Method and device for determining nitrogen oxides and ammonia in exhaust gas, vehicle, computer program product, and computer-readable storage medium

Abstract

In a method for determining nitrogen oxides and ammonia in the exhaust gas of a vehicle, the vehicle having a nitrogen oxide sensor, a lambda signal is received. On the basis of the lambda signal, a lambda value is determined which is representative of the combustion air ratio of the vehicle. A lambda gradient is determined which is representative of whether the lambda value has increased or decreased in comparison with a preliminary value. On the basis of the lambda value and the lambda gradient, it is determined whether a measured value of the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas or whether the measured value of the nitrogen oxide sensor is representative of ammonia in the exhaust gas. An information signal is transmitted which comprises information as to whether the measured value of the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas or whether the measured value of the nitrogen oxide sensor is representative of ammonia in the exhaust gas.

Description

[0001] 202400918

[0002] 1

[0003] Description

[0004] Method and apparatus for determining nitrogen oxides and ammonia in exhaust gas, vehicle, computer program product and computer-readable storage medium

[0005] A method for determining nitrogen oxides and ammonia in the exhaust gas of a vehicle is described. Furthermore, a device for determining nitrogen oxides and ammonia in the exhaust gas of a vehicle is described. A vehicle is also described. A computer program is also described. Finally, a computer-readable storage medium is described.

[0006] Many vehicles are equipped with a catalytic converter, particularly a three-way catalytic converter. These vehicles also have one or more nitrogen oxide sensors, so-called NOx sensors. EU7 legislation makes it mandatory to measure and determine a vehicle's NOx emissions using an in-vehicle NOx sensor. Since an NOx sensor based on the amperometric measurement principle measures the sum of NOx and ammonia (NH3) concentrations in the exhaust gas, as it is cross-sensitive to NH3, it is necessary to separate the NOx and NH3 components from the measured total signal.

[0007] One task to be solved is to help reliably distinguish whether a measured value from a vehicle's nitrogen oxide sensor is representative of nitrogen oxides or of ammonia.

[0008] This task is solved by the method and the subject matter of the independent patent claims. Advantageous embodiments, implementations, and further developments are the subject of the respective dependent patent claims.

[0009] First, the procedure for determining nitrogen oxides and ammonia in the exhaust gas of a vehicle is explained. The vehicle is equipped with a nitrogen oxide sensor.

[0010] The nitrogen oxide sensor is typically located downstream of a three-way catalytic converter. The nitrogen oxide sensor, also known as a NOx sensor, is based on the amperometric measurement principle.

[0011] The procedure determines a lambda value that is representative of the vehicle's air-fuel ratio. 202400918

[0012] 2

[0013] The lambda value, for example, is representative of a linear lambda signal. Alternatively, any other signal representative of the oxygen content or lambda value of the exhaust gas can be used. Examples include a binary lambda signal (measured in volts) or an oxygen concentration (measured in ppm or %).

[0014] The lambda value is determined based on a lambda signal. This signal is provided, for example, by the nitrogen oxide sensor. Lambda specifically represents the air-fuel ratio compared to a stoichiometric combustion mixture. Alternatively, it can also be provided by a second nitrogen oxide sensor.

[0015] The procedure determines a lambda gradient that is representative of whether the lambda value has increased or decreased compared to a previous value.

[0016] Depending on the lambda value and the lambda gradient, it is determined whether a measured value from the nitrogen oxide sensor is representative for nitrogen oxides in the exhaust gas or whether the measured value from the nitrogen oxide sensor is representative for ammonia in the exhaust gas.

[0017] Subsequently, an information signal is sent, which includes information on whether the measured value of the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas or whether the measured value of the nitrogen oxide sensor is representative of ammonia in the exhaust gas.

[0018] In general, the following relationship applies between the emissions downstream of a three-way catalyst and the lambda condition of the three-way catalyst or downstream of the three-way catalyst:

[0019] When operating the catalytic converter with a superstoichiometric air-fuel mixture (lambda > 1, so-called "lean operation" or "lean mixture"), not all nitrogen oxides from combustion can be broken down, as the required reducing agent (e.g., CO) has already been oxidized beforehand. As a result, a certain concentration of NOx is present downstream of the catalytic converter, which can be detected by a nitrogen oxide sensor. The nitrogen oxide sensor can also be referred to as a NOx sensor. NOx is used here as a synonym for nitrogen oxides and is a collective term for gaseous oxides of nitrogen, especially nitric oxide (NO) and nitrogen dioxide (NO2).

[0020] 3

[0021] If the catalytic converter is operated with a substoichiometric air-fuel mixture (lambda < 1, so-called "rich operation"), not all hydrocarbons and carbon monoxide present can be broken down. Additionally, ammonia (NH3) is produced as a side reaction during this rich operation, resulting in an additional pollutant. Consequently, the nitrogen oxide sensor downstream of the catalytic converter measures a certain concentration of NH3. However, all NOx emissions are converted to N2 and CO2 because sufficient amounts of carbon monoxide (CO) are present. Therefore, only NH3 exists downstream of the catalytic converter, and no NOx remains.

[0022] However, this simple approach to distinguishing between NOx and NH3 emissions does not allow for a correct separation between NOx and NH3 under all conditions.

[0023] Firstly, NOx and NH3 formation are subject to certain hysteresis effects, for example, due to the oxygen storage capacity of the catalyst. Secondly, the emission behavior can change with increasing catalyst aging. For these reasons, the simple approach described above for separating NOx and NH3 is insufficient to reliably distinguish between NOx and NH3 emissions.

[0024] According to the invention, in addition to the pure lambda value, the lambda gradient is also used to determine whether the sensor is measuring NOx or NH3. In other words, the information about the "source" of the sensor signal is used, i.e., whether it originates from lean or rich operation. This allows the aforementioned hysteresis effects around the range of lambda = 1 to be taken into account.

[0025] This method allows for a more reliable distinction between NOx and NH3.

[0026] In particular, it is well known from the prior art that the signal from a nitrogen oxide sensor downstream of a three-way catalytic converter can be assigned to NOx or NH3 based on the prevailing lambda value (NOx in lean operation and NH3 in rich operation). However, in practice, this simple approach proves to be incorrect or inaccurate.

[0027] 4 is enough to meet future strict emission regulations (e.g. EU7 emission standard).

[0028] A more detailed differentiation of the signal generated by the nitrogen oxide sensor as NOx or NH3 is necessary to adequately meet future stringent emission regulations (e.g., the EU7 emissions standard). In practice, it has been shown that when passing through lambda = 1 conditions (either from rich to lean or from lean to rich), the direction of the transition must be considered in addition to the currently prevailing lambda value. According to the invention, the direction of the "lambda = 1" transition is taken into account by incorporating the lambda gradient as additional information and by dividing the lambda ranges for determining whether NOx or NH3 emissions are present into four relevant ranges instead of the two ranges known from the prior art. This further refinement of the ranges is necessary to adequately meet future stringent emission regulations (e.g., the EU7 emissions standard).

[0029] To distinguish more reliably between NOx and NH3, it is advantageous not only to define a simple, scalar lambda threshold for the differentiation between NOx and NH3.

[0030] Therefore, according to one embodiment of the method, four lambda threshold values ​​are provided, and depending on a comparison of the lambda value with the four lambda threshold values ​​and the lambda gradient, it is determined whether a measured value of the nitrogen oxide sensor is representative for nitrogen oxides in the exhaust gas or whether the measured value of the nitrogen oxide sensor is representative for ammonia in the exhaust gas, wherein the first lambda threshold value of the four lambda threshold values ​​is greater than the third lambda threshold value, the third lambda threshold value is greater than the second lambda threshold value, and the second lambda threshold value is greater than the fourth lambda threshold value.

[0031] Alternatively, the second lambda threshold can also be greater than the third lambda threshold.

[0032] It has been shown that using four threshold values ​​allows for a significantly more accurate separation between NOx and NH3. In particular, this also makes it possible to consider from which direction (i.e., from which lambda value) the system is moving towards or through the lambda = 1 point (for example, from "rich" to "lean", i.e., from lambda < 1 in 202400918).

[0033] 5

[0034] Towards Lambda > 1). Thus, it is possible that the catalyst already causes NOx emissions at Lambda < 1, or conversely, already produces NH3 emissions at Lambda > 1. This can be reliably accounted for by using four threshold values.

[0035] According to one embodiment of the method, in a first case where the lambda value is greater than the first lambda threshold, the measured value of the nitrogen oxide sensor is determined to be representative of nitrogen oxides in the exhaust gas. In a second case where the lambda gradient is positive and the lambda value is greater than the second lambda threshold and less than the first lambda threshold, the measured value of the nitrogen oxide sensor is determined to be representative of nitrogen oxides in the exhaust gas. In a third case where the lambda gradient is negative and the lambda value is less than the third lambda threshold and greater than the fourth lambda threshold, the measured value of the nitrogen oxide sensor is determined to be representative of ammonia in the exhaust gas. In a fourth case where the lambda value is less than the fourth lambda threshold, the measured value of the nitrogen oxide sensor is determined to be representative of ammonia in the exhaust gas.In all other cases, a final decision valid after the four cases remains active until one of the four cases is fulfilled again.

[0036] According to one embodiment of the method, for the comparison of the lambda value with the four lambda thresholds, an actual lambda value and a predicted lambda value are determined, wherein the actual lambda value is used to determine whether the lambda value is greater than the first lambda threshold or less than the fourth lambda threshold, and the predicted lambda value is used to determine whether the lambda value is greater or less than the second or third lambda threshold.

[0037] The described hysteresis effects can also depend on the speed at which one approaches the lambda = 1 point (i.e., the height of the lambda gradient). The faster one moves from one side (rich or lean) towards lambda = 1, the sooner NOx or NH3 production occurs. To account for this influence, it is advantageous to predict / extrapolate the lambda value several time steps into the future, for example, approximately 200-300 ms.

[0038] The procedure looks like this, for example: 202400918

[0039] 6

[0040] - Calculation of the lambda gradient from the current lambda value at time ii and a lambda value from the past at time ii-x. x is the number of data points one wants to look back at.

[0041] Ideally, one would take 2-3 data points with a sample time of 100 ms, which accordingly also corresponds to 200-300 ms.

[0042] The calculated lambda gradient is then extrapolated into the future to predict the lambda value that will be present in x samples if the current lambda gradient persists. As mentioned above, the recommended extrapolation length is 200-300 ms, or 2-3 samples with a 100 ms sample time. The thresholds are then compared, as mentioned above, not against the current lambda value itself, but against the predicted lambda value. Additionally, the direction of the lambda gradient is taken into account.

[0043] Alternatively, the second and third threshold values ​​can be predetermined depending on the size of the gradient and, for example, stored in a 1D characteristic curve.

[0044] This allows for a more reliable distinction between NOx and NH3.

[0045] According to one embodiment of the method, the determined lambda value is a filtered lambda value.

[0046] It can happen that a supplied lambda signal exhibits significant signal noise. Especially when the catalyst is operating close to lambda = 1, the method is very sensitive to the lambda gradient values ​​and, consequently, to the predicted lambda. This can lead to strong false spikes in the predicted lambda (e.g., a sudden signal increase due to noise can be incorrectly interpreted as a positive lambda gradient).

[0047] To prevent the method from reacting to such short changes in the lambda gradient due to signal noise and falsely detecting an incipient transition from rich to lean (or vice versa), it can be advantageous to filter the noisy lambda signal before calculating the predicted lambda value. 202400918

[0048] 7

[0049] According to one embodiment, to determine the lambda value and the measured value, a lambda measurement signal and a measured value signal are received from the nitrogen oxide sensor, and the two signals are synchronized before further processing. The measured value signal is representative of the concentration of NOx+NH3 molecules.

[0050] It is possible that the two required signals of a NOx sensor (lambda measurement signal and measured value signal) have different response times. In this case, it can be advantageous to delay one of the two signals in time to synchronize their response times.

[0051] The invention is further characterized by a device, wherein the device is configured to carry out the described method or an embodiment of the method.

[0052] The invention is further characterized by the device comprising a vehicle.

[0053] Furthermore, a computer program product is specified, comprising instructions which, when the computer program is executed by a computer, cause it to perform the procedure described herein.

[0054] Furthermore, a computer-readable storage medium is specified on which the computer program described here is stored.

[0055] Exemplary embodiments of the invention are explained in more detail below with reference to the schematic drawings.

[0056] They show:

[0057] Figure 1 shows a flowchart of a program for determining nitrogen oxides and ammonia in exhaust gas.

[0058] Figure 2 shows an exemplary course of ammonia and nitrogen oxides,

[0059] Figure 3 shows a flowchart of another program for determining nitrogen oxides and ammonia in exhaust gas 202400918

[0060] 8

[0061] Figure 4 shows a flowchart of another program for determining nitrogen oxides and ammonia in exhaust gas and

[0062] Figure 5 shows a flowchart of another program for determining nitrogen oxides and ammonia in exhaust gas.

[0063] Elements of the same construction or function are marked with the same reference symbols across all figures.

[0064] Many vehicles are equipped with a catalytic converter, particularly a three-way catalytic converter. These vehicles also have one or more nitrogen oxide sensors, so-called NOx sensors. EU7 legislation makes it mandatory to measure and determine a vehicle's NOx emissions using an in-vehicle NOx sensor. Since an NOx sensor based on the amperometric measurement principle measures the sum of the NOx and NH3 concentrations in the exhaust gas (as it is cross-sensitive to NH3), it is necessary to separate the NOx and NH3 components from the measured total signal.

[0065] Several examples of how to achieve this are shown below. These examples can be combined.

[0066] Figure 1 shows an embodiment of a separation based on lambda information, including consideration of hysteresis effects.

[0067] Figure 3 shows an embodiment of a separation based on NOx / NH3 switching preventer.

[0068] Figure 4 shows an embodiment of a separation based on NOx slip despite grease operation.

[0069] Figure 5 shows an embodiment of a separation based on an NH3 desorption phase / warm-up.

[0070] All these examples can be combined in any way, as none are mutually exclusive. 202400918

[0071] 9

[0072] Figure 1 shows a flowchart of a program for determining nitrogen oxides and ammonia in the exhaust gas of a vehicle.

[0073] The vehicle is equipped with a nitrogen oxide sensor. This sensor is located downstream of a three-way catalytic converter. The nitrogen oxide sensor, also known as a NOx sensor, is based on the amperometric measurement principle.

[0074] The procedure can be carried out, for example, by means of a control device 10.

[0075] The control device 10 comprises, in particular, a processing unit, a program and data memory, and, for example, one or more communication interfaces. The program and data memory and / or the processing unit and / or the communication interfaces can be integrated into a single unit and / or distributed across multiple units. The control device 10 is coupled, in particular, to the nitrogen oxide sensor and the three-way catalytic converter for receiving measurement data from the nitrogen oxide sensor and / or the three-way catalytic converter.

[0076] The control device 10 can also be described as a device for determining nitrogen oxides and ammonia in the exhaust gas.

[0077] The program for determining nitrogen oxides and ammonia in the exhaust gas is stored in particular on the program and data memory of the control device 10.

[0078] The program is started in step S1, in which variables can be initialized if necessary.

[0079] In step S3, a lambda value is determined that is representative of the combustion air ratio of the vehicle.

[0080] The lambda value, for example, is representative of a linear lambda signal. Alternatively, any other signal representative of the oxygen content or lambda value of the exhaust gas can be used. Examples include a binary lambda signal (measured in volts) or an oxygen concentration (measured in ppm or %). 202400918

[0081] 10

[0082] The lambda value is determined based on a lambda signal provided by the nitrogen oxide sensor, which was received in a step (not shown) prior to step S3. Lambda here specifically represents the air-fuel ratio compared to a stoichiometric combustion mixture. Alternatively, it can also be provided by a second nitrogen oxide sensor.

[0083] In step S5, a lambda gradient is determined, which is representative of whether the lambda value has increased or decreased compared to a previous value.

[0084] In step S7, depending on the lambda value and the lambda gradient, it is determined whether a measured value from the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas or whether it is representative of ammonia in the exhaust gas. Furthermore, an information signal is sent, indicating whether the measured value is representative of nitrogen oxides in the exhaust gas or ammonia in the exhaust gas. This information signal is then sent for further processing, for example, to another processing unit or within the device itself.

[0085] Steps S3 and S7 are repeated continuously until the program terminates.

[0086] In step S9, the program is terminated and can be restarted in step S1 if necessary.

[0087] The program offers the following advantages:

[0088] In general, the following relationship applies between the emissions downstream of a three-way catalyst and the lambda condition of the three-way catalyst or downstream of the three-way catalyst:

[0089] When operating the catalytic converter with a superstoichiometric air-fuel mixture (lambda > 1, so-called "lean operation" or "lean mixture"), not all nitrogen oxides from combustion can be broken down, as the required reducing agent (e.g., CO) has already been oxidized beforehand. As a result, a certain concentration of NOx is present downstream of the catalytic converter, which can be detected with a nitrogen oxide sensor. The nitrogen oxide sensor can also be used for 202400918.

[0090] 11 is referred to as a NOx sensor. NOx is used here as a synonym for nitrogen oxides and is a collective term for gaseous oxides of nitrogen, in particular for nitric oxide (NO) and nitrogen dioxide (NO2).

[0091] If the catalyst is operated with a substoichiometric air-fuel mixture (lambda < 1 , so-called “rich operation” or “rich mixture”), not all hydrocarbons and carbon monoxide present can be broken down.

[0092] Additionally, ammonia (NH3) is produced as a side reaction during this rich-fuel operation, resulting in another pollutant. Consequently, the nitrogen oxide sensor downstream of the catalytic converter measures a certain concentration of NH3. However, all NOx emissions are converted to N2 and CO2 because sufficient amounts of fuel gas (CO) are present. Therefore, only NH3 exists downstream of the catalytic converter, and no NOx remains.

[0093] However, this simple approach to distinguishing between NOx and NH3 emissions does not allow for a correct separation between NOx and NH3 under all conditions.

[0094] Firstly, NOx and NH3 formation are subject to certain hysteresis effects, for example, due to the oxygen storage capacity of the catalyst. Secondly, the emission behavior can change with increasing catalyst aging. For these reasons, the simple approach described above for separating NOx and NH3 is insufficient to reliably distinguish between NOx and NH3 emissions.

[0095] In addition to the pure lambda value, the program also uses the lambda gradient to determine whether the sensor is measuring NOx or NH3. In other words, it uses information about the source of the sensor signal, i.e., whether it originates from a lean or rich operating condition. This allows the aforementioned hysteresis effects around lambda = 1 to be taken into account.

[0096] This method allows for a more reliable distinction between NOx and NH3.

[0097] The individual steps are explained in more detail below. 202400918

[0098] 12

[0099] Optionally, four lambda threshold values ​​are provided, and depending on a comparison of the lambda value with the four lambda threshold values ​​and the lambda gradient, it is determined whether a measured value of the nitrogen oxide sensor is representative for nitrogen oxides in the exhaust gas or whether the measured value of the nitrogen oxide sensor is representative for ammonia in the exhaust gas, whereby the first lambda threshold value SW1 of the four lambda threshold values ​​is greater than the third lambda threshold value SW3, the third lambda threshold value SW3 is greater than the second lambda threshold value SW2, and the second lambda threshold value SW2 is greater than the fourth lambda threshold value SW4.

[0100] Alternatively, the second lambda threshold can also be greater than the third lambda threshold.

[0101] The four lambda threshold values ​​are stored, for example, in the control device 10 and were determined through experiments.

[0102] The four lambda threshold values ​​are illustrated in Figure 2.

[0103] Figure 2 shows exemplary trends of ammonia and nitrogen oxides.

[0104] If the lambda value of the system is currently changing from rich to lean, NOx is produced even before passing through the point Lambda = 1. This can be seen in the curve labeled NOx curve.

[0105] If the lambda value of the system is currently moving from lean to rich, NH3 is also produced before passing through the point Lambda = 1. This can be seen in the curve with the reference symbol NH3 curve.

[0106] These hysteresis effects can be covered by the four lambda thresholds and the use of the lambda gradient:

[0107] If lambda is greater than the first lambda threshold SW1, i.e. in very lean operation, then one is certainly in NOx operation.

[0108] If lambda is less than the fourth lambda threshold SW4, i.e., in very rich operation, then NH3 operation is definitely the norm. 202400918

[0109] 13

[0110] The lambda gradient is now taken into account between these two lambda threshold values ​​SW1 and SW4.

[0111] If the lambda value is currently transitioning from rich to lean, the system is on the "NOx curve," and the second lambda threshold value SW2 is used for differentiation. Therefore, if the lambda gradient is positive and the lambda value is greater than the second lambda threshold value SW2 and less than the first lambda threshold value SW1, the measured value from the nitrogen oxide sensor is determined to be representative of nitrogen oxides in the exhaust gas.

[0112] If the lambda value is currently transitioning from lean to rich, the system is on the "NH3 curve," and the third lambda threshold SW3 is used for differentiation. Therefore, if the lambda gradient is negative and the lambda value is less than the third lambda threshold SW3 and greater than the fourth lambda threshold SW4, the reading from the nitrogen oxide sensor is determined to be representative of ammonia in the exhaust gas.

[0113] In all other cases, a final decision valid after the four cases remains active until one of the four cases is fulfilled again.

[0114] Additionally, it can be advantageous to determine and use not only a current lambda value, but also a predicted lambda value. This allows for the optional comparison of the lambda value with the four lambda thresholds, where a current lambda value and a predicted lambda value are determined. The current lambda value is used to determine whether the lambda value is greater than the first lambda threshold SW1 or less than the fourth lambda threshold SW4, while the predicted lambda value is used to determine whether the lambda value is greater than or less than the second or third lambda threshold SW2 and SW3, respectively.

[0115] The described hysteresis effects can also depend on the speed at which one approaches the lambda = 1 point (i.e., the height of the lambda gradient). The faster one moves from one side (rich or lean) towards lambda = 1, the sooner NOx or NH3 production occurs. To account for this influence, it is advantageous to predict / extrapolate the lambda value several time steps into the future, for example, approximately 200-300 ms. 202400918

[0116] 14

[0117] The procedure looks something like this:

[0118] - Calculation of the lambda gradient from the current lambda value at time ii and a lambda value from the past at time ii-x. x is the number of data points one wants to look back at.

[0119] Ideally, one would take 2-3 data points with a sample time of 100 ms, which accordingly also corresponds to 200-300 ms.

[0120] The calculated lambda gradient is then extrapolated into the future to predict the lambda value that will be present in x samples if the current lambda gradient persists. As mentioned above, the recommended extrapolation length is 200-300 ms, or 2-3 samples with a 100 ms sample time. The lambda thresholds are then compared, as mentioned above, not against the current lambda value itself, but against the predicted lambda value. Additionally, the direction of the lambda gradient is taken into account.

[0121] Optionally, the lambda value can be a filtered lambda value. It is possible that a provided lambda signal may exhibit significant signal noise. Especially when the catalyst is operating close to lambda = 1, the method is very sensitive to the lambda gradient values ​​and, consequently, to the predicted lambda. This can lead to strong, false spikes in the predicted lambda (e.g., a sudden signal increase due to noise may be incorrectly interpreted as a positive lambda gradient).

[0122] To prevent the method from reacting to such short changes in the lambda gradient due to signal noise and falsely detecting an incipient transition from rich to lean (or vice versa), it may be advantageous to filter the noisy lambda signal before the calculation of the predicted lambda value takes place.

[0123] Additionally, to determine the lambda value and the measured value, a lambda measurement signal and a measured value signal can optionally be received, which was sent by the nitrogen oxide sensor and the two signals are synchronized before further processing.

[0124] It is possible that the two required signals of a NOx sensor (lambda measurement signal and measured value signal) have different response times. In this case, it may be advantageous to use 202400918

[0125] 15 to delay one of the two signals in time in order to synchronize the response times of both signals.

[0126] Figure 3 shows a flowchart of another program for determining nitrogen oxides and ammonia in the vehicle's exhaust gas.

[0127] The further program can be carried out, for example, using the control device 10 of Figure 1.

[0128] The program and data memory of the control device 10 contains, in particular, the further program for determining nitrogen oxides and ammonia in the exhaust gas.

[0129] The program is started in step S301, in which variables can be initialized if necessary.

[0130] In step S303, measured values ​​from the nitrogen oxide sensor are received. It is then determined whether the measured values ​​show an increasing trend.

[0131] In step S305, if the measured values ​​show an increasing trend, it is determined, depending on a lambda value, whether the measured values ​​of the nitrogen oxide sensor are representative of nitrogen oxides in the exhaust gas or whether the measured values ​​of the nitrogen oxide sensor are representative of ammonia in the exhaust gas. This is done, for example, according to the embodiment shown in Figure 1.

[0132] In step S307, further measured values ​​from the nitrogen oxide sensor are assigned, according to the previous determination, whether the measured values ​​are representative of nitrogen oxides in the exhaust gas or whether they are representative of ammonia in the exhaust gas, nitrogen oxides, or ammonia, until a predefined termination condition is met. Furthermore, an information signal is sent—for further processing in the device or in another processing unit—which includes information on whether the further measured values ​​from the nitrogen oxide sensor are still assigned to nitrogen oxides or ammonia.

[0133] In step S309, the program is terminated and can be restarted in step S301 if necessary. 202400918

[0134] 16

[0135] Especially during short but significant breakthroughs of the lambda condition towards lean or rich (so-called lean or rich excursions), the resulting emission peak (NOx or NH3) can last considerably longer than the lambda breakthrough that caused the emission peak. This means that, for a short time, the correlation between the lambda condition and NOx or NH3 emissions no longer holds (lean = NOx, rich = NH3).

[0136] Under such conditions, massive misinterpretations can occur. For example, a significant lean-run (lambda > 1.03) occurs for a brief moment, say, one second. Prior to this, the system was operating at lambda = 1.000 or even slightly rich at lambda = 0.999. After this one second, the lambda value returns to lambda = 1.000 or possibly already slightly rich again at, for example, lambda = 0.999. However, the NOx peak caused by the brief lean-run typically lasts longer than one second, for example, three seconds. Consequently, starting 1.1 seconds after the lean-run began, when the lambda value is already back at lambda = 1.000 or 0.999, a separation algorithm would incorrectly interpret the still-present NOx peak as NH3 because rich lambda conditions are present.According to this, the algorithm misinterprets the data for a duration of 1.9 seconds, which can lead to a significant miscalculation of emissions.

[0137] To counteract this, the trend of the measured values ​​is taken into account as described above, and measured values ​​from the nitrogen oxide sensor continue to be assigned according to the previous determination of whether the measured values ​​of the nitrogen oxide sensor are representative for nitrogen oxides in the exhaust gas or whether the measured values ​​of the nitrogen oxide sensor are representative for ammonia in the exhaust gas, nitrogen oxides or ammonia, until a predetermined termination condition is met.

[0138] A switching preventer is therefore implemented, which assigns the measured values ​​either to NOx until the termination condition is met, or which assigns measured values ​​to NH3 until the termination condition is met.

[0139] This eliminates misinterpretations during short lambda value changes. 202400918

[0140] 17

[0141] This helps to reliably distinguish whether a measured value from a vehicle's nitrogen oxide sensor is representative of nitrogen oxides or of ammonia.

[0142] The specified termination condition is explained in more detail below.

[0143] The specified termination condition includes, for example, that a trend of the further measured values ​​falls below a specified threshold.

[0144] A suitable termination condition is, in particular, when the measured value is low enough to indicate that the emission peak has passed. This can be achieved either by specifying a fixed threshold, for example, in the range of 100 ppm, or by using a value that is determined based on the emission peak, e.g., factor * peak maximum since activation of the switching preventer (for example, 0.25 * 500 ppm = 125 ppm). The maximum emission peak while the switching preventer is active can be determined and temporarily stored. Combining both conditions is also possible.

[0145] Alternatively or additionally, the specified termination condition includes the requirement that a local minimum was detected during the course of further measurements.

[0146] It is also possible that the measurement signal does not fall below the previously mentioned threshold value because a further signal increase occurs, triggered either by NH3 or NOx. Therefore, another suitable termination condition is that a local minimum is detected during the subsequent measurement sequence. This means that the subsequent second signal increase could potentially be an NH3 peak if a NOx switching preventer is active, or analogous logic for the NH3 switching preventer.

[0147] The detection of a local minimum is achieved, for example, using the following logic:

[0148] - Two samples prior to time ii-2, the concentration signal was larger than the signal at time ii-1 --> concentration signal was decreasing

[0149] - At time ii, the concentration signal is larger than at time ii-1 --> Concentration signal is increasing --> There was a local minimum at time ii-1. 202400918

[0150] 18

[0151] Alternatively or additionally, the specified termination condition includes the requirement that a current lambda value is greater than a specified upper threshold.

[0152] If lambda is very lean, one can be certain that the measured value signal is clearly representative of NOx. Therefore, this can also be used as a termination condition, especially for an NH3 switching preventer.

[0153] Detection is achieved, for example, via a predefined lambda threshold value that is significantly above 1,000, such as the first lambda threshold value SW1 in Figure 2. Alternatively or additionally, a binary lambda signal can be used, for example, if the binary lambda signal is < 50 mV.

[0154] Alternatively or additionally, the specified termination condition includes the requirement that a current lambda value is less than a specified lower threshold.

[0155] If lambda is very rich, one can be certain that the measured value signal is clearly representative of NH3. Therefore, this can also be used as a termination condition, especially for a NOx switching preventer.

[0156] Detection is achieved, for example, via a predefined lambda threshold value that is significantly below 1,000, such as the fourth lambda threshold SW4 in Figure 2. Alternatively or additionally, a binary lambda signal can be used, for example, if the binary lambda signal is > 800 mV.

[0157] The aforementioned termination conditions can be combined in any way.

[0158] To activate the switching preventer, more than one condition can be used to make it even more secure. For example, it can be activated only if at least one measured value was significant enough, specifically if at least one measured value is greater than a minimum threshold. The minimum threshold is, for example, in the range of 100-150 ppm.

[0159] Another condition could be, for example, that no other switching preventer is currently active, e.g., if a calculation is running (202400918).

[0160] 19 would actually activate a NOx switching preventer, but an NH3 switching preventer is currently active, or vice versa.

[0161] Another condition could be, for example, that the measured values ​​show an increasing trend.

[0162] Another condition could be, for example, that NOx has been detected to activate the NOx switching preventer, or that NH3 has been detected to activate the NH3 switching preventer. Detection is carried out, for example, based on the program shown in Figure 1.

[0163] Figure 4 shows an embodiment of a separation based on NOx slip despite grease operation.

[0164] The further program of Figure 4 can be carried out, for example, using the control device 10 of Figure 1.

[0165] The program and data memory of the control device 10 contains, in particular, the further program for determining nitrogen oxides and ammonia in the exhaust gas.

[0166] The program is started in step S401, in which variables can be initialized if necessary.

[0167] In step S403, a lambda value is determined that is representative of the combustion air ratio of the vehicle.

[0168] The lambda value is determined based on a lambda signal, which is provided, for example, by the nitrogen oxide sensor. Alternatively, it can also be provided by a second nitrogen oxide sensor.

[0169] The lambda value, for example, is representative of a linear lambda signal. Alternatively, any other signal representative of the oxygen content or lambda value of the exhaust gas can be used. Examples include a binary lambda signal (measured in volts) or an oxygen concentration (measured in ppm or %). 202400918

[0170] 20

[0171] In step S405, depending on the lambda value, it is determined whether a measured value from the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas or whether the measured value from the nitrogen oxide sensor is representative of ammonia in the exhaust gas. This is done, for example, according to the embodiment shown in Figure 1.

[0172] In step S407, if it has been determined that the measured value of the nitrogen oxide sensor is representative of ammonia in the exhaust gas, a mass flow of the exhaust gas is determined; depending on the mass flow and the lambda value, a nitrogen oxide content factor of the measured value is determined.

[0173] In step S409, depending on the nitrogen oxide content factor, a nitrogen oxide content of the measured value and an ammonia content of the measured value are determined and - for further processing in the device or in another computing unit - an information signal is sent which includes information about the nitrogen oxide content of the measured value and the ammonia content of the measured value.

[0174] In step S411, the program is terminated and can be restarted in step S401 if necessary.

[0175] The described program takes into account a significant property of the catalyst. Especially at high mass flows, it is possible that NOx, in addition to NH3, may be present downstream of the catalyst under rich conditions. This so-called "NOx slippage in rich conditions" can be explained by the fact that at high mass flows, the residence time of the NOx molecules, which are produced during the combustion process in the combustion chambers, is so short in the catalyst that not all NOx molecules can be converted during this brief period. A portion of the NOx molecules, so to speak, slips through the catalyst unconverted and is measured by the nitrogen oxide sensor.

[0176] In the case of NOx slippage in fats, a suitable proportional distribution between NOx and NH3 concentrations must therefore be made in order to correctly determine NOx and NH3.

[0177] This is made possible by the method using the nitrogen oxide fraction factor, which indicates, for example, the proportion of NOx in the measured NOx sensor signal (e.g., a nitrogen oxide fraction factor of 0.3 means that 30% of the signal is NOx and 70% is NH3). 202400918

[0178] 21

[0179] After determining the nitrogen oxide fraction factor, the measured signal can thus be separated into NOx and NH3.

[0180] This is done, for example, according to the following formulas, taking into account the NH3 cross-sensitivity of the nitrogen oxide sensor:

[0181] NOx after separation = measured value * nitrogen oxide fraction factor / (nitrogen oxide fraction factor + (1 - nitrogen oxide fraction factor) * cross-sensitivity factor);

[0182] NH3 after separation = (measured value - NOx after separation) * 1 / cross-sensitivity parameter

[0183] Alternatively, the NOx concentration can be determined not as a percentage of the total measured signal, but in absolute values, i.e., the NOx concentration in ppm. The NH3 concentration is then the measured sum signal – NOx concentration. Here, too, the cross-sensitivity to NH3 is specifically taken into account.

[0184] Thus, the nitrogen oxide content factor can be used to reliably distinguish whether a measured value from the vehicle's nitrogen oxide sensor is representative of nitrogen oxides or of ammonia.

[0185] For example, a temperature representative of the catalyst temperature is also determined, and depending on the temperature, mass flow and lambda value, a nitrogen oxide fraction factor of the measured value is determined.

[0186] The temperature is determined, for example, based on the temperature of the exhaust gas, a component temperature, a pipe wall temperature, or similar factors. It can be measured or based on a model value.

[0187] In addition to the mass flow, NOx slip in rich conditions also depends on the gas temperature downstream of the catalyst and the prevailing lambda value.

[0188] Thus, taking into account the temperature downstream of the catalyst, or the temperature that is representative of the temperature of the catalyst, the nitrogen oxide fraction factor can be determined even more precisely.

[0189] The nitrogen oxide fraction factor can be determined in various ways. 202400918

[0190] 22

[0191] For example, the nitrogen oxide fraction factor is determined using a 3D map, where the 3D map shows nitrogen oxide fraction factors via temperature, lambda and mass flow.

[0192] This makes it possible to determine the nitrogen oxide fraction factors using a single 3D map. The 3D map is stored, for example, in the memory of a control device and was created during a test phase.

[0193] Alternatively or additionally, the nitrogen oxide fraction factor is determined using a first 2D map and a second 2D map, whereby the first 2D map shows nitrogen oxide fraction factors via temperature and mass flow and the second 2D map shows nitrogen oxide fraction factors via lambda and mass flow.

[0194] 2D maps can potentially be stored more easily in control devices than 3D maps. Therefore, using 2D maps can lead to more efficient data acquisition. Other combinations of 2D maps are also possible, such as a 2D map of nitrogen oxide fraction factors based on lambda and temperature, etc.

[0195] Alternatively or additionally, a catalytic converter age can also be used to determine the nitrogen oxide content factor and / or the nitrogen oxide content.

[0196] For example, a catalyst age is determined that is representative of the aging state of a catalyst in the vehicle, and depending on the catalyst age, the temperature, the mass flow and the lambda value, a nitrogen oxide content of the measured value is determined.

[0197] The age of the catalytic converter can be determined, for example, via onboard diagnostics such as the so-called "Oxygen Storage Capacity (OSC) determination" or by measuring engine operating time. Any other method for determining the age of the catalytic converter can also be used.

[0198] For example, the nitrogen oxide fraction factor is determined using the 3D map or the 2D maps and a subsequent correction depending on the catalyst age, or several 3D maps or several first and second 2D maps for catalysts of varying ages are used to determine the nitrogen oxide fraction factor. 202400918

[0199] 23

[0200] The age of the catalytic converter can be taken into account in various ways. The simplest method is to correct the previously used calculation, for example, the determination based on the 3D map, using a correction factor according to the catalytic converter's age. Alternatively, an additional 1D characteristic curve can also be used.

[0201] Alternatively, the aforementioned 3D or 2D maps for differently aged catalysts are stored, and interpolation is performed between the different maps for the different aging states (e.g. linear interpolation, but other interpolation methods or even weighting functions can also be stored).

[0202] Thus, using the program in Figure 4, especially in combination with the program in Figure 1 and / or the program in Figure 3, it is possible to distinguish very accurately between NOx and NH3.

[0203] Figure 5 shows an embodiment of a separation based on an NH3 desorption / heating phase.

[0204] The further program of Figure 5 can be carried out, for example, using the control device 10 of Figure 1.

[0205] The program and data memory of the control device 10 contains, in particular, the further program for determining nitrogen oxides and ammonia in the exhaust gas.

[0206] The program is started in step S501, in which variables can be initialized if necessary.

[0207] In step S503, a temperature signal is received and, depending on the temperature signal, temperature information for a catalytic converter of the vehicle is determined.

[0208] In step S505, depending on the temperature information, it is determined whether the catalyst's activation temperature has been reached, and until the activation temperature is reached, a measured value from the nitrogen oxide sensor is assigned to nitrogen oxides. 202400918

[0209] 24

[0210] In step S507, after reaching the activation temperature, an ammonia desorption phase is determined depending on further temperature information, and during the ammonia desorption phase, and depending on the ammonia desorption phase, the measured value is assigned to nitrogen oxides and / or ammonia.

[0211] In step S509, the program is terminated and can be restarted in step S501 if necessary.

[0212] The program in Figure 5 is executed in particular until the ammonia desorption phase is completed; afterwards, the program in Figure 1 can be executed in combination with the programs in Figures 3 and 4.

[0213] A catalyst can only convert emissions above a certain activation temperature, which is approximately in the range of 200-250°C. Since no conversion takes place before this light-off temperature is reached, and combustion in the cylinders produces only NOx and no NH3, the nitrogen oxide sensor signal measured downstream of the catalyst before the light-off temperature is reached must be caused exclusively by NOx molecules.

[0214] The procedure of the program shown in Figure 5 takes this into account and thus specifies that before reaching the Light-Off threshold, the entire measured sensor signal is interpreted as NOx.

[0215] This method allows for a more reliable distinction between NOx and NH3.

[0216] The light-off threshold can be determined based on the temperature information for the catalyst, especially for the first catalyst if two are installed (closed-coupled and underfloor).

[0217] Any temperature suitable for describing the catalyst light-off state can be selected as the temperature information. Accordingly, the temperature signal and temperature information are representative of, for example:

[0218] - Gas temperature upstream or downstream of the catalyst,

[0219] - Component temperature of the catalyst, 202400918

[0220] 25

[0221] - Pipe jacket temperature of the catalyst canning,

[0222] - Temperature in the catalyst after the first few centimeters (e.g. 5cm) of the catalyst and / or

[0223] - Heat integrals based on the aforementioned temperature quantities - any combinations of the aforementioned variants.

[0224] The temperature signal can be provided in particular by a suitable temperature sensor or by a computing unit that determines the aforementioned variants.

[0225] Additionally, a light-off threshold can be selected that detects the beginning of NOx conversion (if NOx conversion > threshold (e.g. 90%)).

[0226] Another phenomenon taken into account by the program in Figure 5 is the ammonia desorption phase.

[0227] A particular phenomenon during the heating phase of a catalyst is the absorption of NH3 in liquid water within the humid exhaust system and on the surfaces of the catalysts / particulate filters during this heating process. This phenomenon can also be referred to as the ammonia desorption phase and is taken into account here.

[0228] This method allows for a more reliable distinction between NOx and NH3.

[0229] The formation of the NH3 desorption phase involves the following successive phases:

[0230] 1. A front catalyst exceeds light-off. This means that under lean conditions NOx is converted and under rich conditions NH3 can be formed and emitted.

[0231] 2. The NH3 molecules produced in the first catalyst "fly" out of the first catalyst.

[0232] 3. Further downstream, the exhaust system is still partially unheated and contains liquid water as long as the local temperature at the points where water has accumulated remains below 100°C. Liquid water collects particularly on the underside of the catalytic converter and pipes because gravity causes it to accumulate there more readily, and in the spaces between the porous washcoat of the catalytic converter coating. 202400918

[0233] 26

[0234] 4. The NH3 molecules pass downstream of the first catalyst through some of these moist / liquid water accumulations and the NH3 is absorbed in the liquid water.

[0235] 5. As a result, the NH3 molecules produced at rich lambda values ​​do not initially reach a nitrogen oxide sensor located downstream of an underfloor catalyst.

[0236] 6. The adsorbed NH3 is only released again when the liquid water in which the NH3 was dissolved evaporates, i.e., when the local temperature exceeds 100°C. This process is called the NH3 desorption phase and can extend over several minutes until the entire exhaust system, up to the position of the nitrogen oxide sensor, has been completely heated to > 100°C and no liquid water remains in the exhaust system.

[0237] The described behavior is similar to the behavior of a NOx.

[0238] Storage catalyst that stores NOx molecules at low temperatures and releases them again in a delayed manner once a temperature threshold is exceeded.

[0239] The desorption of NH3 molecules is therefore purely dependent on the temperature of the exhaust system and is not linked to the lambda condition of the underfloor catalyst.

[0240] Consequently, a separation algorithm based on lambda cannot assign this desorbed NH3 because the connection between emission formation and the lambda condition is missing.

[0241] It is therefore possible that, during the NH3 desorption phase, the subfloor catalyst is operating at a lean mixture. Consequently, the separation algorithm would incorrectly identify the desorbed NH3 as NOx. For this reason, an additional function, as described here, is necessary to accurately model the NH3 desorption phase.

[0242] The detection of the start of the NH3 desorption phase is designed in such a way that it recognizes when the evaporation of liquid water (and thus NH3 desorption) begins downstream of the first catalyst. Strictly speaking, it is not necessarily the start of desorption that needs to be described, but rather the point in time from which so much NH3 is released that it can no longer be completely reabsorbed further downstream in the liquid water. The first molecules, which occur immediately downstream of the first catalyst, are therefore...

[0243] 27

[0244] TWC are desorbed, namely they are absorbed in the liquid water that is still present downstream, until this water is in turn evaporated.

[0245] The start of the NH3 desorption phase is detected based on the temperature information downstream of the first catalyst. This occurs when the liquid water evaporates at the first points after the first catalyst and the stored NH3 is released.

[0246] Suitable temperature information could include:

[0247] - Gas temperatures,

[0248] - Component temperatures (e.g., of the first catalyst),

[0249] - Pipe jacket temperatures (ideally placed at the points with the most water accumulation),

[0250] - Heat integrals based on the mentioned temperatures,

[0251] - Any combination of the aforementioned information.

[0252] Alternatively or additionally, a dew point model can be used, which describes the absence of dew points (= evaporation of liquid water) at the position after the first catalyst.

[0253] The end of the NH3 desorption phase is characterized by the fact that any liquid water upstream of the nitrogen oxide sensor position has evaporated and therefore no more NH3 is desorbed into the gas system.

[0254] The detection of the end of the NH3 desorption phase must be designed in such a way as to detect when all liquid water upstream of the nitrogen oxide sensor position has evaporated.

[0255] This can again be described using the information mentioned above.

[0256] Additionally, an optional "NH3 Desorption finished" and a "Water dew point reset" flag can be introduced to distinguish whether, for example, the program of Figure 1 or the program of Figure 5 should be executed.

[0257] "NH3 desorption finished": This means that the system detects when the entire exhaust system, up to the position of the nitrogen oxide sensor, is free of liquid water and the NH3 desorption phase is therefore complete. 202400918

[0258] 28

[0259] "Water dew point reset": To reset this "NH3 desorption finished" flag, the exhaust system (or at least parts of it) must again drop below the dew point temperature of water before the next

[0260] The NH3 desorption phase can be detected. This is because once the system is hot and remains above the water dew point, no further NH3 adsorption and desorption can take place in liquid water.

[0261] Both flags can also be described using the information mentioned above, for example.

[0262] The assignment of the measured value to NOx and / or NH3 during the desorption phase can be done in different ways.

[0263] For example, for the assignment, a curve of the measured value is filtered and low-frequency components are assigned to ammonia and high-frequency components to nitrogen oxides.

[0264] If an NH3 desorption phase is detected, a large part of the measured nitrogen oxide sensor signal (sum of NOx+NH3) is caused by NH3 molecules, since the measured nitrogen oxide sensor signal is mainly caused by the desorbed NH3 molecules.

[0265] However, it can happen that NOx and NH3 are predominant simultaneously. The NH3 originates from NH3 desorption, while the NOx is produced because the underfloor catalyst is currently operating at a lean mixture. The separation between the NOx, caused by the lean mixture, and the NH3, resulting from desorption, can be achieved through filtration.

[0266] A fundamental difference between the dynamic behavior of the NH3 desorption peak and a NOx peak due to lean operation is that an NH3 desorption peak has low dynamics and can extend over several seconds to minutes.

[0267] The intermittent, superimposed NOx peaks, on the other hand, occur primarily only briefly (a few seconds), as a robust, well-implemented control system counteracts the lean lambda deviation and regulates to a lambda value close to 1000. 202400918

[0268] 29

[0269] This difference in timing is exploited here. The summed signal is filtered so that only the low-frequency components remain (e.g., using a low-pass filter). This low-frequency component is attributed to NH3. The remaining high-frequency component of the summed signal is then consequently attributed to NOx.

[0270] Additionally, the lambda condition of the underfloor catalyst described above can be taken into account, particularly with regard to the high-frequency components, especially considering the programs shown in Figures 1, 3, and 4, to verify whether it is a NOx peak and thus increase the robustness of the separation. If the lambda condition is lean, the high-frequency component is interpreted as NOx; if the lambda condition is rich, the high-frequency component is still attributed to NH3.

[0271] Alternatively or additionally to the NOx-NH3 split by filtering the sum signal, a scalar fraction factor (e.g., 0.8, which corresponds to 80% NOx and 20% NH3) can be selected, especially if lean lambda conditions have been detected. In this case, a fraction factor is determined, and depending on the fraction factor, a portion of the measured value is assigned to ammonia and a portion to nitrogen oxides.

[0272] Different values ​​can be assigned to the scalar proportion factor, depending on certain influencing factors. For example, it can be selected based on the magnitude of the measured value and / or the prevailing lambda value.

[0273] As described, the programs described can be combined as desired, since each offers advantages for individual phenomena in nitrogen oxide sensor measurement. Thus, all or parts of the programs can also be combined into a single overall program to specifically counteract a particular combination of the presented phenomena. 202400918

[0274] 30

[0275] Reference symbol list

[0276] S1-S9 steps

[0277] S301-S309 steps S401-S411 steps

[0278] S501-S509 steps

[0279] 10 Control device

[0280] SW1 First lambda threshold

[0281] SW2 Second lambda threshold SW3 Third lambda threshold

[0282] SW4 Fourth Lambda Threshold

[0283] NH3 trend NH3 trend

[0284] NOx trend NOx trend

Claims

202400918 31 Patent claims 1. Method for determining nitrogen oxides and ammonia in the exhaust gas of a vehicle, wherein the vehicle has a nitrogen oxide sensor, wherein a lambda signal is received in the method, - depending on the lambda signal, a lambda value is determined that is representative of the combustion air ratio of the vehicle, - a lambda gradient is determined that is representative of whether the lambda value has increased or decreased compared to a previous value, - depending on the lambda value and the lambda gradient, it is determined whether a measured value from the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas or whether the measured value from the nitrogen oxide sensor is representative of ammonia in the exhaust gas, and - an information signal is sent which includes information on whether the measured value of the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas or whether the measured value of the nitrogen oxide sensor is representative of ammonia in the exhaust gas.

2. Method according to claim 1, wherein four lambda threshold values ​​are provided and, depending on a comparison of the lambda value with the four lambda threshold values ​​and the lambda gradient, it is determined whether a measured value of the nitrogen oxide sensor is representative for nitrogen oxides in the exhaust gas or whether the measured value of the nitrogen oxide sensor is representative for ammonia in the exhaust gas, wherein the first lambda threshold value (SW1) of the four lambda threshold values ​​is greater than the third lambda threshold value (SW3), the third lambda threshold value (SW3) is greater than the second lambda threshold value (SW2), and the second lambda threshold value (SW2) is greater than the fourth lambda threshold value (SW4).

3. The method according to claim 2, wherein - in a first case, where the lambda value is greater than the first lambda threshold value (SW1 ), it is determined that the measured value of the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas, - in a second case, where the lambda gradient is positive and the lambda value is greater than the second lambda threshold (SW2) and less than the first lambda threshold (SW1), it is determined that the measured value of the nitrogen oxide sensor is representative of nitrogen oxides in the exhaust gas, 202400918 32 - in a third case, where the lambda gradient is negative and the lambda value is smaller than the third lambda threshold (SW3) and larger than the fourth lambda threshold (SW4), it is determined that the measured value of the nitrogen oxide sensor is representative of ammonia in the exhaust gas, - in a fourth case, where the lambda value is smaller than the fourth lambda threshold (SW4), it is determined that the measured value of the nitrogen oxide sensor is representative of ammonia in the exhaust gas, - in all other cases, a final valid decision after the four cases remains active until one of the four cases is fulfilled again.

4. Method according to claim 2 or 3, wherein for the comparison of the lambda value with the four lambda thresholds, an actual lambda value and a predicted lambda value are determined, wherein the actual lambda value is used to determine whether the lambda value is greater than the first lambda threshold (SW1) or less than the fourth lambda threshold (SW4), and the predicted lambda value is used to determine whether the lambda value is greater or less than the second or third lambda threshold (SW2, SW3).

5. Method according to any of the preceding claims, wherein the determined lambda value is a filtered lambda value.

6. Method according to one of the preceding claims, wherein, for determining the lambda value and the measured value, a lambda measurement signal and a measured value signal are received which was sent by the nitrogen oxide sensor and the two signals are synchronized before further processing.

7. Device, wherein the device is configured to perform the method according to any of the preceding claims.

8. Vehicle comprising the device according to claim 7.

9. Computer program product comprising instructions which, when the program is executed by a computer, cause it to execute the method according to any one of claims 1 to 6.

10. Computer-readable storage medium comprising instructions which, when executed by a computer, cause the computer to execute the method according to any one of claims 1 to 6.

Images