Method, computing unit and computer program for operating an internal combustion engine with at least one catalyst

Abstract

Method (200) for operating an internal combustion engine (110) with at least one catalyst (122), wherein control interventions of a lambda control for regulating an exhaust gas composition of the internal combustion engine are deactivated, comprising: determining a current exhaust gas composition upstream of the at least one catalyst (122), determining a current oxygen level of the at least one catalyst (122) based on the determined current exhaust gas composition, determining (210) a planned control intervention on a composition of an air-fuel mixture supplied to the internal combustion engine (110) based on the determined current oxygen level of the at least one catalyst, determining a current exhaust gas composition (123) downstream of the at least one catalyst (122), determining a future,exhaust gas composition (123) resulting from an air-fuel mixture already supplied to the internal combustion engine (110) downstream of at least one catalyst (122), and reactivating the lambda control and determining (260) a control intervention to be carried out depending on the planned control intervention and the current exhaust gas composition (123) downstream of at least one catalyst (122), and / or depending on the planned control intervention and the future exhaust gas composition (123).

Description

[0001] The present invention relates to a method for operating an internal combustion engine with at least one catalyst, as well as a computing unit and a computer program for carrying it out. Background of the invention

[0002] Incomplete combustion of the air-fuel mixture in a gasoline engine produces, in addition to nitrogen (N2), carbon dioxide (CO2) and water (H2O), a variety of combustion products, including hydrocarbons (HC), carbon monoxide (CO) and nitrogen oxides (NOx), which are emitted. x ) are legally limited. The applicable emission limits for motor vehicles can typically only be met with catalytic exhaust aftertreatment. By using a three-way catalytic converter, the aforementioned pollutant components can be converted.

[0003] A simultaneously high conversion rate for HC, CO and NO xIn three-way catalysts, this is only achieved in a narrow range around the stoichiometric operating point (Lambda = 1), the so-called "catalyst window".

[0004] To operate the catalytic converter within its operating window, a lambda control system is typically used, based on signals from lambda sensors both upstream and downstream of the catalyst. For controlling the lambda value upstream of the catalyst, the oxygen content of the exhaust gas before the catalyst is measured with a lambda sensor. Depending on this measurement, the control system corrects the fuel quantity from the mixture feedforward. For more precise control, the exhaust gas downstream of the catalyst is also analyzed with another lambda sensor. This signal is used for a reference control system that is superimposed on the lambda control upstream of the catalyst. A switching lambda sensor is generally used downstream of the catalyst; this sensor has a very steep characteristic curve at lambda = 1 and can therefore indicate lambda = 1 very accurately.

[0005] In addition to the control system, which generally only compensates for small deviations from Lambda = 1 and is designed to be comparatively slow, a Lambda feedforward control can be used after large deviations from Lambda = 1 to quickly reach the catalyst window again, e.g. after phases with overrun fuel cut-off ("catalyst removal").

[0006] Such control concepts have the disadvantage that they only detect a departure from the catalyst window late, based on the voltage of the switching lambda sensor behind the catalyst.

[0007] An alternative to controlling a three-way catalytic converter based on the signal from a lambda sensor downstream of the converter is controlling the mean oxygen level of the converter. Since this mean level cannot be measured, it can only be modeled. A corresponding model-based control of the level of a three-way catalytic converter is described in DE 10 2016 222 418 A1. A feedforward control for a model-based control of the level of a three-way catalytic converter is described in DE 10 2018 208 683 A1, and a model-based prediction of the feedforward lambda value required when restarting after a phase with inactive control intervention is described in DE 10 2018 217 307 A1. Disclosure of the invention

[0008] According to the invention, a method for operating an internal combustion engine with at least one catalyst, as well as a computing unit and a computer program for carrying it out, are proposed, comprising the features of the independent claims. Advantageous embodiments are the subject of the dependent claims and the following description.

[0009] A method according to the invention for operating an internal combustion engine with at least one catalyst, wherein control interventions of a lambda control for regulating an exhaust gas composition of the internal combustion engine are deactivated, comprises determining a current exhaust gas composition upstream of the at least one catalyst, determining a current oxygen level of the at least one catalyst based on the determined current exhaust gas composition, determining a current exhaust gas composition downstream of the at least one catalyst, determining a planned control intervention on a composition of an air-fuel mixture supplied to the internal combustion engine based on the determined current oxygen level of the at least one catalyst, determining a current exhaust gas composition downstream of the at least one catalyst, determining a future,The system uses the exhaust gas composition downstream of at least one catalyst, resulting from an air-fuel mixture already supplied to the internal combustion engine, to reactivate the lambda control and determine a control intervention to be performed, depending on the planned control intervention and the current exhaust gas composition downstream of at least one catalyst, and / or depending on the planned control intervention and the future exhaust gas composition. This allows any error in the planned control intervention, which may be based, for example, on imprecise model assumptions, to be corrected in order to reinitialize the lambda control with as few adverse effects on emissions behavior as possible after an inactive control intervention. In such situations, a miscalculation of the oxygen level is particularly likely, and thus also a miscalculation of the planned control intervention.

[0010] Determining the future exhaust gas composition downstream of the at least one catalyst, resulting from an air-fuel mixture already supplied to the internal combustion engine, preferably comprises determining the exhaust gas composition downstream of the at least one catalyst for a period whose duration corresponds to the flow time of the exhaust gas from the internal combustion engine as it flows through a flow path from the internal combustion engine to downstream of the at least one catalyst. This ensures that situations in which the exhaust gas composition downstream of the catalyst cannot yet reflect any changes in the air-fuel mixture due to an excessively long dead time or gas transit time are correctly taken into account when determining the control intervention.

[0011] In particular, determining the planned intervention can involve categorization, especially into one or more categories such as weight loss, weight gain, and stoichiometric requirements. This facilitates the assignment of relevant events to the temporal development.

[0012] In advantageous embodiments of the method, a control intervention opposite to the planned control intervention is specified if the current exhaust gas composition corresponds to an exhaust gas composition that can be generated by the planned control intervention, and simultaneously the future exhaust gas composition does not already correspond to an exhaust gas composition that can be generated by the planned control intervention. Thus, for example, if a lean air-fuel mixture is planned based on the calculated fill level, but the current exhaust gas composition downstream of the catalyst is already lean and the future exhaust gas composition downstream of the catalyst does not suggest a rich exhaust gas (or is also lean), a rich air-fuel mixture is specified.

[0013] Alternatively or additionally, a corresponding adjustment can be specified if the current exhaust gas composition corresponds to an exhaust gas composition resulting from an opposing adjustment. For example, if a lean air-fuel mixture is planned based on the calculated fill level and the current exhaust gas composition downstream of the catalytic converter is rich, a lean air-fuel mixture will be specified.

[0014] Alternatively or additionally, a corresponding adjustment to be carried out can be specified if the future exhaust gas composition already corresponds to an exhaust gas composition that can be produced by the planned adjustment. For example, if a lean air-fuel mixture is planned based on the calculated fill level and the future exhaust gas composition downstream of the catalytic converter is rich, a lean air-fuel mixture is specified.

[0015] The three options mentioned above can prevent a misjudgment of the planned intervention from leading to worsened emission behavior.

[0016] During periods without active control intervention, one or more of the following functions are performed: engine overrun or coasting, component protection, and (partial) combustion chamber shutdown. In such situations, the oxygen level in the catalyst can change significantly, making these changes particularly relevant to emissions behavior when the lambda control is reactivated.

[0017] The current and / or future exhaust gas composition is preferably determined using a sensor, in particular one or more lambda sensors. This is a particularly robust method, and such sensors are already required by law, so there are no additional costs involved.

[0018] A computing unit according to the invention, e.g. a control unit of a motor vehicle, is, in particular in terms of programming, equipped to carry out a method according to the invention.

[0019] Implementing a method according to the invention in the form of a computer program or computer program product with program code for carrying out all method steps is also advantageous, as this incurs particularly low costs, especially if an executing control unit is already available for other tasks. Finally, a machine-readable storage medium is provided with a computer program stored on it as described above. Suitable storage media or data carriers for providing the computer program are, in particular, magnetic, optical, and electrical storage media, such as hard drives, flash memory, EEPROMs, DVDs, etc. Downloading a program via computer networks (Internet, intranet, etc.) is also possible. Such a download can be wired or wireless (e.g., via a WLAN network, a 3G, 4G, 5G, or 6G connection, etc.).

[0020] Further advantages and embodiments of the invention will become apparent from the description and the accompanying drawing.

[0021] The invention is schematically illustrated in the drawing using an exemplary embodiment and is described below with reference to the drawing. List of characters Fig. Figure 1 schematically shows a vehicle with an internal combustion engine and a catalytic converter, as it can be used within the scope of the present invention. Fig. Figure 2 shows an advantageous embodiment of a method according to the invention in the form of a highly simplified flowchart. embodiment(s) of the invention

[0022] In Fig. Figure 1 is a vehicle, such as can be used within the scope of the invention, schematically depicted and collectively designated by 100. The vehicle 100 comprises an internal combustion engine 110, here for example with six indicated cylinders, an exhaust system 120 which has several cleaning components 122, 124, e.g., catalysts and / or particulate filters, and a computing unit 130, which is configured to control the internal combustion engine 110 and the exhaust system 120 and is connected to them via a data transmission link. Furthermore, in the illustrated example, the computing unit 130 is connected via a data transmission link to sensors 112, 121, 123, 127, which detect operating parameters of the internal combustion engine 110 and / or the exhaust system 120. It is understood that further sensors may be present, which are not shown.

[0023] In the example shown here, the computing unit 130 includes a data storage unit 132, in which, for example, calculation instructions and / or parameters (e.g., threshold values, characteristic values ​​of the internal combustion engine 110 and / or the exhaust system 120, etc.) can be stored.

[0024] The internal combustion engine 110 drives wheels 140 and can also be driven by the wheels in certain operating phases (e.g. so-called overrun mode)

[0025] In Fig. 2 is an advantageous embodiment of a method according to the invention, schematically represented in the form of a flowchart and generally designated by 200.

[0026] References to vehicle components or parts thereof used in the description of procedure 200 refer in particular to the component described in Fig. 1 vehicle shown 100.

[0027] Method 200 can be used in particular in connection with model-based lambda control of the operation of the internal combustion engine 110. Such model-based lambda control can, for example, include the following sub-functionalities: - Route model - Pre-filter - Feedforward control - Level regulator - Adaptation, whereby procedure 200 particularly concerns feedforward control.

[0028] Since the oxygen level of the catalyst 122 cannot be measured, it is modeled using a linear model. The signal from the lambda sensor 121 upstream of the catalyst 122 is converted into one or more input variables for a catalyst level model. In other words, the current exhaust gas composition upstream of the catalyst 122 is determined, and the current oxygen level of the catalyst 122 is calculated based on this determined exhaust gas composition.

[0029] It is advantageous to convert the signal from lambda sensor 121 into the concentration of one or more exhaust gas components. For example, it is advantageous to convert the lambda value into the concentrations of oxygen (O2), carbon monoxide (CO), hydrogen (H2) and hydrocarbons (HC) upstream of the catalyst 122 using an inlet emission model.

[0030] Using the parameters calculated by the input emission model and optionally additional input parameters (e.g., exhaust gas or catalyst temperatures, exhaust gas mass flow, and the current maximum oxygen storage capacity of the catalyst 122), at least one catalyst fill level is modeled. To represent filling and emptying processes more realistically, the catalyst 122 is preferably divided into several (axial) zones, and the concentrations of the individual exhaust gas components are determined for each of these zones using reaction kinetics. These concentrations can then be converted into a fill level for each zone, preferably the oxygen fill level normalized to the current maximum oxygen storage capacity. The fill levels of individual or all zones can be combined into an overall fill level, reflecting the state of the catalyst 122, using suitable weighting.For example, in the simplest case, the fill levels of all zones can be weighted equally to determine an average fill level. However, with suitable weighting, it can also be taken into account that the fill level in a relatively small area at the outlet of catalyst 122 is crucial for the current exhaust gas composition downstream of catalyst 122, while the fill level in the upstream volume and its development are crucial for the development of the fill level in this small area at the outlet of catalyst 122. For the sake of simplicity, an average oxygen fill level is assumed in the following.

[0031] The concentrations of the individual exhaust gas components at the outlet of the catalyst 122, calculated using the catalyst model, are converted into a signal for the adaptation of the system model. This signal can then be compared with the signal from an exhaust gas sensor 123 located downstream of the catalyst. Preferably, the lambda value downstream of the catalyst 122 is modeled.

[0032] The feedforward control 200 can, for example, be essentially designed as an inversion of the plant model. This has the advantage that the controller only needs to intervene if the fill level of the catalyst 122, modeled using the plant model, deviates from a target fill level trajectory calculated by the feedforward control 200. While the plant model converts the input lambda value upstream of the catalyst 122 into an average oxygen fill level of the catalyst 122, the feedforward control 200 converts the average target oxygen fill level into a corresponding target lambda value upstream of the catalyst 122. The target fill level trajectory takes into account that the permissible lambda adjustment range is limited for various reasons, such as emission minimization and drivability. Therefore, the average target oxygen fill level cannot be reached arbitrarily quickly.

[0033] Preferably, the process model is inverted analytically. However, catalyst 122 is a complex, nonlinear process with time-varying process parameters, which can generally only be represented by a system of nonlinear differential equations. This typically leads to the system of equations for the inverted process model not being solvable analytically. Therefore, as an alternative to the analytical inversion of the process model, a numerical inversion of the process model is also possible.

[0034] When the actuator control is inactive (i.e., when lambda control is deactivated), the feedforward control system observes the input lambda measured by the lambda sensor 121 upstream of the catalyst 122, but cannot actively influence it. This observation is performed so that the feedforward control system can immediately specify an optimal trajectory when the actuator control is reactivated.

[0035] When the variable valve timing is reactivated, a residual amount of exhaust gas remains between the combustion chamber of the internal combustion engine 110 and the lambda sensor 121 upstream of the catalyst 122. This residual amount cannot be observed by the pilot control system, as the pilot control system must already specify a pilot lambda value and a corresponding trajectory for the oxygen level. To achieve optimal activation and control behavior upon reactivation of the variable valve timing, the unobservable residual amount of exhaust gas is taken into account. This is done by predicting the change in the oxygen level that will occur during the transport of the residual exhaust gas from the combustion chamber to the lambda sensor 121 upstream of the catalyst 122 using the linear model in a prediction step 210 of the procedure 200.The feedforward control uses the predicted oxygen level to reinitialize the trajectory at the time of resumption after a phase with inactive control intervention and calculates the corresponding feedforward lambda.

[0036] The mean oxygen level, modeled using the system model, is adjusted to a setpoint that minimizes the probability of lean and rich exhaust gas breakthroughs through the catalyst 122, thus resulting in minimal emissions. The setpoint is preferably pre-filtered. The pre-filtered setpoint for the oxygen level serves as the reference input for both the feedforward control and a controller. The output signals of the feedforward control and the controller are summed. The summed signal represents the target lambda value upstream of the catalyst 122.

[0037] Since the input variables of the plant model, in particular the signal from lambda sensor 121 upstream of the catalytic converter, are subject to uncertainties, the plant model is preferably adapted. Likewise, the feedforward control and, if necessary, controller parameters can be adapted. Such an adaptation is described, for example, in DE 10 2018 251 725 A1.

[0038] As described above, the feedforward control and prediction 210 are based on the plant model. The accuracy of the feedforward control therefore depends on the accuracies of the measured and model parameters incorporated into the plant model. Inaccuracies in these parameters and model simplifications compared to the actual plant can lead to the predicted oxygen level of the catalyst not corresponding to the actual level upon re-engagement after a period of inactive control intervention.

[0039] After a period of inactive control intervention, the catalyst 122 is typically outside the catalyst window (a state with a fill level suitable for pollutant conversion), for example, after overrun fuel cut-off or after lambda adjustment for component protection. Therefore, reactivation after such a period is particularly relevant to emissions. The catalyst 122 must be brought back into the catalyst window as quickly as possible to ensure optimal exhaust gas conversion. An incorrect prediction 210 of the oxygen level must be avoided at all costs, as this would lead to an unsuitable reinitialization of the trajectory for the oxygen level and thus to an unsuitable pilot lambda value upon reactivation. In the worst case, this could result in a lean mixture even though the actual oxygen level in the catalyst is already too high, or in an enrichment when the oxygen level is already too low.That would have increased NO. x This results in emissions or increased CO and HC emissions.

[0040] To prevent this, method 200 provides for a plausibility check of the predicted oxygen level or the associated expected or planned pre-control lambda value using a sensor 123 downstream of the catalyst. Preferably, this sensor 123 is a switching lambda sensor downstream of the catalyst 122.

[0041] Since the effects of an incorrect prediction 210 of the oxygen level are particularly high at very high or very low modeled oxygen levels, the plausibility check 200 can optionally be limited to such situations.

[0042] The plausibility check 200 of the expected or planned input lambda value is performed particularly when, in an evaluation step 220, it is detected that the lambda control or the input control is being reactivated. For this purpose, operating parameters of the internal combustion engine 110 (e.g., engine speed, load demand, fuel mass flow, throttle position, etc.) can be evaluated. Signals from other vehicle components, which are received, for example, via an internal vehicle data network (e.g., CAN), can also indicate a reactivation of the lambda control.

[0043] In a categorization 230, the expected or planned pre-control lambda value is preferably categorized into one of three categories: 'lean mixture', 'stoichiometric requirement', or 'rich mixture'. In the case of a categorization as lean mixture, i.e., if the prediction 210 results in a high expected or planned oxygen level or a lean expected or planned pre-control lambda value, a first plausibility check 242 determines whether significant enrichment has occurred within the period before the lambda control is reactivated. This period is required for exhaust gas transport from the internal combustion engine 110 to the lambda sensor 123 downstream of the catalyst 122. For this purpose, the composition of the air-fuel mixture supplied to the internal combustion engine 110 can be observed, for example, simultaneously with or within the scope of the prediction 210.If the proportion of rich gas components in the exhaust gas produced by the internal combustion engine exceeds a threshold value (e.g. at least 3%, 5% or 7% of the total exhaust gas mass flow in the relevant period), a reinitialization 260 of the feedforward control can be carried out with a correspondingly high feedforward lambda value or a leaning-out request 262.

[0044] If, however, it is determined that rich gas components in the exhaust gas do not exceed or have exceeded the threshold value during the relevant period, a second plausibility check 252 verifies whether the lambda sensor 123 downstream of the catalyst 122 indicates a lean lambda value. If this is the case, it must be assumed that the prediction 210 of the pilot lambda value or the oxygen level was incorrect. Therefore, in this case, the reinitialization 260 is carried out with a low pilot lambda value or a richening request 261. If, on the other hand, no lean lambda value downstream of the catalyst 122 is detected in the second plausibility check 252, the result of the prediction 210 can be considered substantially correct, so that the leaning-out request 262 can be used for the reinitialization 260 of the pilot control.

[0045] Similarly, in the case of categorization 230 as enrichment, a first plausibility check 244 can determine whether a significant leaning out has occurred within the period before the lambda control is reactivated. This period is required for the exhaust gas transport from the internal combustion engine 110 to the lambda sensor 123 downstream of the catalyst 122. If the proportion of lean gas components in the exhaust gas generated by the internal combustion engine 110 exceeds a threshold value (e.g., at least 3%, 5%, or 7% of the total exhaust gas mass flow in the relevant period), the reinitialization 260 of the feedforward control can be carried out with a correspondingly low feedforward lambda value or a fuel enrichment request 264. This enrichment requirement 264 can correspond to the enrichment requirement 261, which is used in the above-described case of a failed plausibility check of the prediction 210 of a slimming requirement.However, a separate, rich input-value lambda 264, which differs from the rich input-value lambda 261, can also be chosen.

[0046] If, however, the first plausibility check 244 determines that lean gas components in the exhaust gas do not exceed or have exceeded the threshold value during the relevant period, a second plausibility check 254 verifies whether the lambda sensor 123 downstream of the catalyst 122 indicates a rich lambda value. If this is the case, it must be assumed that the prediction 210 of the pilot lambda value or the oxygen level was incorrect. Therefore, in this case, reinitialization 260 is performed with a high pilot lambda value or a leaning-out request 265, which may correspond to or differ from the leaning-out request 262 explained above.If, however, no rich lambda value is detected downstream of the catalyst 122 in the second plausibility check 254, the prediction 210 can be considered to be essentially correct, so that the enrichment request 264 can be used for the reinitialization 260 of the lambda control or its feedforward control.

[0047] If a substantially stoichiometric pre-control lambda value is determined as predicted in categorization 230, this can, for example, be used directly as the initialization value 263 for the reinitialization 260 of the lambda control or its pre-control, since in this case no massive miscalculation is to be feared, so that the control may be sufficient to quickly reach the catalyst window. However, even in such a case, it can be advantageous to provide at least one plausibility check, in particular a second plausibility check to verify the lambda value downstream of the catalyst 122, in order to be able to provide a correspondingly corrected pre-control lambda value 263 in the event of a strongly lean or strongly rich exhaust gas lambda value.

[0048] This ensures that in all situations, a breakthrough of rich or lean exhaust gas through the catalyst 122 can be reliably prevented when the feedforward control is restarted after an operational interruption, which has an overall positive effect on the emission behavior of the vehicle 100.

[0049] Situations in which plausibility checks using the probe downstream of the catalytic converter could lead to an incorrect result, namely when a large mixture change (e.g., typically greater than 5%) opposite to the current exhaust gas lambda value (sensor 123) downstream of the catalytic converter has occurred shortly before resumption of operation, and the gas transit time between the combustion chamber and probe 123 downstream of the catalytic converter 122 is longer than the time between the mixture change and resumption of operation, are covered by the respective first plausibility check stages 242 and 244. In these cases, probe 123 has not yet been able to react to the mixture change, and it is possible that a rich or lean exhaust gas lambda value already exists upstream or in the catalytic converter 122, even though probe 123 downstream of the catalytic converter 122 still indicates a lean or rich exhaust gas.

[0050] Large mixture changes are to be expected, for example, in the context of combustion chamber shutdowns (e.g. for gear shift support) where more oxygen is introduced into the catalyst 122, or from rich component protection where more oxygen is carried out of the catalyst 122.

[0051] If, on the other hand, such a mixture change is small in magnitude (e.g., less than 3%), then the sensor 123 downstream of the catalyst 122 may not react to it within the gas transit time between the combustion chamber and the sensor 123. However, in this case, due to the storage capacity of the catalyst 122, it can be assumed that the sensor 123 downstream of the catalyst 122 will not indicate a different direction for the exhaust gas lambda value (rich or lean) than is actually present upstream or in the catalyst 122. Therefore, the second plausibility check stage 252, 254 of the predicted pre-control lambda value will not lead to a misinterpretation with the signal from the sensor 123 downstream of the catalyst 122. Accordingly, the choice of the threshold value can be based on the storage capacity of the respective catalyst 122 used. QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] DE 102016222418 A1

[0007] DE 102018208683 A1

[0007] DE 102018217307 A1

[0007] DE 102018251725 A1

[0037]

Claims

[1] Method (200) for operating an internal combustion engine (110) with at least one catalyst (122), wherein control interventions of a lambda control for controlling an exhaust gas composition of the internal combustion engine are deactivated, comprising: Determining a current exhaust gas composition upstream of the at least one catalyst (122), Determining a current oxygen level of the at least one catalyst (122) based on the determined current exhaust gas composition, Determining (210) a planned control intervention on a composition of an air-fuel mixture supplied to the internal combustion engine (110) based on the determined current oxygen level of the at least one catalyst, Determining a current exhaust gas composition (123) downstream of the at least one catalyst (122), Determining a future exhaust gas composition (123) resulting from an air-fuel mixture already supplied to the internal combustion engine (110) downstream of the at least one catalyst (122) and Reactivating the lambda control and determining (260) a control intervention to be carried out as a function of the planned control intervention and the current exhaust gas composition (123) downstream of the at least one catalytic converter (122), and / or as a function of the planned control intervention and the future exhaust gas composition (123). [2] Method (200) according to claim 1, wherein the determination of the future exhaust gas composition (123) downstream of the at least one catalyst (122) resulting from an air-fuel mixture already supplied to the internal combustion engine (110) comprises determining the exhaust gas composition (123) downstream of the at least one catalyst (122) for a period of time whose duration corresponds to a flow duration of the exhaust gas of the internal combustion engine (110) for flowing through a flow path from the internal combustion engine (110) to downstream of the at least one catalyst (122). [3] Method (200) according to claim 1 or 2, wherein the determination (210) of the planned control intervention comprises a categorization (230), in particular into one or more categories of leaning, enriching and stoichiometric requirement. [4] Method (200) according to one of the preceding claims, wherein a control intervention (261, 265) to be carried out which is opposite to the planned control intervention is determined if the current exhaust gas composition downstream of the at least one catalyst (122) corresponds to an exhaust gas composition which can be generated by the planned control intervention and at the same time the future exhaust gas composition does not already correspond to an exhaust gas composition which can be generated by the opposite control intervention (261, 265). [5] Method according to one of the preceding claims, wherein a control intervention (262, 263, 264) to be carried out corresponding to the planned control intervention is determined if the current exhaust gas composition corresponds to an exhaust gas composition resulting from an opposite control intervention (261, 265). [6] Method according to one of the preceding claims, wherein a control intervention (262, 263, 264) to be carried out corresponding to the planned control intervention is determined if the future exhaust gas composition corresponds to an exhaust gas composition resulting from an opposite control intervention (261, 265). [7] Method (200) according to one of the preceding claims, wherein in the period without active control intervention, one or more of the group consisting of a coasting or sailing operation of the internal combustion engine (110), a component protection function and a (partial) combustion chamber shutdown are carried out. [8] Method (200) according to one of the preceding claims, wherein the current exhaust gas composition and / or the future exhaust gas composition are determined by means of a sensor (112, 121, 123, 127), in particular one or more lambda sensors (121, 123). [9] Computing unit (130) which is configured to carry out all method steps of a method (200) according to one of the preceding claims. [10] Computer program which causes a computing unit (130) to carry out all method steps of a method (200) according to one of claims 1 to 7 when it is executed on the computing unit. [11] A machine-readable storage medium having stored thereon a computer program according to claim 9.

Images