METHOD AND DEVICE FOR OPERATING A GAS SENSOR

Description

Technical area

[0001] The invention relates to gas sensors, and in particular to methods for determining a concentration value taking into account storage-related changes in the sensitivity of gas sensors. Technical background

[0002] Gas sensors are known for determining the concentrations of gaseous substances and can generally be designed as chemical sensors with various sensor principles and sensitive materials. Typically, a chemical reaction or absorption of a gaseous component to be measured at a sensitive layer in the sensor structure leads to a change in the properties of the sensitive layer. This change in the properties of the sensitive layer can be detected and evaluated as a change in an electrical quantity.

[0003] Depending on the sensor material used, the behavior of the gas sensor is more or less stable, and thus the condition of the gas sensor can change during its lifetime depending on environmental conditions. Particularly during storage periods, the sensor properties can change due to influences such as ambient temperature, humidity, ambient air contamination, and the like, to such an extent that a reliable and accurate determination of the gas concentration of a gas component in a measurement gas is no longer possible. This problem occurs primarily with gas sensors that use organic materials, and especially organic semiconductors, in the sensitive layer, since signal-altering molecules can be absorbed or released into the sensitive layer during storage periods.

[0004] Document CN106405 007 B discloses a calibration method for a gas sensor, comprising the steps of: setting up a sensor calibration model, wherein a test chamber provides calibration data for the calibration model, and determining the linear and nonlinear parameters of the influencing factors of the various sensors in the calibration model based on the standard calibration data provided by the test chamber. The calibration model is based on an artificial neural network that can automatically extract the temperature and humidity of the sensor from the experimental data.

[0005] WO 2019 / 128203 A1 discloses a network monitor for toxic and harmful gases with a sensor detection component, comprising: a housing, a main control board in the housing, and a detection module, a data processing module, and a power supply system, each connected to the main control board. The detection module includes two or more toxic and harmful gas sensors to realize real-time detection of a variety of toxic and harmful gases; the main control board is provided with a signal acquisition unit, a wireless transmission unit, a GPS positioning unit, and a data storage unit. The main control board is connected to the Internet of Things-based management system via a wireless transmission unit to transmit the detection data to the management system via the Internet of Things.

[0006] Document CN 109 472 303 A1 discloses a gas sensor drift compensation method based on a self-coding network decision, comprising the following steps: selecting one-dimensional or multi-dimensional different eigenvalues ​​in the gas sensor array data and obtaining corresponding mean values, and selecting and removing the abnormal sampling points of the sensor array whose corresponding eigenvalues ​​are greater or smaller than a certain multiple of the mean value;Constructing a batch self-coding network, the batch self-coding network consisting of a noise reduction self-coder and a joint self-coder, and reducing the preprocessed training sample feature values ​​by a noise reduction self-coder and obtaining the coded features, and combining the stored feature values ​​and the original tags to code the training samples, and classifying the samples by the CART algorithm, comparing and analyzing the classifications before and after sample coding; Disclosure of the invention

[0007] According to the invention, a method for operating a gas sensor according to claim 1 as well as a device and a gas sensor according to the independent claims are provided.

[0008] Further embodiments are specified in the dependent claims.

[0009] According to a first aspect, a method for operating a gas sensor system with a gas sensor having a sensitive layer is provided to provide a concentration variable of a gas concentration of a gas component in a measurement gas, comprising the following steps: Performing a measurement of the gas concentration during a measurement process to obtain a temporal profile of a sensor signal dependent on the gas concentration; determining the concentration value using a data-based sensor model depending on the temporal profile of the sensor signal, wherein the data-based sensor model takes into account sensor behavior outside the measurement process to determine the corrected concentration value.

[0010] The data-based sensor model is designed to determine the concentration value depending on the temporal course of the sensor signal and depending on the sensor behavior outside the measurement process.

[0011] The above method provides for determining a gas concentration of a gas component to be detected in the sample gas in the form of a concentration variable based on the observation of a time-dependent sensor signal during a measurement process involving exposure of a sensitive layer to measurement gas and based on sensor behavior outside the measurement period using a data-based sensor model. The evaluation of sensor behavior outside the measurement period makes it possible to take storage- or aging-related sensor conditions into account when determining the gas concentration.

[0012] In particular, the above method assumes that the sensor behavior outside of the actual measurement process is generally characteristic of the sensor state. Since the actual sensor state cannot usually be directly determined or quantified, the sensor state can be implicitly taken into account when determining the gas concentration of the gas components to be detected, depending on the sensor behavior, using the trainable, data-based sensor model. For example, the behavior of the sensor signal during heating and cooling phases that precede or follow the measurement process can allow conclusions to be drawn about the storage- or aging-related sensor state.

[0013] Thus, a sensor model can be created or provided which, based on the sensor signals during the measurement and based on measurement-independent sensor signals or operating parameters of the gas sensor, contributes to determining a more precise concentration indication of the gas component to be detected in the measurement gas.

[0014] Operating parameters that can be considered include, in particular, device-specific calibration parameters and sensor states, such as sensor temperature, sample gas temperature, and sample humidity. A device-specific calibration parameter can generally specify device- or sensor-specific information, in particular the sensitivity of the gas sensor.

[0015] Furthermore, these operating parameters can also be considered as time series. For example, the temperature profile of the sample gas, the humidity profile of the sample gas, and the profile and humidity of a purge gas used to flush the gas sensor outside of the measurement process can be taken into account.

[0016] Any regression algorithm, such as Lasso, Random Forest, Gaussian process, and neural networks, can be used as a data-based sensor model. Data-based models that are translationally invariant with respect to the duration of individual time periods, such as the measurement process, a preceding or subsequent heating, cooling, and holding phase, depending on the sensor operating mode in the measurement process, are particularly preferred, as is the case with convolutional neural networks, for example.

[0017] In particular, the data-based sensor model can be trained to take into account sensor behavior dependent on aging or degradation of the gas sensor to determine the concentration value.

[0018] The above method makes it possible to better account for changes in sensor conditions caused by storage and aging when determining gas concentrations. This is achieved by evaluating the gas sensor's behavior, which depends on the sensor condition caused by storage and aging, particularly outside the measurement period, so that a precise evaluation of the sensor signal obtained during the measurement process is always possible.

[0019] The above method represents a significant improvement over conventional calibration procedures, which are merely static and cannot adequately account for changing sensor conditions over time. Furthermore, conventional calibration methods are complex and generally cannot be performed on-site.

[0020] In particular, the corrected concentration quantity can be determined depending on a physically modeled concentration quantity and depending on a concentration quantity determined from the sensor model, in particular according to a predetermined weighting function.

[0021] It can be provided that the data-based sensor model receives measurement variables related to the measurement process and measurement characteristics derived therefrom as input variables.

[0022] In particular, the measurement variables may comprise one or more variables resulting from the time-dependent sensor signal during the duration of the measurement process, in particular values ​​of the sensor signal at several sampling times between the start and the end of the measurement process and / or one or more variables derived therefrom, in particular an absolute signal change of the sensor signal between the start and the end of the measurement process and / or the maximum or average gradient of the sensor signal S during the measurement process.

[0023] Furthermore, the measurement variables can comprise one or more variables resulting from the temporal sensor signal during a period outside the period of the measuring process, in particular during a period of time that precedes and / or follows the loading phase, in particular during a period during which a specific sensor state is set, in particular values ​​of the sensor signal at several sampling times between the beginning and the end of the period in question and / or one or more variables derived therefrom, in particular an absolute signal change between the beginning and the end of the period in question and / or the maximum or average gradient of the sensor signal S during the period in question.

[0024] It may be provided that the measurement quantities include one or more of the following quantities: one or more device- or sensor-specific calibration parameters, which in particular indicate a sensitivity of the gas sensor; one or more quantities resulting from a temporal signal of one or more further sensors integrated in the system, in particular a sample gas temperature sensor, a sample gas humidity sensor and a gas sensor temperature sensor, during the period of the measurement process and / or a period outside the period of the measurement process, in particular values ​​of the corresponding signal at several sampling times between the start and the end of the measurement process or the period in question and / or one or more quantities derived therefrom, in particular an absolute signal change between the start and the end of the period in question and / or the maximum or average gradient of the signal of the one or more further sensors integrated in the system during the period in question;and one or more quantities resulting from the difference between the time-dependent signal of one or more sensors integrated in the system, in particular the gas sensor, during the period of the measuring process and / or a period outside the period of the measuring process and the time-dependent signal of the same sensor(s) in the same period of a previous measurement, in particular a calibration measurement, in particular values ​​of the corresponding difference at several sampling times between the start and the end of the measuring process or the period in question and / or one or more quantities derived therefrom. ;

[0025] Furthermore, the measurement characteristics can be derived from the measurement quantities and include one or more of the following quantities: an indication of a signal response by applying a test voltage pulse, a proportionality factor between the temperature and the sensor signal and / or a time constant of the signal response; one or more signal responses caused by sudden changes in the composition and / or pressure of the measurement gas impinging on the sensor surface (31); a baseline value corresponding to a raw gas sensor signal at a defined point in time, in particular at the start of the measurement process; one or more parameters of physical models fitted to the measurement data; a parameter of an exponential function fitted to the sensor signal during the heating or cooling phase; one or more modified measurement variables, in particular by compensating for the temperature influence on the sensor signal by means of a corresponding proportionality factor;one or more discrete values ​​of the sensor signal in time periods before or after the measurement process as a function of temperature; and a difference or a quotient between successively recorded values ​​of at least one of the measured variables, in particular a difference or a quotient between successively recorded values ​​of at least one of the measured variables between a current value of the at least one measured variable and a reference value of the at least one measured variable.

[0026] According to a further aspect, a device for operating a gas sensor system with a gas sensor is provided to provide a concentration variable of a gas concentration of a gas component in a measurement gas, wherein the device is designed to: Performing a measurement of the gas concentration during a measurement process to obtain a temporal profile of a sensor signal dependent on the gas concentration; and determining the concentration value using a data-based sensor model depending on the temporal profile of the sensor signal, wherein the data-based sensor model is trained to take into account sensor behavior outside the measurement process to determine the concentration value.

[0027] The data-based sensor model is designed to determine the concentration value depending on the temporal course of the sensor signal and depending on the sensor behavior outside the measurement process.

[0028] According to a further aspect, a gas sensor system comprising a gas sensor and the above device is provided. Brief description of the drawings

[0029] Embodiments are described in more detail below with reference to the accompanying drawings. They show: Figure 1 shows a schematic representation of a gas sensor; Figure 2 shows a block diagram illustrating a function for determining a gas concentration using a gas sensor according to an example; and Figure 3 shows a functional circuit diagram for determining a gas concentration using a gas sensor according to another embodiment. Description of embodiments

[0030] Figur 1 shows a schematic representation of a gas sensor system 1 with a measuring chamber 2 in which a gas sensor 3 is arranged. The gas sensor 3 has a sensor surface 31 with a sensitive layer, which is exposed to the gas component to be detected and which provides an electrical sensor variable that indicates a gas concentration of the gas component to be detected.

[0031] Controlled by a control unit 4, a measurement gas can be introduced into the measuring chamber 2 via a gas supply line 5, in which the concentration of a gas component to be detected is to be measured. Furthermore, a purge gas can be introduced into the measuring chamber 2 to calibrate the gas sensor 3. Typically, the measurement gas is discharged via another channel, so that the gas sensor 3 is continuously flushed with the measurement gas.

[0032] The control unit 4 can be provided locally or implemented cloud-based and be in communication with the gas sensor 3.

[0033] The gas sensor 3 is provided with a heating device 6 to heat the gas sensor 3 to different temperature levels.

[0034] Gas sensors can employ different measurement principles. Semiconductor gas sensors, in particular, can be operated in a temperature-cycling mode, whereby oxygen accumulates on the sensor surface during a high-temperature phase. The oxygen is subsequently displaced by the gas component to be detected during a low-temperature phase. The conductivity change during the low-temperature phase can then be evaluated accordingly to obtain a gas concentration. Alternatively, the gas concentration of the gas component to be detected in the measurement gas can also be determined by the conductivity of the semiconductor gas sensor in an equilibrium state of the sensor surface of the gas sensor 3.

[0035] Other measuring principles of gas sensors also provide for temporal phases with different temperature exposures.

[0036] The gas sensor 3 is generally designed such that a chemical reaction, adsorption or absorption of a detected gas component to be measured at the sensor surface 31 leads to a change in the properties of the sensor surface 31, which is reflected in a change in an electrical sensor variable (sensor signal), such as in the form of a change in a sensor current, a sensor voltage or a sensor resistance (sensor conductivity).

[0037] During its service life, gas sensor 3 is subject to changes in condition due to external influences such as temperature, humidity, and material exposure. This changes the sensor condition and thus its sensitivity. This results in storage- and aging-related miscalibration of the gas sensors, which must be corrected. Conventional methods require recalibrations, which are time-consuming and only performed at specific times, between which gas sensor 3 is not properly calibrated.

[0038] Basically, a concentration variable is to be provided using the gas sensor system 1. The concentration variable indicates the gas concentration of a gas component to be detected in a measuring gas flowing over the sensor surface 31. The aging influences and environmental influences of a gas sensor 3 are to be taken into account. To determine these influences on the concentration measurement due to a change in a storage- and aging-related sensor condition, a data-based sensor model is used in the control unit 4, which is trained to either to determine a correction value for applying the concentration value derived from the measured sensor value, or to determine the concentration value directly from the course of the electrical sensor value during a concentration measurement and other operating parameters and / or behavioral parameters of the gas sensor 3.

[0039] In Figur 2 As an example, a functional diagram is shown to illustrate the determination of the concentration variable for the gas concentration of a gas component based on a correction variable determined by the data-based sensor model. For this purpose, a raw concentration variable C phys is determined using a physical sensor model 21, which is created based on physical laws and depends on the specific sensor technology. This raw concentration variable indicates a gas concentration for an unchanged (e.g., unaged or in the as-manufactured) gas sensor 3. In principle, the underlying physical sensor model 21 can indicate the sensor sensitivity for a new, unaged gas sensor 3.

[0040] The physical sensor model 21 can, for example, comprise any mathematical function, such as a series expansion, which typically has a large number of model parameters, or the like. The model parameters of the physical sensor model are generally globally optimized and device-specifically calibrated, and thus fixed for the individual gas sensor 3.

[0041] The raw concentration quantity C phys can be determined in the sensor model 21 as a function of the sensor signal S and one or more of the following measurement quantities M1, M2, M3,...: Quantities resulting from the time-dependent sensor signal S during the duration of the measuring process, i.e. during the exposure phase to the measuring gas, such as values ​​of the sensor signal at several sampling times between the start and the end of the measuring process and / or quantities derived therefrom, such as the absolute signal change between the start and the end of the measuring process or the maximum or average gradient of the sensor signal S during the measuring process. Quantities resulting from the temporal sensor signal during a period outside the period of the measuring process, in particular during a period that precedes and / or follows the exposure phase, and in particular during a period during which a specific sensor state is set, in particular during a period of a predetermined temperature profile, in particular a heating, cooling or temperature maintenance phase, such asValues ​​of the sensor signal at multiple sampling points between the beginning and end of the period in question and / or quantities derived therefrom, such as the absolute signal change between the beginning and end of the period in question or the maximum or average slope of the sensor signal S during the period in question; one or more device-specific calibration parameters that contribute to the accuracy of the model by providing device-specific information, in particular the sensitivity of the gas sensor, depending on its type. These calibration parameters can be specified and / or empirically determined.Quantities resulting from a temporal signal of one or more sensors further integrated in the system, such as a sample gas temperature sensor, a sample gas humidity sensor and a gas sensor temperature sensor, during the period of the measurement process and / or a period outside the period of the measurement process, such as values ​​of the sensor signal at several sampling times between the start and the end of the measurement process or the period in question and / or one or more quantities derived therefrom, in particular an absolute signal change between the start and the end of the period in question and / or the maximum or average gradient of the signal of the one or more sensors further integrated in the system during the period in question.one or more quantities resulting from the difference between the time-dependent signal of one or more sensors integrated in the system, in particular the sensor signal S, during the period of the measuring process and / or a period outside the period of the measuring process and the time-dependent signal of the same sensor(s) in the same period of a previous measurement, in particular a calibration measurement, in particular values ​​of the corresponding difference at several sampling times between the start and the end of the measuring process or the period in question and / or one or more quantities derived therefrom.

[0042] For example, the raw concentration size C phys using one of the following models parameterized with the model parameters a,b,c,... from the measured variables M1, M2, M3,... determined in this way with: a linear model according to C phys = aM 1 + bM 2 + cM 3 + d a power model according to C phys = aM 1 b + cM 2 d + eM 3 f + ⋯ + h or a physical model assuming Langmuir adsorption and neglecting desorption processes during the measurement process according to C phys = f M 1 , M 2 , M 3 … , a , b , c …

[0043] In a sensor model block 22, a correction value K is now determined, with which the raw concentration value determined from the physically motivated sensor model 21 C phys in a correction block 23 in order to obtain a corrected concentration quantity C korr. In particular, the correction quantity K can be a correction factor for multiplicatively applying the raw concentration quantity C phys or a correction offset for additive loading of the raw concentration size C phys be.

[0044] The correction value K can be determined by a data-based sensor model in the sensor model block 22. The data-based sensor model can include a regression model, such as a Lasso model, a Random Forest model, a Gaussian process model, or a neural network.

[0045] Selected measurement variables M1, M2, M3 .....and derived measurement characteristics F1, F2, F3... can be used as input variables for the sensor model.

[0046] The data-based sensor model determines the correction quantity K from one or more of the above measurement quantities M1, M2, M3 ....

[0047] The measurement features F1, F2, F3... can be derived from the measurement variables M1, M2, M3 in the measurement feature block 24 and serve to condense the information contained in the measurement variables. For example, the measurement features F1, F2, F3... can include one or more of the following measurement features: a signal response by applying a test voltage pulse, a proportionality factor between the temperature and the sensor signal and / or a time constant of the signal response; signal responses caused by sudden changes in the composition (in particular humidity) and / or pressure of the measurement gas impinging on the sensor surface 31. These changes occur in particular at the beginning or end of the measurement process when the gas composition changes; a baseline value corresponding to a raw gas sensor signal at a defined point in time (e.g., at the start of the measurement process); one or more parameters of physical models that are fitted to the measurement data. For example, a model of the adsorption and desorption kinetics can be fitted to the sensor signal during the application and, if applicable, subsequent holding phase, and the fitted physical parameters (e.g.,Adsorption enthalpy, rate constant) can be used as a measurement characteristic for the sensor model. A parameter of empirical relationships that is fitted to the measured data, such as a parameter of an exponential function that is fitted to the sensor signal during the heating or cooling phase.Modified measurement variables, in particular by compensating for the temperature influence on the raw gas sensor signal by means of a corresponding proportionality factor; One or more discrete values ​​of the sensor signal in time periods before or after the measurement process as a function of temperature. Sensor values ​​converted from relative to absolute gas humidity using the Antoine equation. A difference or a quotient between successively recorded values ​​of at least one of the measurement variables. At least one of the measurement variables, in particular a difference or a quotient between successively recorded values ​​of at least one of the measurement variables. Between a current value of the at least one measurement variable and a reference value of the at least one measurement variable.

[0048] The data-based sensor model is trained with training data that maps measurement characteristics to a corresponding concentration quantity (ground truth). The measurement characteristics are obtained from measurement quantities of gas sensors 3, each exposed to a calibration measurement gas. The concentration quantity corresponds to the gas concentration in the calibration measurement gas, with the gas sensors 3 having been stored under different storage conditions. By back-calculating based on the concentration quantity obtained from the physically motivated model, the data-based sensor model can be correlated accordingly.

[0049] To train the data-based sensor model, experimentally labeled pairs of measurement quantities and measurement characteristics and gas concentration can be generated using a variation of storage histories, i.e., temporal sequences of environmental conditions, e.g., by controlled variation of storage time, temperature and humidity, and using calibration measurement gases with known concentration of the gas component to be detected.

[0050] According to an embodiment as shown in the functional diagram of Figur 3 As shown, the data-based sensor model can directly determine the corrected concentration quantity C korr in a sensor model block 22'. The physical sensor model is implemented in the data-based sensor model together with the correction dependent on the respective sensor state in order to determine the corrected concentration quantity C korr as the model output.

[0051] In a hybrid modeling approach, the physical sensor model and the data-based sensor model can determine corrected concentration variables in parallel. Using a downstream weighting function, both corrected concentration variables can be weighted relative to each other. The weighting can be selected depending on the respective corrected concentration variables, in particular according to a predefined weighting function.

Claims

1. Method for operating a gas sensor system (1) having a gas sensor (3) with a sensitive layer in order to provide a concentration variable of a gas concentration of a gas component in a measurement gas, having the following steps of: - measuring the gas concentration during a measurement operation in order to obtain a temporal progression of a sensor signal (S) that is dependent on the gas concentration; - determining the concentration variable (Ckorr) with the aid of a data-based sensor model on the basis of the temporal progression of the sensor signal (S), wherein the data-based sensor model is trained to take into account a sensor behaviour outside the measurement operation in order to determine the concentration variable (Ckorr), - wherein the data-based sensor model is designed to determine the concentration variable (Ckorr) on the basis of the temporal progression of the sensor signal (S) and on the basis of the sensor behaviour outside the measurement operation.

2. Method according to Claim 1, wherein the data-based sensor model comprises a Gaussian process model, a LASSO algorithm, a random forest algorithm or a neural network, wherein the data-based sensor model is trained to take into account a sensor behaviour in order to determine the concentration variable (Ckorr).

3. Method according to Claim 1 or 2, wherein the concentration variable (Ckorr) is determined on the basis of a physically modelled concentration variable (Cphys) and on the basis of a concentration variable (Ckorr) determined with the aid of the sensor model, in particular according to a predefined weighting function.

4. Method according to one of Claims 1 to 3, wherein the data-based sensor model receives measurement variables associated with the measurement operation and / or measurement features (F1, F2, F3...) derived therefrom as input variables.

5. Method according to Claim 4, wherein the measurement variables (M1, M2, M3 ...) comprise one or more of the following variables: - variables which result from the time-dependent sensor signal (S) during the duration of the measurement operation, in particular values of the sensor signal (S) at a plurality of sampling times between the beginning and the end of the measurement operation and / or one or more variables derived therefrom, in particular an absolute signal change in the sensor signal (S) between the beginning and the end of the measurement operation and / or the maximum or average gradient of the sensor signal (S) during the measurement operation.

6. Method according to Claim 5, wherein the measurement variables (M1, M2, M3 ...) comprise one or more of the following variables: - variables which result from the temporal sensor signal (S) during a period outside the period of the measurement operation, in particular during a duration which precedes and / or follows the application phase, in particular during a period during which a particular sensor state is set, in particular values of the sensor signal (S) at a plurality of sampling times between the beginning and the end of the relevant period and / or one or more variables derived therefrom, in particular an absolute signal change between the beginning and the end of the relevant period and / or the maximum or average gradient of the sensor signal (S) during the relevant period.

7. Method according to Claim 5 or 6, wherein the measurement variables (M1, M2, M3 .....) comprise one or more of the following variables: - one or more device-specific calibration parameters which indicate, in particular, a sensitivity of the gas sensor (3); and / or - one or more variables which result from a temporal signal from one or more further sensors integrated in the system, in particular a measurement gas temperature sensor, a measurement gas humidity sensor and a gas sensor temperature sensor, during the period of the measurement operation and / or a period outside the period of the measurement operation, in particular values of the corresponding signal at a plurality of sampling times between the beginning and the end of the measurement operation or of the relevant period and / or one or more variables derived therefrom, in particular an absolute signal change between the beginning and the end of the relevant period and / or the maximum or average gradient of the signal from the one or more sensors also integrated in the system during the relevant period.

8. Method according to one of Claims 5 to 7, wherein the measurement features (F1, F2, F3...) are derived from the measurement variables (M1, M2, M3 ...) and comprise one or more of the following variables: - an indication of a signal response by virtue of the application of a test voltage pulse, - a proportionality factor between the temperature and the sensor signal (S) and / or a time constant of the signal response; - one or more signal responses which are caused by sudden changes in the composition and / or the pressure of the measurement gas hitting the sensor surface (31); - a baseline value corresponding to a gas sensor raw signal at a defined time, in particular at the time of the beginning of the measurement operation; - one or more parameters of physical models which are fitted to the measurement data; - a parameter of an exponential function which is fitted to the sensor signal (S) during the heating or cooling phase; - one or more modified measurement variables (M1, M2, M3 ...), in particular by virtue of compensating for the temperature influence on the sensor signal (S) by means of a corresponding proportionality factor; - one or more discrete values of the sensor signal in periods before or after the measurement operation as a function of the temperature; and - a difference or a quotient between successively captured values of at least one of the measurement variables (M1, M2, M3 ...), in particular a difference or a quotient between successively captured values of at least one of the measurement variables (M1, M2, M3 ...) between a current value of the at least one measurement variable (M1, M2, M3 ...) and a reference value of the at least one measurement variable (M1, M2, M3 ...).

9. Device for operating a gas sensor system (1) having a gas sensor (3) with a sensitive layer in order to provide a concentration variable (Ckorr) of a gas concentration of a gas component in a measurement gas, wherein the device is designed to: - measure the gas concentration during a measurement operation in order to obtain a temporal progression of a sensor signal (S) that is dependent on the gas concentration; and - determine the concentration variable (Ckorr) with the aid of a data-based sensor model on the basis of the temporal progression of the sensor signal (S), wherein the data-based sensor model is trained to take into account a sensor behaviour outside the measurement operation in order to determine the concentration variable (Ckorr), - wherein the data-based sensor model is designed to determine the concentration variable (Ckorr) on the basis of the temporal progression of the sensor signal (S) and on the basis of the sensor behaviour outside the measurement operation.

10. Gas sensor system (1) having a gas sensor (3) and a device according to Claim 9.

11. Computer program which is configured to carry out all steps of a method according to one of Claims 1 to 8 in a device according to Claim 9.

12. Electronic storage medium on which a computer program according to Claim 11 is stored.