following description in conjunction with the drawings. The components in the drawings are not necessarily drawn to scale, but emphasis is instead placed on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0029] Figure 1 is a schematic diagram of the overall structure of a multi-channel electrochemical detection electrode in an embodiment of the present application;

[0030] Figure 2 is a schematic diagram of the overall process of a multi-channel electrochemical sensor detection method in an embodiment of the present application;

[0031] Figure 3 is a schematic diagram of the process of a specific method for obtaining the initial molar concentration value of the target gas in an embodiment of the present application;

[0032] Figure 4 is a schematic diagram of the overall process of a multi-channel electrochemical sensor detection method in another embodiment of the present application.

[0033] BRIEF DESCRIPTION OF DRAWINGS

[0034] 1, electrode substrate, 2, working electrode, 3, reference electrode, 4, counter electrode. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the protection scope of the present application.

[0036] It should be understood that when an element as a layer, region or plate is referred to as being "on" another element, it can be directly on the other element or intervening elements can also be present. In addition, it should be understood that when an element is referred to as being "connected" to or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements can be present. As used herein the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description of the application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0038] The terms "first", "second", etc. are used herein only to describe the names of the elements, and do not represent a specific number and order.

[0039] Before the specific embodiments of the present application are described, the prior art will be briefly introduced.

[0040] An electrochemical sensor works by reacting with the gas being measured and producing an electrical signal that is proportional to the gas concentration. A typical electrochemical sensor consists of a sensing electrode (or working electrode) and a counter electrode, separated by a thin electrolyte layer. The gas first reacts with the sensor through a small capillary-type opening, then a hydrophobic barrier layer, and finally reaches the electrode surface. This method allows the appropriate amount of gas to react with the sensing electrode to form a sufficient electrical signal, while preventing the electrolyte from leaking out of the sensor.

[0041] The gas that diffuses through the barrier reacts with the sensing electrode, which can use either an oxidation or reduction mechanism. These reactions are catalyzed by the electrode material designed for the gas being measured. Through a resistor connected between the electrodes, a current proportional to the concentration of the gas being measured flows between the anode and the cathode. By measuring this current, the gas concentration can be determined. Because of the current generated in this process, electrochemical sensors are also commonly referred to as amperometric gas sensors or micro fuel cells.

[0042] Existing sensors are susceptible to cross-sensitivity in complex gas mixture environments. For example, both carbon monoxide CO and hydrogen H2 can undergo oxidation reactions on Pt electrodes, resulting in overlapping current signals. Traditional sensors can suppress interference through physical filtration or chemical modification, but the selectivity improvement is limited and cannot achieve simultaneous detection of multiple components. In addition, environmental parameters such as temperature and humidity significantly affect electrode activity. Experiments show that for every 10℃ increase in temperature, the response current fluctuation of Pt electrodes to H2 can reach ±15%, and existing compensation schemes (such as linear interpolation) are difficult to correct non-linear drift in real time.

[0043] Although some studies attempt to introduce machine learning to optimize electrochemical signals, deep learning models have poor generalization ability, and single neural networks cannot integrate time series signals (LSTM) and spatial features (CNN), resulting in weak correlation between multi-channel data. Moreover, the existing algorithm lacks a dynamic updating mechanism, and it relies on offline training, which cannot adapt to electrode aging or environmental mutation scenarios, and the system robustness is insufficient.

[0044] To reduce the cross-sensitivity of electrochemical sensor electrodes, as shown in Figure 1 An embodiment of the multi-channel electrochemical detection electrode includes an electrode substrate 1 and a plurality of working electrodes 2 disposed on the electrode substrate. The surfaces of the plurality of working electrodes are respectively modified with different types of metal nano-catalysts to detect different gas components, and the working electrode surfaces are covered with a PDMS hydrophobic layer to form a hydrophilic / hydrophobic channel separation structure.

[0045] As a preferred technical solution, the multi-channel electrochemical detection electrode further includes a reference electrode 3 and a counter electrode (auxiliary electrode) 4. The electrode substrate is made of 3D printed porous graphene material, and the different types of metal nano-catalysts include Pt / Pd bimetallic catalysts and WO3 / ZnO heterojunction catalysts.

[0046] Of course, the plurality of different types of metal nano-catalysts also include MOFs material catalysts. By modifying the surfaces of the plurality of working electrodes with different types of metal nano-catalysts to detect different gas components, specific catalytic interfaces can be designed for target gases such as H2 (hydrogen), CO (carbon monoxide), and NOx (nitrogen oxides).

[0047] By covering the working electrode surfaces with a microfluidic hydrophobic layer (polydimethylsiloxane-PDMS), gas molecules can be separated by hydrophilic / hydrophobic channel separation, and cross-reactions can be suppressed.

[0048] As a preferred technical solution, the plurality of working electrodes 2 correspond to the reference electrodes 3 one by one, and the plurality of working electrodes 2 correspond to the counter electrodes one by one, each of the working electrodes 2, the reference electrodes 3 corresponding to the working electrodes 2 and the counter electrodes 4 adopts a ring electrode layout, and the working electrodes 2, the reference electrodes 3 and the counter electrodes 4 are distributed at 120 degrees from each other.

[0049] Of course, in order to simplify the structure, a plurality of working electrodes can share one reference electrode and one counter electrode, and the electrical connection relationship between the plurality of working electrodes and the reference electrode is switched by a multi-channel switch. Similarly, a plurality of working electrodes can share one counter electrode, and the electrical connection relationship between the plurality of working electrodes and the counter electrode is switched by a multi-channel switch.

[0050] Specifically, the electrode base can be made of ceramic / Al2O3, and has a circular ring structure with a diameter of 5-10 mm, the working electrode, the reference electrode and the counter electrode are distributed radially symmetrically, the central angle is 120°, forming an equilateral triangle topology. The gap between the electrodes is covered with a 50μm thick PDMS hydrophobic film with a pore size of less than 1μm.

[0051] The working electrode 2, the reference electrode 3 and the counter electrode 4 are distributed at 120 degrees from each other, so that the electric field forms a uniform gradient field around the WE, which can eliminate the edge effect of the traditional linear layout, improve the surface current density of the WE (Working Electrode), and enhance the current response sensitivity. In addition, the ring symmetric structure suppresses environmental electromagnetic interference (EMI), reduces noise level and cross sensitivity.

[0052] In summary, the multi-channel electrochemical detection electrode forms a multi-channel detection electrode array by arranging a plurality of working electrodes on the electrode base. By modifying different types of metal nano catalysts on the surface of the plurality of working electrodes and covering a PDMS hydrophobic layer to form a hydrophilic / hydrophobic channel separation structure, the multi-channel electrochemical detection electrode separates gas molecules through the hydrophilic / hydrophobic channel, which can suppress cross-reaction and reduce cross-sensitivity.

[0053] In an embodiment of the present application, a multi-channel electrochemical sensor comprises the multi-channel electrochemical detection electrode.

[0054] The multi-channel electrochemical sensor further comprises a sampling unit, a voltage amplifier and a current amplifier, the sampling unit is used to collect the actual voltage signal of the voltage detection loop and the actual current signal of the current detection loop; wherein each of the working electrodes, the reference electrodes corresponding to the working electrodes and the voltage amplifier constitute a detection voltage loop, and each of the working electrodes, the counter electrodes corresponding to the working electrodes and the current amplifier constitute a current detection loop.

[0055] The detection voltage circuit is used to detect the detection response voltage of the working electrode to the target gas, and the current detection circuit is used to detect the detection response current of the working electrode to the target gas. After obtaining the detection response voltage and the detection response current, and after amplification processing, the molar concentration value of the target gas can be obtained through table lookup or the like.

[0056] Here, a multi-stage gas path microcavity (aperture gradient: 5 μm→100 nm) can also be designed based on MEMS technology, to separate mixed gases through molecular sieve effect and reduce cross-sensitivity.

[0057] Since the multi-channel electrochemical detection electrode is formed by modifying different types of metal nano-catalysts on the surface of multiple working electrodes and covering a PDMS hydrophobic layer to form a hydrophilic / hydrophobic channel separation structure, gas molecules are separated through the hydrophilic / hydrophobic channel, cross-reactions can be inhibited, cross-sensitivity can be reduced, and therefore the multi-channel electrochemical sensor can also reduce the cross-sensitivity of the electrode detection of the electrochemical sensor and improve the detection accuracy of the molar concentration of the target gas.

[0058] As a preferred technical solution, the multi-channel electrochemical sensor further comprises a concentration calculation unit, which is configured to: obtain a cross-sensitivity correction factor according to the concentration information of the interference gas and the cross-sensitivity coefficient; obtain an environmental interference factor according to the environmental temperature and humidity information; and obtain an initial molar concentration value of the target gas according to the cross-sensitivity correction factor, the environmental interference factor, the target gas basic information, the working electrode basic information, and the detection response current of the working electrode.

[0059] The concentration calculation unit obtains the initial molar concentration value of the target gas according to the formula wherein C i represents the initial molar concentration value of the target gas in the i-th channel, the target gas basic information includes the diffusion coefficient D of the target gas in the electrolyte and the number of electrons transferred n by the target gas molecules participating in the electrochemical reaction, the working electrode basic information includes the effective surface area A of the working electrode and the catalytic reaction order α, I i represents the detection response current of the working electrode in the i-th channel, Σ j≠i (β ij ·C j ) represents the cross-sensitivity correction factor, β ij represents the cross-sensitivity coefficient of the j-th interference gas to the i-th channel, C i represents the molar concentration value of the j-th interference gas, ε(T,H) represents the environmental interference factor, ε represents the temperature and humidity comprehensive interference function, T and H represent the environmental temperature value and the humidity value respectively, and F represents the Faraday constant, represents the time rate of change of the applied potential.

[0060] Specifically, the Faraday constant characterizes the relationship between charge and molar number, and is generally taken as 96485C / mol. A is determined by the working electrode structure, D is affected by temperature and material porosity, and the catalytic reaction order reflects the nonlinear effect of the catalytic activity of the electrode material on the reaction rate. Generally speaking, α is between 0.5-1.0. Used to control the kinetics of electrochemical reactions. ij It is dimensionless and can be determined through calibration experiments. The ambient temperature value affects the electrolyte ion mobility and reaction rate constant, and the ambient humidity value changes the electrolyte conductivity and gas diffusion efficiency.

[0061] The temperature and humidity comprehensive interference function can be calibrated experimentally. Specifically, the following steps are performed: first, a temperature gradient (e.g., -20°C to 80°C) and a humidity gradient (0% to 99% RH) are set in a temperature and humidity chamber to cover the sensor's operating range. Then, the output current / voltage values ​​of the sensor in the absence of target gas (only environmental interference) are recorded as baseline interference data. Finally, the data are fitted using the least squares method to establish a polynomial interference function model. This polynomial interference function model can be expressed as ε(T,H) = α0+α1T+α2H+α3T 2 +α4T·H+α5H 2 +α6T 3 .

[0062] The catalytic reaction order α can be determined by measuring the current response of the logarithmic concentration gradient under constant temperature and humidity conditions, and then by the linear regression slope. Specifically, under constant temperature and humidity conditions, different concentrations of target gas are introduced, and the current response I is recorded. By fitting ln(I) and ln(C i ), the slope is the value of α.

[0063] When α = 1, the current response is linear with concentration, corresponding to a fast reaction on an ideal electrode surface (such as the oxidation of hydrogen by a platinum catalyst). When α = 0.5, the current response is proportional to the square root of the concentration, corresponding to a diffusion-controlled reaction (such as the restricted diffusion of gas in a porous electrode). The larger the α, the more sensitive the current response to concentration changes, making it suitable for trace detection.

[0064] β ij A single interfering gas can be introduced into the sensor, the current changes in each channel can be recorded, and a cross-sensitivity matrix can be constructed to determine the effect.

[0065] pass The coupling between them can distinguish the concentration contribution of target gas and interfering gas, and combine with the subsequent neural network (such as L1 regularization to suppress β ij ) to achieve multi-component analysis. By introducing environmental interference factors, baseline shifts caused by thermal drift and humidity can be corrected.

[0066] After calculating the initial molar concentration value of the target gas of the i-th channel according to the formula After calculating the initial molar concentration value of the target gas of the i-th channel according to the formula

[0067] In summary, the multi-channel electrochemical sensor innovatively introduces the catalytic reaction order α, calibrates the nonlinear reaction order of different catalytic interfaces (such as Pt / Pd bimetal) through experiments, adapts to complex catalytic scenarios, and uniformly models electrochemical reactions, molecular diffusion, and environmental interference, thereby breaking through the limitations of traditional single-variable models and providing physical meaning variables (such as CNN-LSTM network directly calling β ij matrix correction cross interference) as input features of the neural network.

[0068] As shown in the multi-channel electrochemical sensor detection method according to an embodiment of the present application, applied to the multi-channel electrochemical sensor, the method comprises the following steps: Figure 2 S1, for a target gas, obtaining a detection response current of a working electrode of a channel corresponding to the target gas.

[0069] S2, obtaining an initial molar concentration value of the target gas according to the detection response current.

[0070] Specifically, based on the detection response current, the initial molar concentration value of the target gas can be obtained through a table lookup method, or the initial molar concentration value of the target gas can be obtained according to a preset detection response current-initial molar concentration value function formula.

[0071] Of course, based on the detection response current and / or the detection response voltage, a reference molar concentration value can be obtained through a table lookup method, and the first one of the maximum value, the minimum value, the average value, or the weighted average value of the initial molar concentration value and the reference molar concentration value is selected as the final molar concentration value.

[0072] As a preferred technical solution, in step S2, as shown in the multi-channel electrochemical sensor detection method according to an embodiment of the present application, the specific method for obtaining the initial molar concentration value of the target gas according to the detection response current comprises the following steps:

[0073] Figure 3 S21, obtaining a cross-sensitivity correction factor according to concentration information of an interfering gas and a cross-sensitivity coefficient;

[0074] S21, obtaining a cross-sensitivity correction factor according to concentration information of an interfering gas and a cross-sensitivity coefficient; ​

[0075] S22, obtaining an environmental interference factor according to the environmental temperature and humidity information.

[0076] S23, and obtaining an initial molar concentration value of the target gas according to the cross-sensitivity correction factor, the environmental interference factor, the target gas basic information, the working electrode basic information, and the detection response current.

[0077] The initial molar concentration value Wherein, the target gas basic information includes a diffusion coefficient D of the target gas in the electrolyte and an electron transfer number n of the target gas molecules participating in the electrochemical reaction, the working electrode basic information includes an effective surface area A of the working electrode and a catalytic reaction order a, I i represents the detection response current of the working electrode of the i-th channel, Σ j≠i (β ij ·C j ) represents the cross-sensitivity correction factor, β ij represents the cross-sensitivity coefficient of the j-th interfering gas to the i-th channel, C i represents the molar concentration value of the j-th interfering gas, ε(T, H) represents the environmental interference factor, ε represents the temperature and humidity comprehensive interference function, T and H represent the environmental temperature value and the humidity value respectively, and F represents the Faraday constant, represents the time rate of change of the applied potential.

[0078] The multi-channel electrochemical sensor detection method innovatively introduces the catalytic reaction order a, calibrates the nonlinear reaction order of different catalytic interfaces (such as Pt / Pd bimetal) through experiments, adapts to complex catalytic scenarios, and uniformly models the electrochemical reaction, molecular diffusion, and environmental interference, thereby breaking through the limitations of traditional single-variable models, separating the main effect, cross effect, and environmental effect, and providing physical meaning clear variables (such as CNN-LSTM network directly calling β ij matrix correction cross interference) for input features of the neural network.

[0079] As a preferred technical solution, as shown in Figure 4 The multi-channel electrochemical sensor detection method further includes the following steps:

[0080] S3, constructing a CNN-LSTM dual-channel network. The CNN-LSTM dual-channel network includes a CNN branch, an LSTM branch, and a fusion layer.

[0081] S4, performing convolution operation on the current signals of the multi-channel working electrodes through the CNN branch to extract spatial features F CNN .

[0082] Specifically, the spatial features of multi-electrode time domain signals are extracted through convolutional neural networks (CNN) to capture the spatial correlation between different electrode channels. The convolution kernel size is designed to be consistent with the number of channels to ensure that each convolution kernel can cover the input data of all electrode channels, that is, the convolution kernel size = the number of channels. Assuming that the input signal contains N electrode channels, the width of the convolution kernel is N, and the height is the length of the time window (such as 10 time steps), the spatial features across channels are extracted through a sliding window. The current signal of the multi-channel working electrode is convolved, and after pooling and activation function (such as ReLU), the spatial feature vector F is output. CNN .

[0083] S5, model the current signal of the multi-channel working electrode through the LSTM branch and extract the time series feature F LSTM .

[0084] Specifically, one of the functions of the LSTM branch is to use the long-short-term memory network to capture the nonlinear law of gas concentration changes over time, solving the problem that traditional methods are insufficient in modeling dynamic concentration evolution. The time series data of each electrode channel (such as the current value sampled once per second) is input to the LSTM in a sequence of [T×N] (T is the number of time steps and N is the number of channels). The LSTM uses hidden states to memorize long-term dependencies and ultimately outputs a time series feature vector F LSTM .

[0085] S6, fusing the spatial features and temporal features to obtain the predicted concentration value C of the kth type of gas k =σ(W c ·[F CNN ,F LSTM ]+b c ); where σ represents the activation function, W c represents the weight matrix, b c Represents the bias vector.

[0086] The fusion layer fuses the features of the CNN and LSTM through a fully connected layer (FC), and uses an attention mechanism (optional) to weight important features and output the final prediction result. The bias vector is used to improve the nonlinear expression capability of the model.

[0087] [F CNN ,F LSTM ] means concatenating the feature vectors of CNN and LSTM into a higher-dimensional feature vector, preserving the complementarity of spatial and temporal information, W c ·[F CNN ,F LSTM ]+b cThis represents a weighted summation of the concatenated features and maps them to the target output space (e.g., gas concentration dimension). The activation function is typically a Sigmoid activation function, which maps the output to a probability interval (0, 1) and is suitable for normalized concentration prediction.

[0088] More specifically, when detecting mixed gases H2 and CO, the CNN branch extracts the specific response F of the Pt electrode to H2. CNN , the LSTM branch captures the fluctuation of CO concentration over time F LSTM After fusion, C H2 and C CO Corrected the C in single channel respectively H2 and C CO cross interference.

[0089] Here, the CNN extracts spatial coupling effects between electrodes (such as cross-sensitivity), while the LSTM models the time-varying characteristics of gas concentration. The two complement each other to enhance robustness. By adaptively adjusting the contribution of spatial and temporal features through the weight matrix, overfitting caused by the dominance of a single modality can be avoided.

[0090] The initial molar concentration value of the target gas has certain cross-interference and environmental noise. Relative to the initial molar concentration value of the target gas, by constructing a CNN-LSTM dual-channel network to obtain the predicted concentration value of the k-th gas, the cross-interference and environmental noise in the single channel can be eliminated.

[0091] In general, the constructed CNN-LSTM dual-channel network uses the CNN-LSTM dual-channel architecture to model the spatial correlation and temporal evolution of electrochemical signals respectively, and finally outputs high-precision detection results through an interpretable fusion layer, which can achieve collaborative optimization of multimodal features and thus improve the detection accuracy of gas concentration.

[0092] Preferably, after obtaining the predicted concentration value of the k-th type of gas, the final molar concentration value of the target gas is C'=γ1·C k +γ2·C i Where γ1 and γ2 represent the weights of the predicted concentration and initial molar concentration of the kth target gas, respectively. By setting γ1 and γ2, we can better balance the weights of the predicted and initial molar concentrations of the target gas. This overcomes the limitations of traditional single-variable models and enables the coordinated optimization of multimodal features, thereby improving gas concentration detection accuracy.

[0093] Preferably, in step S3, the specific method of constructing the CNN-LSTM dual-channel network includes the following steps:

[0094] S31, construct loss function

[0095] S32, constructing a CNN-LSTM dual-channel network according to the loss function;

[0096] wherein N represents the number of samples, C true represents the true gas concentration value, represents the predicted gas concentration value, ||·||2 represents the L2 norm, λ represents the regularization coefficient, M represents the number of cross-sensitivity coefficients, and |·| represents the L1 regularization.

[0097] Specifically, represents the concentration prediction error term, which is used to measure the prediction accuracy of the model, and by minimizing the difference between the predicted value and the true value, the neural network output is ensured to be close to the true concentration distribution. The L2 norm is more sensitive to large errors, which can effectively suppress the influence of outliers.

[0098] represents the cross-sensitivity regularization term, which sums the absolute values of the cross-sensitivity coefficients through L1 regularization, forcing the model to reduce the weight of irrelevant interfering gases, reducing the crosstalk error between multiple gases (such as excluding CO signal interference when detecting H2), preventing the model from overfitting to noise or abnormal samples, and improving the generalization ability of the sensor in complex environments.

[0099] For the CNN-LSTM dual-channel network, terminal data local training can be realized by an edge computing device (such as Raspberry Pi).

[0100] The loss function is designed specifically for multi-channel electrochemical sensors and optimized for the following problems:

[0101] 1. Multi-gas interference problem: the response of electrodes to different gases has cross-sensitivity (such as the overlap of Pt / Pd bimetallic catalyst response to H2 and CO), and the regularization term sparsifies the β ij coefficient matrix, suppressing the influence of invalid channels.

[0102] 2. Dynamic environment adaptability: the L2 regularization term ensures the main concentration prediction accuracy, and the L1 regularization term dynamically compresses the interference signal weight, adapting to temperature and humidity fluctuations or changes in gas mixture concentration.

[0103] 3. Hardware resource constraints: L1 regularization can simplify the model complexity and reduce the computational power requirements of embedded devices (such as Raspberry Pi), adapting to edge computing scenarios.

[0104] As a preferred solution, when the current rate of change of any two channels |ΔI i / Δt|>δ and the correlation coefficient ρ ij <μ, the full-channel scanning mode is activated. Wherein the correlation coefficient Cov(I i ,I j) represents the current signal covariance of channels i and j, σ i j respectively represent the current signal standard deviation of channels i and j. δ and μ respectively represent the preset current change threshold and correlation coefficient threshold. Generally, the current change threshold is 0.1 μA / s, and the correlation coefficient threshold is 0.3.

[0105] When ρ ij < μ, it can be understood that the change trends of the two-channel signals are significantly different, different gas components (such as H2 and CO) can be detected, or there is nonlinear interference (such as different effects of temperature and humidity fluctuations on different electrodes); when ρ ij ≥ μ, the signal change trends are similar, which can be caused by the same gas or common mode interference (such as power supply noise), and full-channel scanning is not needed.

[0106] In a mixed gas environment, different gases can exist at the same time (such as H2 and CO coexisting). Full-channel scanning can synchronously collect all electrode signals, avoiding single-channel misjudgment. For example, Pt / Pd electrodes are sensitive to H2, and WO3 / ZnO electrodes are sensitive to CO, so multi-channel data fusion is needed to distinguish the gas types. By comparing the differences between multi-channel signals, a neural network dynamic correction algorithm (such as a CNN-LSTM hybrid network) is used to separate cross-sensitive signals, reduce the false alarm rate, and suppress cross interference.

[0107] That is, when the current change rate |ΔI i / Δt|> δ and the correlation coefficient ρ ij < μ of any two channels, the full-channel scanning mode is activated, which can realize multi-gas collaborative detection and suppress cross interference.

[0108] The technical features of the above-described embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described, but as long as the combinations of the technical features do not contradict, they should be considered as within the scope of the present disclosure.

[0109] The above-described embodiments only express several embodiments of the present disclosure, which are described in detail and specifically, but should not be construed as limiting the scope of the patent. It should be noted that for those skilled in the art, without departing from the concept of the present disclosure, some modifications and improvements can be made, which are within the scope of the present disclosure. Therefore, the scope of protection of the present patent should be subject to the appended claims.​

Claims

1. A multi-channel electrochemical detection electrode, comprising an electrode substrate and a plurality of working electrodes arranged on the electrode substrate, characterized in that: The surfaces of the multiple working electrodes are respectively modified with different types of metal nanocatalysts to detect different gas components. The surfaces of the working electrodes are covered with a PDMS hydrophobic layer to form a hydrophilic / hydrophobic channel separation structure.

2. A multi-channel electrochemical detection electrode according to claim 1, characterized in that: The electrode substrate is made of 3D printed porous graphene material, and the different types of metal nanocatalysts include Pt / Pd bimetallic catalysts and WO3 / ZnO heterojunction catalysts.

3. A multi-channel electrochemical sensor, characterized in that The invention comprises the multi-channel electrochemical detection electrode according to any one of claims 1 to 2.

4. A multi-channel electrochemical sensor according to claim 3, characterized in that: The invention also includes a concentration calculation unit, wherein the concentration calculation unit is used to: Obtain the cross-sensitivity correction factor based on the concentration information of the interfering gas and the cross-sensitivity coefficient; Obtain environmental interference factors based on environmental temperature and humidity information; And the initial molar concentration value of the target gas is obtained according to the cross-sensitivity correction factor, the environmental interference factor, the target gas basic information, the working electrode basic information and the detection response current of the working electrode.

5. A multi-channel electrochemical sensor detection method, applied to the multi-channel electrochemical sensor according to any one of claims 3-4, characterized in that: The steps include: For the target gas, obtaining a detection response current of a working electrode of a channel corresponding to the target gas; An initial molar concentration value of the target gas is obtained according to the detection response current.

6. A multi-channel electrochemical sensor detection method according to claim 5, characterized in that: The specific method of obtaining the initial molar concentration value of the target gas according to the detection response current includes the following steps: Obtain the cross-sensitivity correction factor based on the concentration information of the interfering gas and the cross-sensitivity coefficient; Obtain environmental interference factors based on environmental temperature and humidity information; And the initial molar concentration value of the target gas is obtained according to the cross-sensitivity correction factor, the environmental interference factor, the target gas basic information, the working electrode basic information and the detection response current.

7. A multi-channel electrochemical sensor detection method according to claim 6, characterized in that: The following steps are also included: Build a CNN-LSTM dual-channel network; The CNN branch performs convolution operations on the current signals of multi-channel working electrodes to extract spatial features; The current signal of the multi-channel working electrode is modeled through the LSTM branch to extract the timing features; The spatial features and the temporal features are fused to obtain the predicted concentration value of the k-th type of gas.

8. A multi-channel electrochemical sensor detection method according to claim 7, characterized in that: The specific method of building a CNN-LSTM dual-channel network includes the following steps: Construct loss function; A CNN-LSTM dual-channel network is constructed based on the loss function.

Images