Gas concentration detection method, device, equipment and storage medium

Abstract

The invention discloses a gas concentration detection method, device and equipment and a storage medium. The method comprises the following steps: acquiring sensing data points generated by two sensors which are based on a spectral absorption principle and have higher detection precision and higher response speed at the same time to detect the concentration of target gas in real time; then converting each sensing data point into a concentration data point with a feature item comprising a timestamp by utilizing a pre-trained BP neural network and combining with a Lambert-Beer law; determining available measurement data of the preset Kalman filter at the current time step from all the concentration data points according to each timestamp; and finally, acquiring a concentration predicted value of the target gas at the current time step by using a preset Kalman filter, and updating the state of the preset Kalman filter according to the available measurement data and the current concentration predicted value, so as to obtain a concentration estimated value of the target gas at the current time step. Therefore, both high detection precision and high response speed are realized.

Description

Technical Field

[0001] The present application relates to the field of gas detection technology, and in particular to a gas concentration detection method, device, equipment and storage medium. Background Art

[0002] Gas detection technology based on the principle of spectral absorption is an important technical branch in the field of gas detection. It mainly uses the absorption characteristics of gas molecules to light of a specific wavelength to detect the type and concentration of gas. For gas sensors that apply the principle of spectral absorption, response speed and detection accuracy are two important indicators to measure their performance.

[0003] Most of the gas detection instruments based on the spectral absorption principle currently on the market use a single gas sensor for detection. In theory, a single gas sensor can also have both a fast response speed and high detection accuracy. However, in practical applications, depending on the specific application scenarios and optimization goals, it is usually necessary to weigh and make targeted choices between response speed and detection accuracy in the design and optimization of the sensor. As a result, the use of a single sensor for detection often results in the following situations: those with a fast response speed generally have lower detection accuracy, while those with higher detection accuracy generally have a slower response speed. This makes it difficult for existing gas detection instruments based on the spectral absorption principle to simultaneously meet the requirements of fast response and high detection accuracy, limiting their scope of application and performance improvement. Summary of the invention

[0004] The embodiments of the present invention provide a gas concentration detection device and method to solve the problem that the existing gas detection instrument based on the spectrum absorption principle is difficult to meet the requirements of fast response and high detection accuracy at the same time.

[0005] In a first aspect, an embodiment of the present invention provides a method for detecting gas concentration, the method comprising:

[0006] When the first sensor and the second sensor simultaneously start to detect the concentration of the target gas contained in the gas to be detected and at least one of the two generates a sensing data point, the sensing data point is acquired in real time;

[0007] Using a pre-trained back propagation BP neural network in combination with Beer-Lambert's law, each of the sensor data points is converted into a concentration data point, and the characteristic item of each of the concentration data points includes a timestamp;

[0008] Determining available measurement data of a preset Kalman filter at a current time step from all the concentration data points according to each of the timestamps;

[0009] Using the preset Kalman filter to obtain a predicted concentration value of the target gas at the current time step, and updating the state of the preset Kalman filter according to the available measurement data and the current predicted concentration value to obtain an estimated concentration value of the target gas at the current time step;

[0010] The first sensor and the second sensor are both gas sensors based on the spectral absorption principle, the detection accuracy of the first sensor is higher than that of the second sensor, and the response speed of the first sensor is lower than that of the second sensor.

[0011] In a second aspect, an embodiment of the present invention further provides a gas concentration detection device, the device comprising:

[0012] A data acquisition module, used for acquiring the sensing data point in real time when the first sensor and the second sensor simultaneously start to detect the concentration of the target gas contained in the gas to be detected and at least one of the two generates a sensing data point;

[0013] A data processing module, used for converting each of the sensing data points into a concentration data point by using a pre-trained back propagation BP neural network in combination with Lambert-Beer's law, wherein the characteristic item of each of the concentration data points includes a timestamp;

[0014] A data determination module, used to determine available measurement data of a preset Kalman filter at a current time step from all the concentration data points according to each of the timestamps;

[0015] a concentration estimation module, configured to obtain a concentration prediction value of the target gas at the current time step by using the preset Kalman filter, and to update the state of the preset Kalman filter according to the available measurement data and the current concentration prediction value, so as to obtain a concentration estimation value of the target gas at the current time step;

[0016] The first sensor and the second sensor are both gas sensors based on the spectral absorption principle, the detection accuracy of the first sensor is higher than that of the second sensor, and the response speed of the first sensor is lower than that of the second sensor.

[0017] In a third aspect, an embodiment of the present invention further provides a gas concentration detection device, the device comprising:

[0018] a first sensor, a second sensor, a memory, and one or more processors;

[0019] Wherein, the one or more processors are respectively connected to the first sensor, the second sensor and the memory;

[0020] The first sensor and the second sensor are used to simultaneously start detecting the concentration of the target gas contained in the gas to be tested and generate sensing data points;

[0021] The memory is used to store one or more programs;

[0022] The one or more programs are executed by the one or more processors to implement the gas concentration detection method as described in the first aspect of the embodiment of the present invention.

[0023] In a fourth aspect, an embodiment of the present invention further provides a storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the gas concentration detection method as described in the first aspect of the embodiment of the present invention is implemented.

[0024] The embodiment of the present invention firstly acquires in real time the first sensor and the second sensor, both of which are based on the principle of spectral absorption, but one has higher detection accuracy and the other has faster response speed, and simultaneously detects the concentration of the target gas contained in the gas to be tested to generate sensor data points; then, the pre-trained BP neural network is combined with the Lambert-Beer law to convert each sensor data point into a concentration data point whose characteristic item includes a timestamp; then, according to each timestamp, the available measurement data of the preset Kalman filter at the current time step is determined from all the concentration data points; finally, the preset Kalman filter is used to obtain the concentration prediction value of the target gas at the current time step, and the state of the preset Kalman filter is updated according to the available measurement data and the current concentration prediction value, so as to obtain the concentration estimation value of the target gas at the current time step. Thus, the embodiment of the present invention realizes the data fusion processing of the first sensor and the second sensor, so that when detecting the concentration of the target gas, the detection accuracy achieved is higher than the detection accuracy of the second sensor, and the response speed achieved is faster than the response speed of the first sensor. Compared with the gas concentration detection scheme using only the first sensor and only the second sensor, the embodiment of the present invention achieves the balance between high detection accuracy and fast response speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flow chart of a gas concentration detection method provided in Embodiment 1 of the present invention;

[0026] Figure 2 It is a flow chart of a gas concentration detection method provided in the second embodiment of the present invention;

[0027] Figure 3 is a structural schematic diagram of a gas concentration detection device provided in Embodiment 3 of the present invention;

[0028] Figure 4 is a structural schematic diagram of a gas concentration detection device provided by Embodiment 4 of the present invention;

[0029] Figure 5 This is a structural example diagram of a gas concentration detection device provided in Example 4 of the present invention. DETAILED DESCRIPTION

[0030] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. In addition, the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict. It is also necessary to explain that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all structures.

[0031] Embodiment 1

[0032] Figure 1 This is a flow chart of a gas concentration detection method provided by an embodiment of the present invention. This embodiment can be applied to the sensor data points of two gas sensors based on the spectral absorption principle, and the BP neural network is used to extract the characteristic values ​​and the Kalman filter is used to perform data fusion processing to achieve high detection accuracy and fast response speed. The method can be performed by a gas concentration detection device, which can be implemented by hardware and / or software and is generally integrated on the gas concentration detection equipment.

[0033] It is understandable that for a gas sensor based on the principle of spectral absorption, when light of a specific wavelength passes through a gas chamber containing gas molecules, the gas molecules absorb the light of the specific wavelength, resulting in attenuation of the light intensity. According to the Lambert-Beer law, this attenuation is proportional to the concentration of the gas. Therefore, by obtaining the change in the intensity of the transmitted light, the concentration of the gas can be calculated.

[0034] Most of the existing gas detection instruments based on the spectral absorption principle use a single gas sensor, and optimize the structure, material and / or compensation algorithm of the single sensor for specific application scenarios to achieve fast response speed or high detection accuracy. However, this method often loses one thing while focusing on the other, and fails to achieve a good balance between the two.

[0035] To this end, an embodiment of the present invention proposes a gas concentration detection scheme, which uses two gas sensors that are both based on the spectral absorption principle but one has higher detection accuracy and the other has faster response speed, and simultaneously detects the concentration of the target gas contained in the gas to be tested, and obtains the sensor data points generated by the two gas sensors in real time; then, the nonlinear mapping and fast feature extraction characteristics of the BP neural network are used to quickly extract the characteristic values ​​of the generated sensor data points to obtain data points that mainly contain the desired characteristic values, thereby greatly simplifying the amount of data to be processed subsequently; and then, Lambert-Beer's law is applied to convert them into data points containing the detected concentration values; finally, the real-time estimation and data fusion characteristics of the Kalman filter are used to perform real-time data fusion processing on the sensor data points of the two gas sensors to obtain the real-time concentration estimation value of the target gas. In this way, the complementary characteristics of the two gas sensors can be combined to achieve a balance between high detection accuracy and fast response speed.

[0036] It should be noted that the gas sensor based on the spectral absorption principle mainly uses the absorption characteristics of gas molecules to light of a specific wavelength to detect the type and concentration of the gas. Its main technical classification includes: non-dispersive infrared (NDIR, Non-Dispersive InfraRed) gas detection technology, tunable laser diode laser absorption spectroscopy (TDLAS, Tunablediode laser absorption spectroscopy) gas detection technology, photoacoustic spectroscopy gas detection technology, Fourier transform infrared (FTIR, Fourier Transform Infrared) gas detection technology, etc. The two gas sensors used in the embodiment of the present invention are not limited to belonging to the same technical classification as mentioned above, as long as both belong to the gas sensor based on the spectral absorption principle.

[0037] In addition, there are also multi-sensor fusion solutions in the prior art, but most of them are to achieve multi-dimensional data fusion of different types of sensors (for example, in the field of autonomous driving technology, the detection data of the same target by the vehicle-mounted ultrasonic radar, laser radar, camera, etc. are fused), while the solution of the embodiment of the present invention is to fuse two gas sensors that are both based on the principle of spectral absorption, and the solution of the embodiment of the present invention focuses on utilizing the complementarity of the two sensors in terms of response speed and detection accuracy. Furthermore, using two sensors of the same type (both based on the principle of spectral absorption for gas concentration detection) can provide redundant information. Therefore, by fusing the measurement results of the two sensors, the measurement error and noise influence of a single sensor can be reduced, and the reliability of the measurement results can be improved. Even if the measurement results of the two sensors are not completely consistent, the fused data can provide more stable and consistent results.

[0038] In addition, in the embodiments of the present invention, the response speed and detection accuracy can be understood as the system response speed and system detection accuracy of the detection device or equipment integrated with the sensor, and the effect achieved by the embodiments of the present invention is also compared from the system level. The response speed can be characterized by the response time. For the sensor, its response time can be understood as the time from the start of detection to the output of the original data (such as voltage or current signal); and for the detection device or equipment, the response time can be understood as the time from the start of detection to the output of the indicated value (such as gas concentration). For the detection accuracy, it can be understood as a comprehensive indicator including but not limited to accuracy (used to characterize the closeness of the measurement result to the true value), precision (used to characterize the repeatability and consistency of the measurement result), resolution (used to characterize the minimum change that the system can detect) and sensitivity (used to characterize the system's ability to respond to changes in the input signal).

[0039] like Figure 1 As shown, the gas concentration detection method provided by the embodiment of the present invention includes the following steps:

[0040] S100: When the first sensor and the second sensor simultaneously start to detect the concentration of the target gas contained in the gas to be detected and at least one of the first sensor and the second sensor generates a sensing data point, the sensing data point is acquired in real time.

[0041] The first sensor and the second sensor are both gas sensors based on the spectral absorption principle, the detection accuracy of the first sensor is higher than that of the second sensor, and the response speed of the first sensor is lower than that of the second sensor.

[0042] The sensor data point refers to the output data of the sensor, which can be the raw data of the sensor or the structured output data containing multiple features obtained by processing the raw data. Taking the NDIR sensor as an example, its output data can include but is not limited to the voltage signal or current signal used to characterize the magnitude of the incident light intensity and the outgoing light intensity, sensor status information, uplink time, downlink time, peak time, peak intensity, and noise data.

[0043] Exemplarily, the structured form of the sensor data point can be expressed as:

[0044] Sensor data point = (first feature, second feature, third feature, ...)

[0045] It is understandable that the solution of the embodiment of the present invention is to fuse the data of the first sensor and the second sensor through Kalman filtering, so as to utilize the complementarity of the two sensors in terms of detection accuracy and response speed, and achieve a balance between high detection accuracy and fast response speed. Therefore, in order to ensure the effectiveness of data fusion, the consistency of the detection objects of the two sensors and the consistency of the detection start time should be ensured.

[0046] In step S100, since the response speeds of the two sensors are different, in order to ensure the timeliness of data acquisition, when any sensor generates a sensing data point, the generated sensing data point is acquired immediately.

[0047] S110, using a pre-trained back propagation BP neural network in combination with Beer-Lambert's law, converting each of the sensing data points into a concentration data point, wherein the characteristic item of each of the concentration data points includes a timestamp.

[0048] It is understandable that BP neural network, as a classic multi-layer feedforward neural network, has powerful feature extraction and nonlinear mapping capabilities. The feature extraction characteristics of BP neural network are mainly reflected in its ability to automatically extract features that have an important impact on the target variable from the input data without the need to manually design feature extraction methods. The nonlinear mapping characteristics of BP neural network are mainly reflected in its ability to learn the complex nonlinear relationship between input data and output data.

[0049] In the embodiment of the present invention, by means of the feature extraction characteristics and nonlinear mapping characteristics of the BP neural network, it is possible to quickly extract the desired feature values, such as timestamp, incident light intensity, outgoing light intensity, etc., from the original sensor data, thereby achieving dimensionality reduction of the original sensor data, thereby greatly reducing the complexity and order of magnitude of subsequent data processing. And by means of the nonlinear mapping characteristics of the BP neural network, the original characteristics of the sensor can be reflected in the data after feature extraction, for example, the high detection accuracy characteristics of the first sensor can be reflected in the feature items after feature extraction of the sensor data points of the first sensor, and the fast response speed characteristics of the second sensor can be reflected in the feature items after feature extraction of the sensor data points of the second sensor.

[0050] It should be noted that, although the BP neural network may require more computing resources and time during the training phase, once the training is completed and deployed in the actual system, it can quickly process data in a real-time environment. The parallel processing capability of the BP neural network enables it to quickly perform forward propagation calculations, realize real-time processing of data, and thus quickly extract eigenvalues. In the embodiments of the present invention, the BP neural network is mainly applied, and its training process is not described in detail here. At the same time, the embodiments of the present invention do not limit the training method and process of the BP neural network.

[0051] The sensors used in the embodiments of the present invention are all sensors based on the principle of spectral absorption, and the Lambert-Beer law describes the relationship between the degree of absorption of light when it passes through a gas and the gas concentration. The absorption rate (absorbance) can be obtained by the incident light intensity and the outgoing light intensity. Therefore, after the BP neural network is used to extract the features of the sensor data points, the Lambert-Beer law can be applied to obtain the corresponding concentration data according to the absorption rate.

[0052] For the timestamp, the sensor can mark the time of data collection in the output data (i.e., the sensor data point) as the timestamp, or the time point when the sensor data point is input into the BP neural network can be marked as the timestamp through step S110. It can be understood that the role of the timestamp is to provide a unified time reference, so as long as the time reference selected for all sensor data points is consistent.

[0053] S120: Determine available measurement data of a preset Kalman filter at a current time step from all the concentration data points according to each of the timestamps.

[0054] According to the working principle of the Kalman filter, the measurement value at the current moment is needed when updating the state, and the purpose of step S120 is to obtain the measurement value at the current moment (the available measurement data at the current time step can be understood as the measurement value at the current moment) to pave the way for step S130.

[0055] The time step can be understood as the basic time unit used to describe the system state update. Each time step corresponds to an update of the system state and the processing of the measurement data. The length of the time step can be fixed or dynamically changing, depending on the application scenario and system design. In the Kalman filter, each time step corresponds to a prediction and update operation. In a multi-sensor system, the response time of different sensors may be different, and the time step can be used to align these data to ensure that all sensor data are fused and processed on the same time basis.

[0056] Optionally, the step size of the time step is set not to exceed the response time of the second sensor.

[0057] The response time of the sensor in the embodiment of the present invention can be understood as the time from sampling to generating a sensing data point by the sensor.

[0058] It is understandable that, since the second sensor has a faster response speed (ie, a shorter response time), by setting the time step size not to exceed the response time of the second sensor, each sensor data point can be fully utilized to update the system status in a timely manner.

[0059] In one embodiment, the length of the time step can be dynamically adjusted to adapt to changes in system state or sensor sampling rate. For example, when the system detects a rapid change in gas concentration, the length of the time step can be shortened to increase the system response speed.

[0060] S130. Utilize the preset Kalman filter to obtain the predicted concentration value of the target gas at the current time step, and update the state of the preset Kalman filter according to the available measurement data and the current predicted concentration value to obtain the estimated concentration value of the target gas at the current time step.

[0061] It can be understood that the working principle of Kalman filtering mainly includes the following two steps:

[0062] 1) Prediction step: predict the state at the current moment based on the system's dynamic model (state transfer equation) and the state estimate at the previous moment; predict the error covariance at the current moment.

[0063] 2) Update step: Calculate the Kalman gain based on the measured value and predicted value at the current moment; use the Kalman gain to update the state estimate and error covariance.

[0064] It should be noted that the Kalman filter does not require pre-training, but requires parameter setting and initialization. This is because the core of the Kalman filter is the system model (state transfer equation and observation equation) and noise characteristics (covariance matrix of process noise and measurement noise), and the system model and noise characteristics need to be parameterized and initialized before using the Kalman filter. The embodiment of the present invention focuses on the application of the Kalman filter, so the specific method of parameter setting and initialization of the Kalman filter is not limited.

[0065] The embodiment of the present invention first obtains the sensor data points generated by the first sensor and the second sensor starting to detect the concentration of the target gas at the same time in real time; then uses the pre-trained BP neural network combined with the Lambert-Beer law to convert each sensor data point into a concentration data point whose characteristic item includes a timestamp; then determines the available measurement data of the preset Kalman filter at the current time step from all the concentration data points according to each timestamp; finally, uses the preset Kalman filter to obtain the concentration prediction value of the target gas at the current time step, and updates the state of the preset Kalman filter according to the available measurement data and the current concentration prediction value, so as to obtain the concentration estimation value of the target gas at the current time step. Thus, the embodiment of the present invention realizes the data fusion processing of the first sensor and the second sensor, so that when detecting the concentration of the target gas, the detection accuracy achieved is higher than the detection accuracy of the second sensor, and the response speed achieved is faster than the response speed of the first sensor. Compared with the gas concentration detection scheme using only the first sensor and only the second sensor, the embodiment of the present invention achieves the balance between high detection accuracy and fast response speed.

[0066] Furthermore, based on the above embodiment, the first sensor and the second sensor may be further optimized:

[0067] The first sensor includes a first air chamber, through which the first sensor collects the gas to be measured; the second sensor includes a second air chamber, through which the second sensor collects the gas to be measured; the optical path length of the first air chamber is longer than the optical path length of the second air chamber.

[0068] It can be understood that both the first sensor and the second sensor are gas sensors based on the principle of spectral absorption, and therefore both are provided with gas chambers. By setting the optical path length of the first gas chamber of the first sensor to be longer than the optical path length of the second gas chamber of the second sensor, the absorption path of the gas molecules in the first gas chamber is longer, and the diffusion path of the gas molecules in the second gas chamber is shorter, so that the first sensor can detect more fully and can detect a target gas with a lower concentration, and the second sensor can detect changes in gas concentration more quickly.

[0069] Through this optimized embodiment, the high detection accuracy characteristics of the first sensor and the fast response speed characteristics of the second sensor are further enhanced, thereby further optimizing the overall detection accuracy and response speed of the system.

[0070] Embodiment 2

[0071] Figure 2It is a flow chart of a gas concentration detection method provided in the second embodiment of the present invention. This embodiment is further optimized on the basis of the first embodiment. In this embodiment, the pre-trained back propagation BP neural network is combined with the Lambert-Beer law to convert each of the sensor data points into a concentration data point, and the feature item of each of the concentration data points includes a timestamp. Specifically, when the sensor data points include a first sensor data point generated by the first sensor, the BP neural network is used to extract feature values ​​of the first sensor data point to obtain a first feature data point; wherein the feature items of the first feature data point include: a first timestamp, a first incident light intensity, and a first exit light intensity; according to the first incident light intensity and the first exit light intensity, combined with the Lambert-Beer law, the first feature data point is converted into a first concentration data point; when the sensor data points include a second sensor data point generated by the second sensor, the BP neural network is used to extract feature values ​​of the second sensor data point to obtain a second feature data point; wherein the feature items of the second feature data point include: a second timestamp, a second incident light intensity, and a second exit light intensity; according to the second incident light intensity and the second exit light intensity, combined with the Lambert-Beer law, the second feature data point is converted into a second concentration data point.

[0072] Furthermore, this embodiment concretizes the method of determining the available measurement data of the preset Kalman filter at the current time step from all the concentration data points according to each of the timestamps as follows: according to each of the timestamps, select the concentration data point whose time marked by the timestamp belongs to the current time step from all the concentration data points as the available concentration data point; when the available concentration data points simultaneously include concentration data points converted from the sensor data points generated by the first sensor and the second sensor, time-align all the available concentration data points, and determine the available concentration data points after time alignment as the available measurement data.

[0073] like Figure 2 As shown, the gas concentration detection method provided in this embodiment includes the following steps:

[0074] S200: When the first sensor and the second sensor simultaneously start to detect the concentration of the target gas contained in the gas to be detected and at least one of the first sensor and the second sensor generates a sensing data point, the sensing data point is acquired in real time.

[0075] S210. When the sensor data points include the first sensor data point generated by the first sensor, use the BP neural network to extract feature values ​​of the first sensor data point to obtain a first feature data point; wherein the feature items of the first feature data point include: a first timestamp, a first incident light intensity, and a first outgoing light intensity.

[0076] It can be understood that the sensor data point itself contains multiple feature items, and feature extraction through BP neural network is to obtain the desired feature items, such as timestamp, incident light intensity and outgoing light intensity.

[0077] S220 . According to the first incident light intensity and the first exiting light intensity, in combination with Beer-Lambert's law, convert the first characteristic data point into a first concentration data point.

[0078] In one embodiment, step S220 may include the following steps S2201 to S2203:

[0079] S2201. Convert the first characteristic data point into a first absorption rate data point according to the first incident light intensity and the first exiting light intensity.

[0080] The absorption rate calculation formula can be expressed as: Among them, A is the absorptivity, I0 is the incident light intensity, and I is the outgoing light intensity.

[0081] According to the absorbance calculation formula, the absorbance can be calculated when the incident light intensity and the exit light intensity are known. Therefore, the purpose of step S2201 is to calculate the first absorbance from the first incident light intensity and the first exit light intensity according to the absorbance calculation formula, and to obtain the first absorbance data point by replacing the first incident light intensity and the first exit light intensity in the feature item of the first feature data point with the first absorbance.

[0082] S2202: Normalize the first absorption rate data point to obtain a first normalized data point.

[0083] It is understandable that no matter it is a sensor data point, a characteristic data point, an absorbance data point or a concentration data point, it contains multiple characteristic items, and the dimensions of different characteristic items may be different, and the order of magnitude may also be quite different. Through normalization processing, data of different magnitudes can be scaled to the same range (for example, 0 to 1), and the dimension difference can be eliminated, so that different characteristic items have the same scale. In addition, normalization can also reduce the impact of noise in the data points on the system, thereby improving the measurement accuracy of the system.

[0084] For the subsequent Kalman filtering steps, the initial normalization process can achieve the following beneficial effects:

[0085] 1) Improve the numerical stability of the Kalman filter and avoid numerical errors;

[0086] 2) Improve the accuracy of state estimation and ensure that the data from each sensor contributes evenly to the state estimation;

[0087] 3) Ensure the consistency of multi-sensor data fusion and eliminate dimensional differences;

[0088] 4) Improve the computational efficiency of Kalman filtering and avoid unnecessary calculations;

[0089] 5) Simplify the adjustment process of Kalman filter parameters and improve model performance.

[0090] S2203. Apply Beer-Lambert's law to convert the first normalized data point into the first concentration data point, wherein the characteristic item of the first concentration data point includes the first timestamp and a first converted concentration value.

[0091] Lambert-Beer's law can be expressed as: Among them, A is the absorptivity, I0 is the incident light intensity, I is the outgoing light intensity, c is the gas concentration, ε is the molar absorption coefficient, and L is the optical path length.

[0092] According to Lambert-Beer law, we can get: That is, when the first absorbance, the molar absorption coefficient ε, and the optical path length are known, the first concentration value can be inverted. Therefore, the purpose of step S2203 is to invert the first concentration value by applying the Lambert-Beer law from the first absorbance according to the Lambert-Beer law, and to obtain the first concentration data point by replacing the first absorbance in the characteristic item of the first absorbance data point with the first concentration value.

[0093] Optionally, a normalization operation may also be performed after obtaining the concentration data points.

[0094] Optionally, before converting the first characteristic data point into the first concentration data point, low-pass filtering is performed on the first characteristic data point.

[0095] It is understandable that the eigenvalues ​​extracted by the BP neural network are usually some combination or transformation of the original data. Although the purpose of feature extraction is to extract useful features, the extracted eigenvalues ​​may still contain noise, which may come from sensor noise, data processing errors, etc. Therefore, by low-pass filtering the eigenvalues ​​extracted by the BP neural network, it can smooth the signal, remove high-frequency noise, and improve the stability of the eigenvalues, thereby further improving the detection accuracy of the system.

[0096] S230. When the sensor data points include the second sensor data points generated by the second sensor, the BP neural network is used to extract feature values ​​of the second sensor data points to obtain second feature data points; wherein the feature items of the second feature data points include: a second timestamp, a second incident light intensity, and a second outgoing light intensity.

[0097] S240: According to the second incident light intensity and the second exit light intensity, in combination with Beer-Lambert's law, convert the second characteristic data point into a second concentration data point.

[0098] In one embodiment, step S240 may include the following steps S2401 to S2403:

[0099] S2401. Convert the second characteristic data point into a second absorption rate data point according to the second incident light intensity and the second exit light intensity.

[0100] S2402: Normalize the second absorption rate data point to obtain a second normalized data point.

[0101] S2403: Apply Beer-Lambert's law to convert the second normalized data point into the second concentration data point, wherein the characteristic item of the second concentration data point includes the second timestamp and the second converted concentration value.

[0102] The implementation process, purpose and effect of steps S2401 to S2403 may refer to the above steps S2201 to S2203 and will not be repeated here.

[0103] In addition, since the response speeds of the first sensor and the second sensor are different, the time when the two generate sensor data points is inconsistent for detecting the gas to be tested at the same time. Therefore, the sensor data points acquired in real time may only include sensor data points generated by one of the two, or may include sensor data points generated by both. Then, when the sensor data points generated by both are included at the same time, steps S210 to S240 can be executed in parallel or in series. In order to improve the response speed of the system, parallel execution can be used to shorten the processing time. For parallel processing, the BP neural network can realize parallel input (multiple input nodes), parallel processing (parallel processing within a layer), and parallel output (multiple output nodes).

[0104] In one embodiment, after converting the second normalized data point into the second concentration data point, it further includes: correcting the second converted concentration value according to a preset correction algorithm, and updating the second concentration data point according to the corrected second converted concentration value.

[0105] It is understandable that although the second sensor has a fast response speed, it is not as good as the first sensor in terms of stability and detection accuracy. Therefore, specific measures need to be taken to correct its measurement results before performing the Kalman filter fusion operation.

[0106] In one embodiment, the preset correction algorithm includes but is not limited to: temperature compensation, humidity compensation, cross-sensitivity compensation, drift correction, denoising and filtering, outlier detection and missing value filling, etc.

[0107] By correcting the second conversion concentration data of the second sensor, at least the following effects can be achieved:

[0108] 1) Improve measurement accuracy: compensate for the impact of sensor errors, environmental factors and data quality on measurement results;

[0109] 2) Improve data quality: remove noise, drift, outliers and missing data to improve data integrity and reliability;

[0110] 3) Improve system robustness: Improve the system's adaptability to environmental changes and sensor failures;

[0111] 4) Support data fusion: Improve the consistency and comparability of data and better support the data fusion of two sensors;

[0112] 5) Improve Kalman filter performance: Provide high-quality data input to improve the state estimation accuracy of the Kalman filter.

[0113] S250 , according to each of the timestamps, selecting from all the concentration data points the concentration data point whose time marked by the timestamp belongs to the current time step as an available concentration data point.

[0114] S260. When the available concentration data points include concentration data points converted from the sensing data points generated by the first sensor and the second sensor, time-align all the available concentration data points, and determine the time-aligned available concentration data points as the available measurement data.

[0115] It is understood that the Kalman filter is a recursive algorithm for estimating the state of a dynamic system. The Kalman filter assumes that all input data is on the same time basis and that there is a clear causal relationship between the measured data and the system state. If the input data is not synchronized, the Kalman filter may incorrectly associate the measured data at the current moment with the state estimate at the previous moment, resulting in inaccurate state estimation.

[0116] Since the first sensor and the second sensor start to detect the gas to be tested at the same time, the time point of the first sampling of the two is the same, but since the response time of the two sensors is different, the time of generating the sensor data point is different. If the time point of data acquisition is used as the timestamp, the time points of subsequent sampling may no longer be synchronized; or if the time when the sensor data point is obtained is used as the timestamp, the time points when the sensor data points generated by the two sensors are obtained may also be out of sync.

[0117] Therefore, in order to ensure the time synchronization of different sensor data input to the Kalman filter, when the available concentration data points include concentration data points converted from the sensor data points generated by the first sensor and the second sensor, all available concentration data points should be time-aligned.

[0118] In one embodiment, the time alignment is performed based on the timestamp of the concentration data point (i.e., the second concentration data point) converted from the sensor data point generated by the second sensor. For the missing concentration data point at a certain reference time point, the value of the missing concentration data point can be estimated by using methods including but not limited to interpolation or resampling.

[0119] Optionally, when the available concentration data points only include concentration data points converted from sensing data points generated by the first sensor or the second sensor, all the available concentration data points are determined as the available measurement data.

[0120] It is understandable that if the step size of the set time step is greater than the response time of the second sensor, the current time step may contain more than two available concentration data points, including multiple measurements from different sensors or from the same sensor.

[0121] The core assumption of the Kalman filter is that there is only one measurement data point in each time step. For the case where multiple measurement data are contained in a time step, the Kalman filter will not make a separate prediction and update for each data point, but will combine all the measurement data points for a state update, for example, weighted averaging or covariance matrix fusion of multiple measurement data. (Exemplary, weighted averaging: the measurement data can be weighted averaged according to the detection accuracy or reliability of each sensor, for example, a larger weight is set for the first sensor with higher detection accuracy; covariance matrix fusion: multiple measurement data are fused using the covariance matrix to calculate the comprehensive measurement value and comprehensive covariance).

[0122] Therefore, if the available data points only include the first concentration data point or the second concentration data point, it means that the current time step only includes data from the first sensor or the second sensor. At this time, even if there are multiple concentration data points, time alignment is not required.

[0123] S270. Utilize the preset Kalman filter to obtain the predicted concentration value of the target gas at the current time step, and update the state of the preset Kalman filter according to the available measurement data and the current predicted concentration value to obtain the estimated concentration value of the target gas at the current time step.

[0124] The embodiment of the present invention first obtains the sensor data points generated by the first sensor and the second sensor for simultaneously detecting the concentration of the target gas contained in the gas to be tested; then, the first data points generated by the first sensor and the second sensor generated by the second sensor are processed by using the pre-trained BP neural network combined with the Lambert-Beer law, and the first concentration data points and the second concentration data points are converted respectively; then, the concentration data points whose time marked by the timestamp belongs to the current time step are selected as the available concentration data points, and when the available concentration data points contain the first concentration data points and the second concentration data points at the same time, all the available concentration data points are time-aligned, so that the available concentration data points after time alignment are determined as the available measurement data; finally, the concentration prediction value of the target gas at the current time step is obtained by using the preset Kalman filter, and the state of the preset Kalman filter is updated according to the available measurement data and the current concentration prediction value, so as to obtain the concentration estimation value of the target gas at the current time step. Therefore, the embodiment of the present invention realizes the data fusion processing of the first sensor and the second sensor, so that when detecting the concentration of the target gas, the detection accuracy achieved is higher than the detection accuracy of the second sensor, and the response speed achieved is faster than the response speed of the first sensor. Compared with the gas concentration detection scheme using only the first sensor and only the second sensor, the embodiment of the present invention achieves both high detection accuracy and fast response speed. In addition, the embodiment of the present invention ensures the accuracy and effectiveness of data fusion processing by time-aligning the available concentration data points, thereby further improving the detection accuracy.

[0125] Embodiment 3

[0126] Figure 3 This is a schematic diagram of the structure of a gas concentration detection device provided by an embodiment of the present invention. This embodiment can be applied to the sensor data points of two gas sensors based on the spectral absorption principle, and the BP neural network is used to extract the characteristic value and the Kalman filter is used to perform data fusion processing to achieve high detection accuracy and fast response speed. The device can be implemented by hardware and / or software, and is generally integrated on the gas concentration detection equipment.

[0127] like Figure 3 As shown, the device includes: a data acquisition module 300, a data processing module 310, a data determination module 320 and a concentration estimation module 330. Among them,

[0128] The data acquisition module 300 is used to acquire the sensing data point in real time when the first sensor and the second sensor simultaneously start to detect the concentration of the target gas contained in the gas to be detected and at least one of the two generates the sensing data point;

[0129] The data processing module 310 is used to convert each of the sensing data points into a concentration data point by using a pre-trained back propagation BP neural network in combination with the Lambert-Beer law, wherein the characteristic item of each of the concentration data points includes a timestamp;

[0130] A data determination module 320, configured to determine available measurement data of a preset Kalman filter at a current time step from all the concentration data points according to each of the timestamps;

[0131] A concentration estimation module 330 is used to obtain a concentration prediction value of the target gas at the current time step by using the preset Kalman filter, and to update the state of the preset Kalman filter according to the available measurement data and the current concentration prediction value to obtain a concentration estimation value of the target gas at the current time step;

[0132] The first sensor and the second sensor are both gas sensors based on the spectral absorption principle, the detection accuracy of the first sensor is higher than that of the second sensor, and the response speed of the first sensor is lower than that of the second sensor.

[0133] Based on the above embodiments, the data processing module 310 includes:

[0134] A first feature extraction unit is used for, when the sensor data points include a first sensor data point generated by the first sensor, extracting feature values ​​from the first sensor data point using the BP neural network to obtain a first feature data point; wherein the feature items of the first feature data point include: a first timestamp, a first incident light intensity, and a first outgoing light intensity;

[0135] A first concentration conversion unit, configured to convert the first characteristic data point into a first concentration data point according to the first incident light intensity and the first exit light intensity in combination with the Beer-Lambert law;

[0136] A second feature extraction unit is used for, when the sensor data points include a second sensor data point generated by the second sensor, extracting feature values ​​from the second sensor data point using the BP neural network to obtain a second feature data point; wherein the feature items of the second feature data point include: a second timestamp, a second incident light intensity, and a second outgoing light intensity;

[0137] The second concentration conversion unit is used to convert the second characteristic data point into a second concentration data point according to the second incident light intensity and the second exit light intensity in combination with the Beer-Lambert law.

[0138] Based on the above embodiments, the first feature extraction unit includes:

[0139] a first conversion subunit, configured to convert the first characteristic data point into a first absorbance data point according to the first incident light intensity and the first exit light intensity;

[0140] A first normalization subunit, configured to perform normalization processing on the first absorbance data point to obtain a first normalized data point;

[0141] The second conversion subunit is used to apply Beer-Lambert's law to convert the first normalized data point into the first concentration data point, wherein the feature item of the first concentration data point includes the first timestamp and a first converted concentration value.

[0142] Based on the above embodiments, the second feature extraction unit includes:

[0143] a third conversion subunit, configured to convert the second characteristic data point into a second absorbance data point according to the second incident light intensity and the second exit light intensity;

[0144] A second normalization subunit, used for performing normalization processing on the second absorption rate data point to obtain a second normalized data point;

[0145] The fourth conversion subunit is used to apply the Beer-Lambert law to convert the second normalized data point into the second concentration data point, wherein the feature item of the second concentration data point includes the second timestamp and the second converted concentration value.

[0146] Based on the above embodiments, the second feature extraction unit further includes:

[0147] The concentration correction subunit is used to correct the second converted concentration value according to a preset correction algorithm, and update the second concentration data point according to the corrected second converted concentration value.

[0148] Based on the above embodiments, the data determination module 320 includes:

[0149] A data point selection unit, configured to select, according to each of the timestamps, from all the concentration data points, a concentration data point whose time is marked by a timestamp belonging to the current time step as an available concentration data point;

[0150] A data determination unit is used to time-align all the available concentration data points when the available concentration data points include concentration data points converted from the sensing data points generated by the first sensor and the second sensor, and determine the time-aligned available concentration data points as the available measurement data.

[0151] Optionally, the first sensor includes a first air chamber, through which the first sensor collects the gas to be measured; the second sensor includes a second air chamber, through which the second sensor collects the gas to be measured; and the optical path length of the first air chamber is longer than the optical path length of the second air chamber.

[0152] The gas concentration detection device provided in the embodiment of the present invention can execute the gas concentration detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0153] Embodiment 4

[0154] Figure 4 is a schematic diagram of the structure of a gas concentration detection device provided by the fourth embodiment of the present invention, such as Figure 4 As shown, the gas concentration detection device includes a processor 40, a first sensor 41, a second sensor 42, a memory 43, an input device 44 and an output device 45; the number of the processor 40 in the gas concentration detection device can be one or more. Figure 4 A processor 40 is taken as an example; the processor 40, the first sensor 41, the second sensor 42, the memory 43, the input device 44 and the output device 45 can be connected by a bus or other means. Figure 4 The example of connecting through bus is taken in the following.

[0155] The first sensor 41 and the second sensor 42 are used to simultaneously detect the concentration of the target gas contained in the gas to be detected and generate sensing data points.

[0156] The memory 43, as a computer-readable storage medium, can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the gas concentration detection method in the embodiment of the present invention (for example, the data acquisition module 300, the data processing module 310, the data determination module 320 and the concentration estimation module 330 in the gas concentration detection device). The processor 40 executes various functional applications and data processing of the gas concentration detection device by running the software programs, instructions and modules stored in the memory 43, that is, realizes the above-mentioned gas concentration detection method.

[0157] The memory 43 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory 43 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 43 may further include a memory remotely arranged relative to the processor 40, and these remote memories may be connected to the device / terminal / server via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0158] The input device 44 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the device / terminal / server. The output device 45 may include a display device such as a display screen.

[0159] Figure 5 FIG. 4 is a structural example diagram of a gas concentration detection device provided by Embodiment 4 of the present invention. Figure 5 As shown, the gas concentration detection device includes: a first sensor 50, a second sensor 51, a first preprocessing module 52, a second preprocessing module 53, a data determination module 54 and a concentration estimation module 55. Among them,

[0160] The first sensor 50 is used to detect the concentration of the target gas contained in the gas to be tested and generate a first sensing data point;

[0161] The second sensor 51 is used to start detecting the concentration of the target gas simultaneously with the first sensor 50 and generate a second sensing data point;

[0162] A first preprocessing module 52 is connected to the first sensor 50 and is used to obtain the first sensor data point in real time, and use a pre-trained back propagation BP neural network in combination with Lambert-Beer's law to process the first sensor data point to obtain a first concentration data point, wherein the characteristic item of the first concentration data point includes a first timestamp;

[0163] A second preprocessing module 53 is connected to the second sensor 51, and is used to obtain the second sensor data point in real time, and use the BP neural network in combination with the Lambert-Beer law to process the second sensor data point to obtain a second concentration data point, wherein the characteristic item of the second concentration data point includes a second timestamp;

[0164] A data determination module 54, connected to the first preprocessing module 52 and the second preprocessing module 53, respectively, for determining available measurement data of a preset Kalman filter at a current time step from the first concentration data point and the second concentration data point according to each of the first timestamp and the second timestamp;

[0165] A concentration estimation module 55, connected to the data determination module 54, is used to obtain a concentration prediction value of the target gas at the current time step by using the preset Kalman filter, and to update the state of the preset Kalman filter according to the available measurement data and the current concentration prediction value to obtain a concentration estimation value of the target gas at the current time step;

[0166] Among them, the first sensor 50 and the second sensor 51 are both gas sensors based on the principle of spectral absorption. The detection accuracy of the first sensor 50 is higher than that of the second sensor 51, and the response speed of the first sensor 50 is lower than that of the second sensor 51.

[0167] Embodiment 5

[0168] Embodiment 5 of the present invention further provides a storage medium containing computer executable instructions, wherein the computer executable instructions are used to execute a gas concentration detection method when executed by a computer processor, the method comprising:

[0169] When the first sensor and the second sensor simultaneously start to detect the concentration of the target gas contained in the gas to be detected and at least one of the two generates a sensing data point, the sensing data point is acquired in real time;

[0170] Using a pre-trained back propagation BP neural network in combination with Beer-Lambert's law, each of the sensor data points is converted into a concentration data point, and the characteristic item of each of the concentration data points includes a timestamp;

[0171] Determining available measurement data of a preset Kalman filter at a current time step from all the concentration data points according to each of the timestamps;

[0172] Using the preset Kalman filter to obtain a predicted concentration value of the target gas at the current time step, and updating the state of the preset Kalman filter according to the available measurement data and the current predicted concentration value to obtain an estimated concentration value of the target gas at the current time step;

[0173] The first sensor and the second sensor are both gas sensors based on the spectral absorption principle, the detection accuracy of the first sensor is higher than that of the second sensor, and the response speed of the first sensor is lower than that of the second sensor.

[0174] Of course, the computer executable instructions of a storage medium including computer executable instructions provided by an embodiment of the present invention are not limited to the operations of the method described above, but can also execute related operations in the gas concentration detection method provided by any embodiment of the present invention.

[0175] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (FLASH), a hard disk or an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0176] It is worth noting that in the embodiment of the above-mentioned gas concentration detection device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0177] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A gas concentration detection method, characterized in that: include: When the first sensor and the second sensor simultaneously start to detect the concentration of the target gas contained in the gas to be detected and at least one of the two generates a sensing data point, the sensing data point is acquired in real time; Using a pre-trained back propagation BP neural network in combination with Beer-Lambert's law, each of the sensor data points is converted into a concentration data point, and the characteristic item of each of the concentration data points includes a timestamp; Determining available measurement data of a preset Kalman filter at a current time step from all the concentration data points according to each of the timestamps; Using the preset Kalman filter to obtain a predicted concentration value of the target gas at the current time step, and updating the state of the preset Kalman filter according to the available measurement data and the current predicted concentration value to obtain an estimated concentration value of the target gas at the current time step; The first sensor and the second sensor are both gas sensors based on the spectral absorption principle, the detection accuracy of the first sensor is higher than that of the second sensor, and the response speed of the first sensor is lower than that of the second sensor.

2. The gas concentration detection method according to claim 1, characterized in that: The method of using the pre-trained back propagation BP neural network and combining the Lambert-Beer law to convert each of the sensor data points into a concentration data point includes: When the sensor data points include a first sensor data point generated by the first sensor, the BP neural network is used to extract feature values ​​of the first sensor data point to obtain a first feature data point; wherein the feature items of the first feature data point include: a first timestamp, a first incident light intensity, and a first outgoing light intensity; According to the first incident light intensity and the first outgoing light intensity, in combination with Beer-Lambert's law, converting the first characteristic data point into a first concentration data point; When the sensor data points include the second sensor data points generated by the second sensor, the BP neural network is used to extract feature values ​​of the second sensor data points to obtain second feature data points; wherein the feature items of the second feature data points include: a second timestamp, a second incident light intensity, and a second outgoing light intensity; According to the second incident light intensity and the second exiting light intensity, in combination with the Beer-Lambert law, the second characteristic data point is converted into a second concentration data point.

3. The gas concentration detection method according to claim 2, characterized in that: The first concentration data point is obtained according to the first characteristic data point in combination with Beer-Lambert's law, including: Converting the first characteristic data point into a first absorption rate data point according to the first incident light intensity and the first exit light intensity; Normalizing the first absorbance data point to obtain a first normalized data point; The first normalized data point is converted into the first concentration data point by applying Beer-Lambert's law, wherein the feature item of the first concentration data point includes the first time stamp and a first converted concentration value.

4. The gas concentration detection method according to claim 2, characterized in that: The second concentration data point is obtained according to the second characteristic data point in combination with the Beer-Lambert law, including: Converting the second characteristic data point into a second absorption rate data point according to the second incident light intensity and the second exit light intensity; Normalizing the second absorbance data point to obtain a second normalized data point; The second normalized data point is converted into the second concentration data point by applying Beer-Lambert's law, wherein the feature item of the second concentration data point includes the second time stamp and the second converted concentration value.

5. The gas concentration detection method according to claim 4, characterized in that: After converting the second normalized data point into the second concentration data point, the method further comprises: The second converted concentration value is corrected according to a preset correction algorithm, and the second concentration data point is updated according to the corrected second converted concentration value.

6. The gas concentration detection method according to claim 1, characterized in that: Determining available measurement data of a preset Kalman filter at a current time step from all the concentration data points according to each of the timestamps includes: According to each of the timestamps, a concentration data point whose time marked by the timestamp belongs to the current time step is selected from all the concentration data points as an available concentration data point; When the available concentration data points include concentration data points converted from sensing data points generated by the first sensor and the second sensor, all the available concentration data points are time-aligned, and the time-aligned available concentration data points are determined as the available measurement data.

7. The gas concentration detection method according to any one of claims 1 to 6, characterized in that: The first sensor comprises a first air chamber, and the first sensor collects the gas to be measured through the first air chamber; The second sensor comprises a second air chamber, and the second sensor collects the gas to be measured through the second air chamber; The optical path length of the first air chamber is longer than the optical path length of the second air chamber.

8. A gas concentration detection device, characterized in that: include: A data acquisition module, used for acquiring the sensing data point in real time when the first sensor and the second sensor simultaneously start to detect the concentration of the target gas contained in the gas to be detected and at least one of the two generates a sensing data point; A data processing module, used for converting each of the sensing data points into a concentration data point by using a pre-trained back propagation BP neural network in combination with Lambert-Beer's law, wherein the characteristic item of each of the concentration data points includes a timestamp; A data determination module, used to determine available measurement data of a preset Kalman filter at a current time step from all the concentration data points according to each of the timestamps; a concentration estimation module, configured to obtain a concentration prediction value of the target gas at the current time step by using the preset Kalman filter, and to update the state of the preset Kalman filter according to the available measurement data and the current concentration prediction value, so as to obtain a concentration estimation value of the target gas at the current time step; The first sensor and the second sensor are both gas sensors based on the spectral absorption principle, the detection accuracy of the first sensor is higher than that of the second sensor, and the response speed of the first sensor is lower than that of the second sensor.

9. A gas concentration detection device, characterized in that: include: a first sensor, a second sensor, a memory, and one or more processors; Wherein, the one or more processors are respectively connected to the first sensor, the second sensor and the memory; The first sensor and the second sensor are used to simultaneously start detecting the concentration of the target gas contained in the gas to be tested and generate sensing data points; The memory is used to store one or more programs; The one or more programs are executed by the one or more processors to implement the gas concentration detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the gas concentration detection method according to any one of claims 1 to 7 is implemented.

Images