A multi-channel olfactory gas analysis monitoring device for ancient building monitoring

Abstract

The application discloses a kind of multi-channel smell-sensitive gas analysis monitoring equipment for ancient building monitoring, including detection unit, control unit, data processing unit;Detection unit includes a plurality of independent smell-sensitive detection sensors, smell-sensitive detection sensor includes a detection channel and a signal acquisition module, signal acquisition module includes sine bias current operational amplifier and signal conversion module;Control unit includes central controller and temperature control module, central controller sends trigger signal to temperature control module and each signal acquisition module synchronously, and temperature control module provides independent heating power control for the heater of each detection channel;Data processing unit receives the detection signal obtained by each signal acquisition module, utilizes data fusion algorithm to analyze and process, and outputs gas analysis result.The application can effectively distinguish different components in mixed gas, significantly improve the identification ability of harmful gas to ancient building characteristics.

Description

Technical Field

[0001] This invention relates to the field of gas monitoring, and specifically to a multi-channel olfactory gas analysis and monitoring device for monitoring ancient buildings. Background Technology

[0002] The long-term preservation of ancient buildings directly depends on their microenvironment, among which the composition of ambient gases is a significant factor contributing to their deterioration. For example, sulfur dioxide in the air combines with moisture to form acid, which corrodes the exterior stone and metal artifacts; high concentrations of carbon dioxide indicate excessive human activity and may lead to secondary hazards such as temperature and humidity fluctuations. Continuous and precise monitoring of characteristic harmful gases in the environment of ancient buildings is a crucial step in implementing preventative protection.

[0003] At present, gas detection equipment, especially olfactory sensors based on metal oxide semiconductors, generally have the following problems: (1) a single sensor responds to a variety of gases, making it difficult to distinguish them; (2) a single-channel design means that the entire system will fail once the probe or circuit fails; (3) in order to achieve accurate analysis, it is often necessary to build a system composed of multiple instruments, which is complex and has high monitoring costs. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-channel olfactory gas analysis and monitoring device for monitoring ancient buildings, which can accurately identify the type and concentration of gases and improve the ability to identify harmful gases characteristic of ancient buildings.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A multi-channel olfactory gas analysis and monitoring device for monitoring ancient buildings includes a detection unit, a control unit, and a data processing unit, wherein:

[0007] The detection unit includes multiple independent olfactory sensors. Each olfactory sensor includes a detection channel and a signal acquisition module. The detection channel includes a gas chamber, an olfactory sensor probe, a heater, and a micro-differential pressure sensor. The gas chamber has an inlet and an outlet. The olfactory sensor probe is disposed within the gas chamber and is made of sintered semiconductor metal oxide material. The heater is disposed within the gas chamber to heat the gas to be tested flowing through it. The micro-differential pressure sensor is used to sense whether there is gas flow in the gas chamber. The signal acquisition module is electrically connected to the olfactory sensor probe and is used to acquire the change in electrical signal generated by the olfactory sensor probe after contacting the gas to be tested. The signal acquisition module includes a sinusoidal bias current operational amplifier and a signal conversion module. The sinusoidal bias current operational amplifier is used to provide a stable bias for the olfactory sensor probe and acquire its voltage signal. The signal conversion module is used to convert the analog voltage signal output by the sinusoidal bias current operational amplifier into a digital voltage signal, which is a voltage value that changes with time.

[0008] The control unit includes a central controller and a temperature control module. The central controller is connected to the micro differential pressure sensors of each detection channel. When the micro differential pressure sensor detects gas flowing into the gas chamber, it sends a sensing signal to the central controller. After receiving the sensing signal, the central controller synchronously sends trigger signals to the temperature control module and each signal acquisition module to start the heating and signal acquisition process of each detection channel. The temperature control module is connected to the heater of each detection channel and is used to provide independent heating power control for the heater of each detection channel, so that each detection channel is in a different temperature environment according to the detection needs. The temperature control module uses a power control method for temperature control.

[0009] The data processing unit is communicatively connected to the central controller. The detection signals acquired by the signal acquisition modules of each olfactory sensor are sent to the data processing unit through the central controller. The data processing unit uses a data fusion algorithm to analyze and process the data, and outputs the gas analysis results, which include the composition and concentration of the gas.

[0010] Each detection channel, under different temperature environments, detects the gas to be tested through the corresponding signal acquisition module to obtain detection signals. The data processing unit extracts features from the detection signals corresponding to each detection channel based on the response curve to obtain feature vectors. Then, the feature vectors corresponding to each detection channel are concatenated to obtain a high-dimensional fusion feature vector. The high-dimensional fusion feature vector is input into the gas analysis model to obtain analysis results. The feature vector includes at least the maximum response value, response slope, integral area, peak time, detection channel ID, and detection channel temperature label.

[0011] The gas analysis model is specifically represented as follows:

[0012]

[0013] in, The probability of the gas category. Weights for gas classification Bias for gas classification This is the predicted value of gas concentration. As the weight for concentration prediction, This is the bias for concentration prediction. To share feature vectors, To share the nonlinear transformation function of the backbone network, For high-dimensional fused feature vectors, To share parameters, For the first i Feature vectors of each detection channel N The total number of detection channels. For the first i The maximum response value of each detection channel For the first i The response slope of each detection channel, For the first i The integral area of ​​each detection channel. For the first i Peak time for each detection channel.

[0014] Preferably, it also includes a host computer, which is communicatively connected to the data processing unit and the control unit. The host computer includes a visual interface and a parameter setting module. The parameter setting module is used to provide users with custom parameter settings. The custom parameters include at least heating parameters and acquisition parameters. The heating parameters include at least heating power, heating rate, and heating start threshold.

[0015] Preferably, data fusion algorithms are used for analysis and processing, which is achieved through the following methods:

[0016] The digital voltage signals from each olfactory sensor are preprocessed to obtain the time series data to be analyzed.

[0017] Feature extraction is performed on the time series data to be analyzed to obtain feature vectors for each detection channel. The feature vectors are determined based on the response curves and include at least the maximum response value, response slope, integral area, peak time, detection channel ID, and detection channel temperature label.

[0018] The feature vectors of each detection channel are concatenated to obtain a high-dimensional fused feature vector;

[0019] The high-dimensional fused feature vector is input into the gas analysis model, and the gas analysis results are output.

[0020] Preferably, the gas analysis model is pre-trained using the following method:

[0021] Based on standard gases with known composition and concentration, various olfactory detection sensors are used to detect voltage response data at different channels and temperatures. Feature extraction is performed on the voltage response data of each detection channel, and then the data are spliced ​​together to obtain a high-dimensional fusion feature vector, forming a labeled sample set.

[0022] The labeled sample set is input into the model to be trained, and the nonlinear mapping relationship between the standard fusion feature vector and the gas category and concentration is learned.

[0023] Preferably, the detection channel introduces the gas to be tested using a natural aspiration method, and utilizes the cavity differential pressure principle to make the gas flow.

[0024] Preferably, the semiconductor metal oxide material is formed by sintering tin dioxide doped with palladium chloride.

[0025] Preferably, the central controller is configured to: determine that there is gas flow and generate the trigger signal when the pressure difference sensed and fed back by the micro differential pressure sensor continuously exceeds a first preset threshold for a first set time; determine that the gas flow stops when the pressure difference sensed and fed back by the micro differential pressure sensor continuously falls below a second preset threshold for a second set time, and control the temperature control module to turn off the heater.

[0026] Preferably, the central controller is configured to assign a unique time offset to each detection channel, and after receiving a trigger signal, each detection channel delays its corresponding time offset before starting heating to achieve time-sharing start-up.

[0027] Preferably, the temperature control module uses a high-precision source meter to achieve power control, with heating power control accuracy at the milliwatt level and temperature control accuracy reaching ±0.5℃; the signal conversion module uses an analog-to-digital converter with a resolution of not less than 16 bits and a sampling frequency of not less than 1Hz.

[0028] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art:

[0029] 1. This invention adopts a multi-channel independent olfactory detection design, with each channel independently set to work at different temperatures, simulating parallel sensing by "multiple nostrils". It utilizes the differences in response characteristics of different gases at different temperatures to actively stimulate gas characteristics and enhance the ability to identify the components of mixed gases.

[0030] 2. This invention constructs a "characteristic response pattern" or "chemical fingerprint" of a gas by combining the gas responses of multiple detection channels under different temperature environments. Then, through data fusion AI models, the gas type and concentration are accurately identified from the detected "chemical fingerprint." In other words, this invention achieves "fingerprint-level" high-precision identification of specific components in the gas mixture to be tested. This allows different gases that originally had similar response characteristics on a single channel to exhibit differentiated multi-dimensional response patterns under multiple temperature channels. This multi-dimensional response pattern constitutes a unique "chemical fingerprint" for each gas, fundamentally solving the problem of "poor selectivity and difficulty in distinguishing multiple gases" with a single sensor.

[0031] 3. The gas analysis model of the present invention does not directly use the original voltage curve, but extracts feature vectors (maximum response value, response slope, integral area, peak time) that can represent its dynamic characteristics. Then, the feature vectors from all detection channels are spliced ​​together to form a high-dimensional fused feature vector, which can comprehensively reflect the global behavior of the gas under all detection conditions.

[0032] 4. The multi-channel design of this invention is also a redundant design. If one sensor fails, the remaining sensors can still work, which improves the reliability and fault tolerance of the system in long-term unattended monitoring scenarios.

[0033] 5. This invention adopts a highly integrated design, using a single device to achieve multi-parameter measurement functions that traditionally require multiple devices, thus reducing system complexity, deployment difficulty, and overall monitoring costs.

[0034] 6. This invention uses a micro differential pressure sensor to intelligently sense gas flow, and only starts the heating and sampling process when gas flows in, which greatly reduces the power consumption of the device and improves the device's endurance in ancient building monitoring scenarios. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the system architecture of the present invention;

[0036] Figure 2 This is a schematic diagram of the structural framework of the acquisition and control section of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Example

[0038] refer to Figure 1 and Figure 2As shown, this invention discloses a multi-channel olfactory gas analysis and monitoring device for monitoring ancient buildings, comprising a detection unit, a control unit, a data processing unit, and a host computer, wherein:

[0039] The detection unit includes multiple independent olfactory sensors. Each olfactory sensor includes a detection channel and a signal acquisition module, wherein:

[0040] The detection channel includes a gas chamber, an olfactory sensor probe, a heater, and a micro-differential pressure sensor. The gas chamber is made of corrosion-resistant polytetrafluoroethylene (PTFE) to reduce gas adsorption and residue. The gas chamber is designed as a flat, elongated shape with an inlet and an outlet, utilizing the cavity differential pressure principle to achieve natural aspiration. The olfactory sensor probe is located within the gas chamber and is made of a sintered semiconductor metal oxide material; in this embodiment, the semiconductor metal oxide material is tin dioxide doped with palladium chloride. The heater is located within the gas chamber to heat the gas flowing through it. The micro-differential pressure sensor detects the presence of gas flow in the gas chamber; its two pressure detection ports are precisely connected to the vicinity of the gas chamber's inlet and outlet via thin-diameter conduits, respectively.

[0041] The signal acquisition module is electrically connected to the olfactory sensor probe and is used to acquire the changes in electrical signals generated by the olfactory sensor probe after contact with the gas to be tested. The signal acquisition module includes a sinusoidal bias current operational amplifier and a signal conversion module. The sinusoidal bias current operational amplifier is used to provide a stable bias for the olfactory sensor probe and acquire its voltage signal. The signal conversion module is used to convert the analog voltage signal output by the sinusoidal bias current operational amplifier into a digital voltage signal, which is a voltage value that changes over time. In this embodiment, the signal conversion module uses an analog-to-digital converter with a resolution of not less than 16 bits and a sampling frequency of not less than 1Hz.

[0042] The control unit includes a central controller and a temperature control module, wherein:

[0043] The central controller is connected to the micro-differential pressure sensors of each detection channel. When the micro-differential pressure sensor detects gas flowing into the gas chamber, it sends a sensing signal to the central controller. Upon receiving the sensing signal, the central controller synchronously sends trigger signals to the temperature control module and each signal acquisition module to initiate the heating and signal acquisition process of each detection channel. The central controller is configured as follows: when the pressure difference sensed and reported by the micro-differential pressure sensor continuously exceeds a first preset threshold for a first set time, it determines that there is gas flow and generates a trigger signal; when the pressure difference sensed and reported by the micro-differential pressure sensor continuously falls below a second preset threshold for a second set time, it determines that the gas flow has stopped and controls the temperature control module to shut down the heater; a unique time offset is assigned to each detection channel, and each detection channel, after receiving the trigger signal, delays its corresponding time offset before starting heating to achieve time-sharing start-up.

[0044] The temperature control module is connected to the heater of each detection channel to provide independent heating power control for each channel, ensuring that each channel operates at a different temperature according to the detection requirements. The temperature control module uses a power control method for temperature control. This allows for the dynamic allocation of different heating powers and temperature profiles to different channels based on the properties of the gas being tested and the detection needs, thereby proactively and optimally stimulating the characteristics of different gases, rather than simply passively controlling the temperature independently.

[0045] In this embodiment, the temperature control module uses a high-precision source meter to achieve power control, with heating power control accuracy at the milliwatt level and temperature control accuracy reaching ±0.5℃.

[0046] The data processing unit is communicatively connected to the central controller. The detection signals acquired by the signal acquisition modules of each olfactory sensor are sent to the data processing unit through the central controller. The data processing unit uses a data fusion algorithm to analyze and process the data, and outputs the gas analysis results, which include the composition and concentration of the gas.

[0047] The above-mentioned analysis and processing using data fusion algorithms is achieved through the following methods:

[0048] The digital voltage signals from each olfactory sensor are preprocessed to obtain the time-series data to be analyzed. The preprocessing mainly involves using a first-order low-pass digital filter to remove high-frequency noise and normalizing the data to eliminate baseline drift.

[0049] Feature extraction is performed on the time-series data to be analyzed to obtain feature vectors for each detection channel. The feature vectors are determined based on the response curve and include at least the maximum response value, response slope, integral area, time to peak, detection channel ID, and detection channel temperature label. The maximum response value is the maximum voltage value on the response curve, the response slope is the maximum first derivative of the rising segment of the response curve (reflecting the response speed), the integral area reflects the total response quantity, and the time to peak is the time required from the start of the response to reaching the peak value.

[0050] The feature vectors of each detection channel are concatenated to obtain a high-dimensional fused feature vector;

[0051] The high-dimensional fused feature vector is input into the gas analysis model, and the gas analysis results are output. The gas analysis model is specifically represented as follows:

[0052]

[0053] in, The probability of the gas category. Weights for gas classification Bias for gas classification This is the predicted value of gas concentration. As the weight for concentration prediction, This is the bias for concentration prediction. To share feature vectors, To share the nonlinear transformation function of the backbone network, For high-dimensional fused feature vectors, To share parameters, For the first i Feature vectors of each detection channel N The total number of detection channels. For the first i The maximum response value of each detection channel For the first i The response slope of each detection channel, For the first i The integral area of ​​each detection channel. For the first i Peak time for each detection channel.

[0054] The gas analysis model described above is pre-trained using the following method: based on standard gases with known composition and concentration, each olfactory sensor is used for detection to obtain voltage response data at different channels and temperatures. Features are extracted from the voltage response data of each detection channel, and then concatenated to obtain a high-dimensional fusion feature vector, forming a labeled sample set. The labeled sample set is then input into the model to be trained to learn the nonlinear mapping relationship between the standard fusion feature vector and the gas category and concentration.

[0055] The gas analysis model described above is actually a multi-task deep learning model, and its specific structure is as follows:

[0056] The shared backbone network consists of a three-layer fully connected neural network (128 -> 64 -> 32 neurons) and uses the ReLU activation function. Its function is to extract the shared feature vector Z from the high-dimensional fused feature vector.

[0057] The classification output head receives the shared feature vector Z and outputs the probability distribution of gas types through a softmax function. ;

[0058] The regression output head receives the same shared feature vector Z and outputs the predicted gas concentration through a linear transformation layer. .

[0059] The host computer is connected to the data processing unit and the control unit respectively. The host computer includes a visual interface and a parameter setting module. The parameter setting module is used to provide users with custom parameter settings. The custom parameters include at least heating parameters and acquisition parameters. The heating parameters include at least heating power, heating rate and heating start threshold. The heating start threshold includes the first preset threshold, the first set time, the second preset threshold and the second set time mentioned above.

[0060] The core idea of ​​this invention is to treat multiple independent detection channels as multiple "nostrils," and use an AI model to learn the "fingerprint" responses of different gases under different channels (representing different detection conditions), thereby making extremely accurate judgments. In practical use, multiple detection channels are configured to detect the same gas under different conditions in parallel (the same gas is activated at different temperatures and produces differentiated response signals) to obtain multidimensional response signals. Then, a gas analysis model is used to accurately identify the component type and concentration of the gas to be tested.

[0061] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A multi-channel olfactory gas analysis monitoring device for ancient building monitoring, characterized in that, The device comprises a detection unit, a control unit and a data processing unit, wherein: The detection unit comprises a plurality of independent olfactory detection sensors, each of which comprises a detection channel and a signal acquisition module. The detection channel comprises a gas chamber, an olfactory sensor probe, a heater and a micro differential pressure sensor. The gas chamber has an air inlet and an air outlet. The olfactory sensor probe is arranged in the gas chamber and is made of sintered semiconductor metal oxide material. The heater is arranged in the gas chamber and is used to heat the gas flowing through. The micro differential pressure sensor is used to sense whether the gas chamber has gas flow. The signal acquisition module is electrically connected to the olfactory sensor probe and is used to acquire the electrical signal change generated by the olfactory sensor probe after contacting the gas to be detected. The signal acquisition module comprises a sinusoidal bias current operational amplifier and a signal conversion module. The sinusoidal bias current operational amplifier is used to provide stable bias for the olfactory sensor probe and acquire the voltage signal thereof. The signal conversion module is used to convert the analog voltage signal output by the sinusoidal bias current operational amplifier into a digital voltage signal. The digital voltage signal is a voltage value that changes over time. The control unit comprises a central controller and a temperature control module. The central controller is connected to the micro differential pressure sensor of each detection channel. When the micro differential pressure sensor senses that the gas chamber has gas flowing in, it sends a sensing signal to the central controller. After receiving the sensing signal, the central controller sends a trigger signal to the temperature control module and each signal acquisition module synchronously to start the heating and signal acquisition process of each detection channel. The temperature control module is connected to the heater of each detection channel and is used to provide independent heating power control for the heater of each detection channel so that each detection channel can be in different temperature environments according to the detection needs. The temperature control module uses power control method for temperature control. The data processing unit is in communication connection with the central controller. The detection signals acquired by the signal acquisition modules of the olfactory detection sensors are sent to the data processing unit through the central controller. The data processing unit analyzes and processes the detection signals using a data fusion algorithm and outputs a gas analysis result. The gas analysis result includes the composition and concentration of the gas. Each detection channel detects the gas to be detected under different temperature environments through the corresponding signal acquisition module to obtain detection signals. The data processing unit extracts features from the detection signals of each detection channel based on a response curve to obtain a feature vector. Then, the data processing unit splices the feature vectors of each detection channel to obtain a high-dimensional fusion feature vector. The high-dimensional fusion feature vector is input into a gas analysis model to obtain an analysis result. The feature vector at least includes a maximum response value, a response slope, an integral area, a peak value reaching time, a detection channel ID and a detection channel temperature label. The gas analysis model is specifically represented as: wherein, is a probability of gas category, is a weight of gas category, is a bias of gas category, is a gas concentration prediction value, is a weight of concentration prediction, is a bias of concentration prediction, is a shared feature vector, is a nonlinear transformation function of shared backbone network, is a high-dimensional fusion feature vector, is a shared parameter, is a feature vector of the i th detection channel, N is a total number of detection channels, is a maximum response value of the i th detection channel, is a response slope of the i th detection channel, is an integral area of the i th detection channel, is a peak reaching time of the i th detection channel.

2. The multi-channel olfactory gas analysis monitoring device for ancient building monitoring according to claim 1, characterized in that: The application also comprises a host computer, which is in communication connection with the data processing unit and the control unit respectively, and comprises a visual interface and a parameter setting module, the parameter setting module being used to provide the user with the setting of self-defined parameters, the self-defined parameters at least including heating parameters and acquisition parameters, the heating parameters at least including heating power, temperature rising rate and heating start threshold.

3. The multi-channel olfactory gas analysis monitoring device for monitoring ancient buildings according to claim 1 or 2, characterized in that, The data fusion algorithm is used for analysis and processing, and the following methods are used to realize the same: The digital voltage signals from each olfactory detection sensor are preprocessed to obtain time series data for analysis; Feature extraction is performed on the time series data for analysis to obtain feature vectors of each detection channel, the feature vectors being determined based on response curves and at least including maximum response value, response slope, integral area, peak value reaching time, detection channel ID and detection channel temperature label; The feature vectors of each detection channel are spliced to obtain a high-dimensional fusion feature vector; The high-dimensional fusion feature vector is input into a pre-trained gas analysis model to output a gas analysis result.

4. The multi-channel olfactory gas analysis monitoring device for monitoring ancient buildings according to claim 3, wherein, The gas analysis model is pre-trained by the following methods: Based on standard gases with known components and concentrations, each olfactory detection sensor is used for detection to obtain voltage response data at different channels and different temperatures, feature extraction is performed on the voltage response data of each detection channel, and then high-dimensional fusion feature vectors are spliced to form a labeled sample set; The labeled sample set is input into a to-be-trained model to learn the nonlinear mapping relationship between the standard fusion feature vector and the gas category and concentration.

5. A multi-channel olfactory gas analysis monitoring device for monitoring ancient buildings according to claim 4, characterized in that: The detection channel uses a natural air suction method to introduce the gas to be detected, and uses the cavity differential pressure principle to make the gas flow.

6. The multi-channel olfactory gas analysis monitoring device for monitoring ancient buildings according to claim 4, characterized in that: The semiconductor metal oxide material is sintered from tin dioxide doped with palladium chloride.

7. A multi-channel odorant gas analysis monitoring device for monitoring ancient buildings according to claim 4, characterized in that, The central controller is configured to: when the pressure difference perceived and fed back by the micro pressure difference sensor continuously exceeds the first preset threshold for a first set time, it is determined that there is gas flow and the trigger signal is generated; when the pressure difference perceived and fed back by the micro pressure difference sensor continuously falls below the second preset threshold for a second set time, it is determined that the gas flow stops, and the temperature control module controls the heater to be turned off.

8. A multi-channel olfactory gas analysis monitoring device for monitoring ancient buildings according to claim 7, characterized in that, The central controller is configured to: assign a unique time offset to each detection channel, and after receiving the trigger signal, each detection channel delays the corresponding time offset and then starts heating to realize time-sharing start.

9. The multi-channel olfactory gas analysis monitoring device for monitoring ancient buildings according to claim 1, characterized in that: The temperature control module uses a high-precision source table to realize power control, the heating power control accuracy is milliwatt level, and the temperature control accuracy reaches ±0.5℃; the signal conversion module uses an analog-to-digital converter, the resolution of which is not less than 16 bits, and the sampling frequency is not less than 1 Hz.

Images