Adapting the reducing agent dosage in the catalytic converter using the sliding mode method

Abstract

The invention relates to a method for regulating the dosing of a reducing agent in an SCR catalytic converter, the method comprising: determining (160) a setpoint effectiveness (η) for the catalytic converter SP ); according to the rated effectiveness (η SP ) calculates (220) an initial dosage (V) for the reducing agent; determines (310) a current value of a switching function (S) of a sliding mode controller, wherein the switching function is based on the rated effectiveness (η SP ) and actual effectiveness (η PV ) and the time derivative of the current deviation; and adapting (330) the initial dosage according to the value of the switching function (S) to obtain the resulting dosage for the reducing agent (r NH3us ).

Description

Technical Field

[0001] The invention relates to a method for adapting the metering of a reducing agent in an SCR catalytic converter by means of a sliding mode regulator, as well as a computer program and a computing unit for carrying out the method. Background Art

[0002] Various measures are known for reducing nitrogen oxides in the exhaust gas of internal combustion engines. In particular, catalysts are used based on the principle of selective catalytic reduction (SCR), in which nitrogen oxides (NO, NO 2 , summarized as NO x ) is converted into nitrogen (N 2 For this purpose, an aqueous urea solution is injected into the exhaust gas in the exhaust gas line, from which carbon dioxide CO is generated under suitable temperature conditions. 2 and ammonia NH 3 Instead of the urea-water solution described, other reducing agents which are suitable for the selective reduction in the SCR catalytic converter are generally also conceivable.

[0003] Effective reduction of nitrogen oxides, especially in order to comply with the requirements for nitrogen oxides and CO 2 In order to meet the increasingly stringent emission limit values, it is important to adapt the injected amount of urea solution as accurately as possible to the instantaneous nitrogen oxide emissions of the engine and the operating conditions in the catalyst. In the case of too little dosing, nitrogen oxides cannot be reduced to a sufficiently effective degree, while in the case of too much dosing, excess ammonia may undesirably enter the exhaust gas after the catalyst and thus into the ambient air, which is also called ammonia slip (Ammoniak-Schlupf). As a dosing device, for example, a known dosing valve with a suitable controller can be used, which is usually arranged in the inlet area of ​​the catalyst. However, other injection positions are also conceivable, and likewise, a combination of multiple dosing devices, for example at different positions in the exhaust gas aftertreatment system, is conceivable.

[0004] The SCR catalysts known today store ammonia or NH3 by adsorption on the catalyst surface. 3 The more reducing agent is supplied to the catalyst, the more NO x The more successful the conversion is. 3 If the storage capacity of the SCR catalyst is not exhausted, the overdosed reducing agent will be stored. 3 In terms of NH 3 Filling level (NH3 If less reducing agent is available than is required for the conversion of the nitrogen oxides currently present in the exhaust gas, the stored reducing agent is consumed for the conversion of the nitrogen oxides and thus reduces the NH 3 Fill level.

[0005] In order to achieve the desired accuracy of the dosing, different measures for control based on software and hardware are meaningful. This is especially true for situations in which the system is in a dynamic transition from one operating point to another. Different calculation methods for urea dosing are possible. For example, the total filling level of ammonia adsorbed on the SCR catalyst or the effectiveness of the catalyst can be used as a control variable (Führungsgröße). When the total filling level is used, this means that there is a temperature-dependent setpoint value for the filling level of the catalyst, which is adjusted. Therefore, at high temperatures, it is often switched to a control based on the effectiveness, so that the effectiveness represents the setpoint value. The total dosing for the reducing agent or urea solution is then obtained as the sum of the precontrol variable and the regulator share. The precontrol variable can be calculated by comparing it with the nitrogen oxides (NO x ) reaction, through desorption from the catalyst surface and the adsorbed NH 3 The losses caused by oxidation are calculated using a physical-chemical model and are provided as a pilot control variable, ie as the basic amount of reducing agent to be replenished. The controller portion can then be calculated by the proportional controller as the deviation between the setpoint filling level and the actual filling level of the catalytic converter.

[0006] For example, a temperature-dependent NH 2 for the catalytic converter can be specified for controlling the SCR system. 3 Rated fill level where the NH 3 The setpoint filling level is high at low temperatures and decreases towards high temperatures. If the driver accelerates the motor vehicle, a change occurs from low to high temperatures, which usually leads to an increase in nitrogen oxide emissions. If the setpoint filling level is now adjusted, a dosing pause is mandatory in this phase, which lasts until the lower target filling level is approximately reached at higher temperatures.

[0007] However, in view of the NH 3 In the case of such a control of the total filling level, it is not taken into account how the filling level is distributed geographically to the catalytic converters depending on the situation or operating state.

[0008] As described above by the acceleration phase and the resulting higher temperatures than in the first example case, in the case of a sudden reduction in the target filling level, further injection of reducing agent is stopped until the targeted filling level is reached, so that the filling level exceeding the catalytic converter is reduced relatively evenly and no NH 3 Further deposit of.

[0009] In contrast, in the second exemplary case, as soon as the dosing is replenished again, for example because a higher temperature is maintained quasi-steadily and the actual filling level falls below the setpoint filling level in the meantime, the catalytic converter is usually refilled from the front, that is, in the region of the injection nozzle located upstream, so that a higher adsorption also takes place in this region. 3 In the case of an overall fill level, for example in the first case, a further local fill level distribution is realized, in which NH 3 The distribution shifts further upstream.

[0010] Since there are temperature differences at certain locations in the catalytic converter, with the temperature generally decreasing towards the downstream, the overall effectiveness can also be differentiated between different situations with the same overall filling level by means of such a different filling level distribution. If areas with a lower temperature are filled to a greater extent (i.e., for example, in the case of a metering pause), the effectiveness can, for example, be temporarily reduced. Summary of the invention

[0011] According to the invention, a method for adapting the metering of a reducing agent in an SCR catalytic converter and a computer program and a computing unit for carrying out the method are proposed having the features of the independent patent claims. Advantageous embodiments are the subject matter of the dependent claims and the following description.

[0012] In particular, a control method for metering is proposed, in which a setpoint effectiveness level for the catalytic converter is first determined, an initial metering level for the reducing agent is then calculated as a function of the setpoint effectiveness level, and a current value of a switching function of a sliding mode controller is subsequently determined, wherein the switching function is based on a current deviation between the setpoint effectiveness level and the actual effectiveness level and a time derivative of the current deviation; and wherein the initial metering level is then adapted as a function of the current value of the switching function in order to thereby obtain a resulting metering level for the reducing agent.

[0013] In this way, the total effectiveness of the catalytic converter is used as a control variable for the control process. If the actual effectiveness differs from the setpoint effectiveness, the sliding mode controller can ensure that the actual effectiveness approaches the setpoint effectiveness, for example (depending on the specific form of the switching function) in a PT1-like manner (ie, like a delay element).

[0014] According to one specific embodiment, the calculation of the initial dosing can be carried out as follows: firstly, a target filling level is calculated as a function of the target effectiveness and then, based on the catalytic converter model, the expected loss of reducing agent within a predefined time period is calculated with the calculated target filling level. The expected loss can then be defined as the initial dosing. Different detailed modeling can be used and, for example, ammonia slip, oxidation and nitrogen oxide conversion can be taken into account and combined as a loss term.

[0015] According to one specific embodiment, the resulting metering amount can be calculated by determining the value of a switching function and, if the value of the switching function is greater than zero, calculating the resulting metering amount by increasing the initial metering amount by a predetermined excess amount, and, if the value of the switching function is less than zero, calculating the resulting metering amount by reducing the initial metering amount by a predetermined excess amount. This simple control makes it possible in a simple manner to use the value for the effectiveness calculated by evaluation to achieve that the effectiveness level is shifted to the desired value without overshooting.

[0016] According to another embodiment, the resulting dosage can be determined as follows:

[0017] ,in

[0018] r NH3us is the amount obtained by the regulator,

[0019] S is the switching function of the sliding mode regulator,

[0020] M is the predetermined excess amount,

[0021] V is the initial dosage, and

[0022] is a predetermined value close to zero.

[0023] By introducing a saturation function for excess quantities, a high frequency of switching of the metering can be avoided.

[0024] According to another exemplary embodiment, the initial dosage can be multiplied by an adaptation factor in addition to the saturation function mentioned above. The adaptation factor can be defined, for example, by the following formula:

[0025] ,in

[0026] f is the adaptation factor;

[0027] F is a constant that is optionally related to the temperature and mass flow in the catalyst;

[0028] S is the switching function of the sliding mode regulator; and

[0029] sgn() stands for sign function.

[0030] The further adaptation ensures that, in the case of a positive value of the switching function, the pilot control variable is increased and the actual value is brought closer to the setpoint value.

[0031] In this context, the excess quantity used for adapting the initial metering quantity can be determined in all variants, for example, according to one of the following possibilities:

[0032] - predefining the excess amount by means of a characteristic diagram which is dependent on the mass flow and / or the temperature in the catalytic converter;

[0033] - calculating the excess amount from the difference between the set filling level and the actual filling level of the catalytic converter, wherein the set filling level is calculated from the set effectiveness;

[0034] - Calculate the excess based on the difference between the rated effectiveness and the actual effectiveness;

[0035] - forming a minimum value from a plurality of predefined excess quantities and / or calculated excess quantities.

[0036] However, the excess amount can also be determined in another way or by a combination of these possibilities, for example by a predefined value in a conventional filling level controller, which determines the excess amount as a function of different catalytic converter parameters.

[0037] According to a preferred embodiment, an initial setpoint level of effectiveness, for example specified in a characteristic diagram, and at least one of a specified setpoint filling level of the catalytic converter and a specified maximum filling level of the catalytic converter can be taken into account when determining the setpoint level of effectiveness. Thus, for example, an excessively high or low catalytic converter filling level, which could occur due to a simple specification of the level of effectiveness, for example at very low temperatures or very low mass flows, can be prevented.

[0038] In all functions and method steps, at least a portion of the input values ​​of the method can be predetermined by at least one characteristic curve. Various values ​​such as a setpoint fill level, a setpoint effectiveness, a maximum fill level, a time constant of a sliding mode controller, a value of a saturation function, ... Etc. Likewise, not one characteristic map may be predefined for each of these values, but rather more than one characteristic map may be predefined, for example as a function of catalytic converter aging.

[0039] In all calculations, it is possible to use at least partially a locally differentiated catalyst model for calculating at least one filling level and / or effectiveness of the catalyst, wherein the catalyst is divided into a plurality of axial segments in the catalyst model. Such a model, for example a plate model with a plurality of catalyst plates arranged in succession, allows efficient and real-time-capable calculations without model inversions that may be numerically unstable and is therefore economical in terms of computing time and storage requirements. In particular, the effectiveness can be optimally designed in the case of dynamic transitions.

[0040] A computing unit according to the present invention, for example a control unit of a motor vehicle, is configured, in particular in terms of programming, to carry out the method according to the present invention.

[0041] The implementation of the method according to the invention in the form of a computer program or computer program product with program code for executing all method steps is also advantageous, since this results in particularly low costs, in particular if the executing control device is also used for other tasks and is therefore present anyway. Suitable data carriers for providing the computer program are in particular magnetic, optical and electrical memories, such as hard disks, flash memories, EEPROMs, DVDs and others. It is also possible to download the program via a computer network (Internet, intranet, etc.).

[0042] Further advantages and embodiments of the invention are apparent from the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The invention is schematically illustrated in the drawings by means of exemplary embodiments and is described below with reference to the drawings.

[0044] Figure 1 An exhaust gas aftertreatment system is shown in which an embodiment of the present invention can be used;

[0045] Figure 2 A first method portion according to an embodiment for determining a nominal degree of effectiveness is shown;

[0046] Figure 3 shows an extended method part according to an embodiment for determining an NH3 pilot control variable; and

[0047] Figure 4 The method steps for adapting the metering according to one embodiment of the present invention are shown. DETAILED DESCRIPTION

[0048] Figure 1An exhaust gas aftertreatment system is shown by way of example, in which an embodiment of the present invention can be used. Here, only the main components are described and shown. As a structure, a standard exhaust gas system in a motor vehicle with an internal combustion engine and an SCR catalyst 70 can be used. The figure shows an internal combustion engine 1 by way of example, which includes an engine block 10 and an exhaust gas channel 30. The exhaust gas aftertreatment system can include multiple components, wherein a first catalyst, such as an oxidation catalyst 40, is first shown in the flow direction of the exhaust gas, which catalyst is not important for the present invention. Similarly, a particle filter or other elements may also be present here alternatively or additionally. An SCR catalyst 70 is connected thereto, before which a reducing agent can be introduced by means of a dosing unit 90 for reducing nitrogen oxides in the exhaust gas flow. In order to monitor the nitrogen oxide concentration in the exhaust gas, nitrogen oxide sensors 60, 80 can be arranged in the exhaust gas channel 30 in the flow direction before and / or after the SCR catalyst 70, respectively.

[0049] The nitrogen oxide sensors 60, 80 are connected to a control unit 51. The control unit 51 can be, for example, an integral component of the engine control unit 50 or can also form a separate control unit for regulating the exhaust system. In particular, the control unit 51 can include a software- and / or hardware-based unit, with which the metering of the reducing agent can be controlled by the metering unit 90. The schematically shown emission model 20 of the internal combustion engine can be used in this control method and other control methods.

[0050] In order to be able to optimize the reducing agent (e.g. NH) also taking into account the dynamic transitions between different operating points 3 For the dosing of SCR catalysts (or urea-water solutions), locally subdivided catalyst models can also be used. This is particularly suitable for modeling the SCR catalyst and the conditions present therein, such as the temperature distribution, in a positionally discretized manner. For this, for example, a model can be selected in which the catalyst is divided into a plurality of sections or plates, which are arranged, for example, axially one after the other in the flow direction. The individual plates can then be simulated individually and related to one another (verknüpfen) using the same model equations, in such a way that the molecules or output variables flowing out of the upstream plate form the input variables of the next plate located downstream. For example, the catalyst can be divided into ten modeled plates with the same parameters, which can then be simulated appropriately. Alternatively, when different types of catalysts are arranged in a housing, the plates can also have different parameters, such as different frequency factors and different activation energies for the reactions being carried out. The modeling can be carried out as broadly or in detail as desired, for example when considering the oxidation of ammonia on the catalyst surface or the oxidation of No to be carried out.x The reactions are more precisely divided into standard SCR, fast SCR and 4:3 or "NO 2 -SCR". Such details of catalytic converter modeling are known in the art and can be included together with the common models and equations. The control method according to the invention does not depend on detailed modeling, but can be applied to each of these models for the respective considered partial catalytic converter part (for example a catalytic converter plate).

[0051] Details on the modeling can be found, for example, in the following publications:

[0052] - Cristian Ciardellia, Isabella Nova, Enrico Tronconi, BrigitteKonrad, DanielChatterjee, Karlheinz Ecke, Michel Weibel: SCR-DeNOx for dieselengine exhaust aftertreatment: unsteady-state kinetic study and monolithreactor modeling, Chemical Engineering Science Volume 59, Issues 22–23;

[0053] - Chatterjee D., Burkhardt T., Bandl-Konrad B., Braun T., TronconiE., Nova I.,Ciardelli C. (2005): Numerical simulation of ammonia SCRcatalytic converters: model development and application, SAE Technical Paper2005-01-0965;

[0054] - Rana Faltsi and Jayesh Mutyal: Confidence in Modeling SCRAftertreatmentSystems, Automation Simulation World Congress 2012; and

[0055] - E. Tronconi, A. Cavanna, P. Forzatti: Unsteady analysis of NOReduction overSelective Catalyst Reduction – De-NOxMonolith Catalysts, Ind.Eng. Chem. Res 1998, 37, 2341-2349.

[0056] For each of the catalyst segments (slices) formed in this way, the filling level m and the degree of effectiveness η can be converted to one another in a simple manner by means of suitable functions. In order to determine the desired degree of effectiveness, the catalyst can also be modeled in a simplified manner as a single slice with a correspondingly adapted size, since possible inaccuracies can be taken into account in a simple manner by adapting the characteristic curve diagram of the model.

[0057] According to one embodiment of the invention, the total effectiveness can now be used as a control variable for the dosing control of the reducing agent. The actual effectiveness can be modeled, while the setpoint effectiveness can be predetermined, for example, as a function of the exhaust gas mass flow and temperature. If the actual value and the setpoint value of the effectiveness deviate from each other, it can be attempted to bring the actual effectiveness close to the setpoint value in the form of a 1st-order delay element (PT1 element). To achieve this, a regulator with a variable structure, such as a sliding mode regulator, can be used, for example.

[0058] Figure 2 The schematic diagram shows the nominal effectiveness level (set point, SP) η according to the first embodiment. sp Determination of the fill level m. In this case, advantage is taken of the fact that in a monolithic model, i.e. in the case of modeling as a single piece with a correspondingly adapted size (1-Brick model), the filling level m and the degree of effectiveness η can be easily converted to one another. For this purpose, the following function is used, which converts the filling level m into the degree of effectiveness η resulting from the fixation of the filling level. The simplification according to the piece is possible because the determination of the setpoint value is relatively insensitive with regard to the conversion accuracy, and possible deviations can be easily taken into account in the characteristic curve family before the conversion. Alternatively, however, the modeling and calculation of the filling level according to the degree of effectiveness can also be carried out with a locally divided filler model, for example with a multi-piece model, by inverting the model using numerical methods.

[0059] As input variables for determining the desired degree of effectiveness, the average catalyst temperature T (or the temperature of the relevant section) and the exhaust gas volume flow dVol and / or the exhaust gas mass flow dmEG can be used first. The desired desired degree of effectiveness η can be predetermined by the characteristic diagram 100. sp0 However, controlling the metering solely with such a predetermined setpoint effectiveness can lead, in particular at low temperatures and an overestimated effectiveness, to very high catalytic converter filling levels, which in turn increase the risk of slippage downstream of the catalytic converter.

[0060] Therefore, the maximum filling level η can be additionally predefined as an input variable. max , so as to avoid slippage as much as possible. For this purpose, the characteristic map 120 can be predefined again for the maximum filling level, wherein the average catalyst temperature T and the exhaust gas mass flow dmEG can be introduced as input variables again. The filling level can be converted into an effective degree in the function module 140 by a corresponding model as already described.

[0061] Furthermore, a low mass flow (e.g. during idling) can result in the effectiveness level itself no longer being sufficiently convincing, since it can then also be very close to 100% under unfavorable conditions. In order to avoid that the fill level drops too much under these conditions and thus that there is not enough NH 3 The buffer is available for use in the next acceleration phase with higher quality streaming. sp0 The other characteristic diagram 110 is used. The degree of effectiveness is also recalculated in the function module 130 by the model based on these values.

[0062] Overall, the setpoint level of effectiveness obtained from the characteristic diagram 100 can then first be compared in block 150 with the level of effectiveness 130 obtained by calculation from the setpoint filling level, and the maximum of these two values ​​can then be selected for further modeling. The value obtained in this way can then be compared in block 160 with the level of effectiveness obtained by calculation from the maximum filling level, and the minimum of these two values ​​can then be used as the setpoint level of effectiveness n obtained. sp In other alternatives, instead of using a maximum and a minimum value, a corresponding adaptation of another validity level, for example a mean value, can also be carried out based on the two values ​​in each case.

[0063] However, it is also possible in principle to dispense with the two limiting modules for the setpoint fill level 110 and the maximum fill level 120 (or one of the two modules) and to carry out the method only with a simpler or different calculation of the setpoint effectiveness.

[0064] For all described steps, the exhaust gas mass flow dVol and the exhaust gas volume flow dmEG can alternatively be used as input variables.

[0065] Figure 3 Schematically showing the Figure 2 Rated effectiveness η sp The determination of is used together to determine the pre-control amount, that is, the other parts of the embodiment of the present invention of the set initial dosage. For this purpose, the rated effectiveness η obtained at the end of the steps described above can be obtained again by the function in step 200. sp The desired filling level can then be used to calculate in the model 210 how much NH 3 Losses occur. Different mechanisms can be considered, depending on the level of detail, preferably modeling: how much NH 3 Escape (dmNH3slip), how much NH 3 and nitrogen oxides are converted by reaction (dmNOxReac), and how much NH 3 Oxidation (dmOxi). The sum of all modeled loss terms then yields the pilot control variable as output of module 220, ie the initial dose that must be injected in order to keep the filling level of the catalytic converter and thus the effectiveness of the catalytic converter constant given the calculated losses.

[0066] Analogously to steps 130 and 140 , for the calculation of the filling level 200 and / or the modeling of the loss term 210 , it is again appropriate to use a simplified 1-Brick model for modeling the catalytic converter; however, it is also possible to use a model of a catalytic converter with multiple sections for one or more of these calculation steps, in particular if there is sufficient computing power in the computing unit used.

[0067] Figure 4 The steps of a still further refined embodiment of the invention for determining or adapting the metering of a reducing agent in an SCR catalytic converter are shown by way of example. Functional blocks that have already been described in detail in conjunction with the previous figures are again provided with the same reference numerals.

[0068] Here, it is shown that the combined Figure 2 and Figure 3The method parts described above are now combined with corresponding regulators for adapting the initial metering or pre-control quantity. Here, the degree of effectiveness is used as the control variable. Based on the regulator output of the regulator 330, it is thus obtained that the value previously obtained in step 220 is as follows: Figure 3 The quantity adapted in this way results in a quantity for injecting the reducing agent into the catalytic converter or an equivalent for providing the reducing agent, ie, for example, NH 3 or form NH 3 The resulting adapted dosage of the urea-water solution.

[0069] In this connection, it is advantageous to refine the total effectiveness, for example, without overshoot and similar first-order delay elements (PT1 elements), and basically approach the rated value. This makes it possible to effectively avoid long metering pauses, for example, in the case of temperature rise. However, the calculations inferred for this case may be very complex. On the contrary, the same behavior can be achieved relatively simply with a sliding mode controller, in which different solutions for sliding mode regulation of metering can be used. The basic principles of the sliding mode control method are known in the profession and can be obtained, for example, from the document "A quick introduction to sliding mode control and its applications", (Università degliStudi diCagliari, Dipartimento di ingegneria elettrica ed elettronica, 2008".

[0070] According to a simple first specific embodiment, a first-order sliding mode controller can be used, wherein as a basic idea an excess quantity M is determined which is then subtracted from or added to a previously determined initial metering quantity depending on the currently determined demand.

[0071] In this case, in step 310 , a switching function S is defined as follows as a function of the control difference:

[0072] ,in

[0073] At the rated effective level and actual effectiveness The deviation between ;

[0074] express The time derivative of T 63 is the time constant,

[0075] It may optionally also be mass flow-dependent and / or temperature-dependent.

[0076] The time constant can be predetermined, for example, by means of a characteristic diagram 340. The actual effectiveness is determined according to the model 300, while the setpoint effectiveness is determined from Figure 2 The steps 100 to 160 shown in FIG.

[0077] Alternatively, it is also possible to make the actual effectiveness level like two first-order (PT1) delay elements connected in series. In this variant, the effectiveness level increase is more strongly limited at the beginning, which can be used to prevent NH 3 Escape makes sense.

[0078] As a “sliding mode” a system is then defined with the condition S = 0. The regulator thus “slides” along the surface.

[0079] In this regard, in a simple embodiment it is provided that if S>0, the resulting metering quantity is calculated by adding the excess quantity M to the previously determined initial metering quantity V,

[0080] ;

[0081] If S < 0, the resulting dosage is calculated by subtracting the excess amount M from the previously determined initial dosage V.

[0082] .

[0083] In this case, the excess amount can also be within the range of the initial metering amount, so that in the case of S<0 it results that the resulting metering amount is zero or close to zero.

[0084] Here, we can now target the effectiveness The control deviations and their derivatives each use variables calculated analytically according to the model 300 , for example according to a multi-slice model with 10 slices. Therefore, a computationally complex model inversion is not necessary.

[0085] However, due to the two-point characteristic of this simple sliding mode controller, the dosage can be switched between the two defined dosages at a high frequency (chattering). Since preferably a very large overage should be used in order to ensure the fill level build-up and the degree of effectiveness under all circumstances, in this case a high frequency of switching between high injection quantities and dosage pauses results (if the overage corresponds, for example, to the pilot control quantity), which should ultimately be equivalent to halving the dosage frequency.

[0086] To prevent this, a saturation function can be used in an extended embodiment instead of a pure addition or subtraction of the excess amount. Here, the resulting dosing is determined as follows:

[0087]

[0088] That is, it is the sum of the pilot control variable plus the excess variable multiplied by the saturation function, where ε is a small value close to zero.

[0089] This variant with a saturation function can prevent an unstable switch between the two metering quantities and replace it with a stable approximation.

[0090] Another embodiment of the sliding mode method according to the invention is based on adapting the pilot control variable by a factor in addition to using a saturation function in order to prevent high-frequency switching of the dosage. For this purpose, the pilot control variable is multiplied by an adaptation factor f, for which the following relationship applies again:

[0091] .

[0092] In this case, F denotes an application constant, which may optionally also be temperature-dependent and mass flow-dependent, sgn() denotes a sign function, and S denotes the already described switching function of the controller.

[0093] If S is positive, the adaptation factor f is therefore increased in order to increase the pilot control variable and bring the actual value closer to the setpoint value.

[0094] In this embodiment with an adapted pilot control variable in the regulator, the resulting metering quantity at the regulator output is then obtained as follows:

[0095] .

[0096] The applicable excess variable M for all of the mentioned control methods can be determined, as is known in the prior art, based on a maximum admissible intervention in the pilot control variable, which is used in conventional filling level controllers for SCR catalytic converters, depending on mass flow, temperature, filling level, raw nitrogen oxide emissions, etc., wherein the maximum admissible intervention modifies the pilot control variable additively.

[0097] Alternatively, it is also possible to determine the additional excess amount M in part 320 according to other approaches. For this purpose, for example, a characteristic curve family can be used in which the exhaust gas mass flow and the temperature T are used as input variables. Likewise, in an extended option, the data from such a characteristic curve family can be multiplied by the difference between the rated filling level and the actual filling level in order to determine the maximum possible or meaningful excess amount. Instead of the filling level difference, it is also possible to determine the difference between the rated effectiveness and the actual effectiveness in a similar manner and likewise multiply it by the characteristic curve family. In general, any combination of these approaches is also possible, for example by forming a minimum value from a plurality of the methods mentioned, wherein a separate characteristic curve family or parameter set is provided (vorhalten) for the regeneration state of the upstream connected particle filter. As the rated filling level m sp In this case, the following filling level applies, which in turn can be calculated using a suitable model, such as the 1-Brick model 200 according to Figure 2 It is calculated based on the obtained rated effectiveness level ηsp.

[0098] Based on the three described method parts, the resulting metering can then be determined in step 330 according to a further embodiment of the present invention according to one of the methods described for a sliding mode controller. .

[0099] In this case, other variables stored or predetermined by a characteristic diagram can be introduced into the method, for example the already described time constant T with characteristic diagram 340 for setting the convergence speed of regulator 310. 63 This factor can additionally depend on the catalyst temperature and / or the exhaust gas mass flow. Here, too, a plurality of characteristic curves can be provided as a function of the aging or operating mode of the catalyst.

[0100] For the adaptation of the saturation curve as described, a characteristic diagram for the value ε can also be predefined, which in turn can depend on the temperature and / or the mass flow. For these values ​​of these parameters, the transition from the negative control limit to the positive control limit can be designed to be harder or softer during the metering adaptation in step 330 .

[0101] All the parameters described should preferably be taken into account as a function of the aging of the catalytic converter, even if the aging is not taken into account in detail in the examples shown. For this purpose, for example, two versions can be present for each of the characteristic curves 100, 110, 120, 340, 350 described, which are respectively suitable for a new catalytic converter and a marginally aged catalytic converter, wherein the characteristic curves are then interpolated linearly or exponentially between the characteristic curves depending on the estimated catalytic converter age. Other possibilities are also conceivable in order to take into account aging effects in modeling and control.

[0102] It should be understood that the Figures 2 to 4 The different steps described as modules or function blocks are implemented, for example, as steps and functions in suitable software for regulating the dosage and / or using a microcontroller, FPGA or other suitable elements that can implement the described processing of input and output values. Likewise, other functions and steps can be used, for example, to improve these models or to introduce other parameters.

Claims

1. A method for regulating the dosing of a reducing agent in an SCR catalyst, the method comprising: Determine the rated effectiveness η for the catalytic converter SP ; According to the rated effectiveness η SP Calculating an initial dosage V for the reducing agent; Determine the current value of the switching function S of the sliding mode controller, wherein the switching function is based on the setpoint effectiveness η SP and the actual effectiveness η PV The current deviation Δη between and the time derivative of the current deviation, wherein the switching function S is defined as follows: in Δη is the rated effectiveness level η SP and the actual effectiveness η PV The deviation between D=d SP -or PV ; represents the time derivative of Δη, and T 63 is the time constant; and The initial dosage V is adapted according to the value of the switching function S to obtain the resulting dosage r for the reducing agent NH3us , The calculation of the initial dosage includes: calculating a nominal fill level based on the nominal degree of effectiveness; calculating an expected loss of reducing agent within a predefined time period based on a catalytic converter model with the calculated target filling level; and The expected loss of reducing agent is defined as the initial dosage V, Wherein the adaptation of the initial dosage comprises: determining a value of the switching function S; and If the value of the switching function is greater than zero, the resulting dosage is obtained by increasing the initial dosage V by a predetermined excess amount M; and If the value of the switching function is less than zero, the resulting metering quantity is obtained by reducing the initial metering quantity V by a predetermined excess quantity M.

2. The method according to claim 1, wherein the adapting of the initial dosage comprises: The dosage is calculated by the following formula: where r NH3us is the resulting metering of the controller, S is the switching function of the sliding mode controller, M is the predetermined excess quantity, V is the initial metering quantity, and ε is a predetermined value close to zero.

3. The method according to claim 2, wherein the method further comprises: Prior to the calculation of the resulting dosage, the initial dosage V is multiplied by an adaptation factor, wherein the adaptation factor is defined by the following formula: Therein, f is the adaptation factor; F is a constant which is optionally related to the temperature and the mass flow in the catalyst; S is the switching function of the sliding mode regulator; and sgn() denotes a sign function.

4. The method according to any one of claims 1 to 3, wherein the excess amount is determined based on one of the following: - the excess is predetermined by a characteristic diagram which is dependent on the mass flow and / or the temperature in the catalytic converter; - calculating the excess amount as a function of a difference between a setpoint filling level and an actual filling level of the catalytic converter, wherein the setpoint filling level is calculated as a function of the setpoint effectiveness; - calculating the excess amount based on the difference between the rated effectiveness and the actual effectiveness; - forming a minimum value from a plurality of predefined excess quantities and / or calculated excess quantities.

5. The method according to any one of claims 1 to 3, wherein the method further comprises: The setpoint level of effectiveness η is determined as a function of an initial setpoint level of effectiveness and at least one of a predetermined setpoint filling level of the catalytic converter and a predetermined maximum filling level of the catalytic converter. SP .

6. The method according to claim 1, wherein at least part of the input values ​​of the method are predetermined by at least one characteristic curve field (100, 110, 120, 340, 350).

7. The method according to any one of claims 1 to 3, wherein a locally differentiated catalyst model is at least partially used to calculate at least one filling level and / or effectiveness of the catalyst, wherein the catalyst is divided into a plurality of axial segments in the catalyst model. 8 . A computing device for regulating the metering of a reducing agent in an SCR catalytic converter, the computing device being designed to carry out all method steps of the method according to claim 1 .

9. A computer program product, comprising a computer program which, when executed on a computing device, causes the computing device to perform all method steps of the method according to any one of claims 1 to 7.

10. A machine-readable storage medium having a computer program stored thereon, which, when executed on a computing device, causes the computing device to execute all method steps of the method according to any one of claims 1 to 7.

Images