DE102018216094B3 - Method for regulating an exhaust gas aftertreatment system and internal combustion engine, set up to carry out such a method - Google Patents

Abstract

The invention relates to a method for regulating an exhaust gas aftertreatment system (7) for reducing nitrogen oxides in an exhaust gas stream treated with the exhaust gas aftertreatment system (7), a metering device (13) for metering a reducing agent into the exhaust gas stream upstream of a catalytic converter (9) being activated, wherein the metering device (13) is controlled as a function of a target loading size of the catalyst (9), the target loading size being determined on the basis of a phenomenological model of a relationship between a relative loading of the catalyst (9) and a conversion rate of the catalyst (9), and wherein Control of the metering device (13) a physical memory balance is used.

Description

The invention relates to a method for regulating an exhaust gas aftertreatment system for reducing nitrogen oxides, and to an internal combustion engine which is set up to carry out such a method.

Methods for regulating an exhaust gas aftertreatment system for reducing nitrogen oxides, also referred to as SCR systems for short, are typically based either on simplified, control engineering replacement models, for example proportional elements with or without dead time and / or delay elements, or - alternatively - on detailed physical models. Both approaches have difficulties. Control engineering replacement models can be easily determined, for example by so-called jump tests, but due to their simplicity only describe the dynamic behavior of an SCR system approximately. In particular, they find their limits in transient operating states of the SCR system. In contrast, detailed physical models generally describe the system behavior correctly. However, they are based on a large number of system parameters that can only be determined with difficulty and with great effort. The use of such models in series control units for internal combustion engines may not be possible due to the required computing times. Therefore, simplifying assumptions are made in part; For example, a spatial expansion of an SCR catalytic converter in the flow direction is neglected during the implementation. In this case, however, the physical models lose accuracy. It is also problematic that physical models are very difficult to adapt to a changed real system behavior, since, in particular due to the large number of model parameters, it remains unclear which parameters can be meaningfully adapted to the changed, real behavior of the SCR system.

Ultimately, it is in particular not possible to use the approaches discussed here to specifically influence the storage loading of an SCR catalytic converter with a reducing agent, in particular ammonia. In the case of the simplified, control-technical replacement models, this fails due to the lack of dynamics or the insufficient physical accuracy, in which processes of ammonia storage are not shown; in the case of physical models due to the high level of complexity and the lack of knowledge as to which parameters are to be regarded as essential, the computing and storage effort on a control unit. The targeted influencing of the storage loading is desirable, however, since the system behavior of an SCR catalytic converter is strongly influenced by the loading of the catalytic converter surface with reducing agent.

Out DE 10 2012 203 539 A1 shows a method for correcting an estimated amount of NH ₃ stored in the SCR. In one example, the SCR efficiency is determined based on a NO _x sensor output, and the estimated amount of NH ₃ is corrected based on the SCR efficiency. Engine emissions can be reduced by improving the estimated NH ₃ storage level, at least under some conditions.

The object of the invention is to create a method for regulating an exhaust gas aftertreatment system for reducing nitrogen oxides and an internal combustion engine set up to carry out such a method, the disadvantages mentioned not occurring.

The object is achieved by creating the subject matter of the independent claims. Advantageous refinements result from the subclaims.

The object is achieved in particular by creating a method for regulating an exhaust gas aftertreatment system for reducing nitrogen oxides in an exhaust gas stream aftertreated with the exhaust gas aftertreatment system, with a metering device for metering a reducing agent - including the metering of a reducing agent precursor product which is in the exhaust gas stream to the reducing agent is implemented - is to be understood - in the exhaust gas stream upstream of a catalytic converter, which is set up in particular for the catalytic reduction of nitrogen oxides, in particular for the selective catalytic reduction of nitrogen oxides, in particular as an SCR catalytic converter, the metering device being controlled as a function of a desired load size of the Catalyst is controlled. The target loading size is determined using a phenomenological model of a relationship between a relative loading of the catalyst and a conversion rate of the catalyst. A physical memory balance calculation is used to control the metering device. In this way, in particular, a physical modeling and a phenomenological modeling of the catalyst behavior are skilfully linked, the advantages of the phenomenological consideration on the one hand and the physical consideration on the other hand being combined with one another, the respective disadvantages virtually canceling each other out.

In particular, the phenomenological model enables the target load size to be determined simply and quickly, while the physical memory balance calculation enables the metering device to be controlled in a realistic and precise manner, even for the real system. A physical memory balance calculation is much less complex than a full physical model and can therefore also be calculated quickly and with little storage effort. The method is therefore particularly characterized by high dynamics, and it is also very suitable for transient operating states of the exhaust gas aftertreatment system.

A relative loading of the catalyst with the reducing agent, in particular a relative loading of a catalyst surface of the catalyst, is preferably used as the target loading variable. Alternatively or additionally, it is possible that a target conversion rate of a catalytic converter and / or a target nitrogen oxide concentration - which is to be understood in particular as a target nitrogen oxide partial pressure - in the post-treated exhaust gas stream downstream of the catalyst. The target conversion rate and the target nitrogen oxide concentration are also linked to the relative reducing agent loading of the catalytic converter and can therefore be used as target loading parameters.

In the context of the method, the relative loading of the catalyst is preferably influenced, in particular controlled, preferably regulated.

The target load variable is preferably determined as a function of the operating point, in particular as a function of a current operating point of an internal combustion engine having the exhaust gas aftertreatment system, using the phenomenological model. This is advantageous because the conditions for the reduction of nitrogen oxides on the catalytic converter strongly depend on the instantaneous operating point of the internal combustion engine, in particular on an instantaneous exhaust gas mass flow, an instantaneous exhaust gas temperature, an instantaneous raw nitrogen oxide concentration upstream of the catalytic converter in the exhaust gas stream, and / or further parameters.

The phenomenological model is preferably a heuristic and / or a stationary model of the relationship between the relative loading of the catalyst with reducing agent and the conversion rate of the catalyst. Such a model is both easy to set up and also quick, easy and calculable without a large amount of memory.

To control the metering device, the phenomenological model is preferably used in addition to the physical memory balance calculation. In this way, the metering device can be controlled with particularly high accuracy and dynamics.

According to a development of the invention, it is provided that a polynomial function, in particular a polynomial, of the conversion rate as a function of the relative loading of the catalyst is used as the phenomenological model. This represents a simple and, in particular, easily calculable form of the phenomenological model.

According to a preferred embodiment, the polynomial function is a third order polynomial. It has been found that this level of complexity is sufficient for a sufficiently precise description of the catalyst behavior, while at the same time the computation and storage effort is low.

wobei η(t) die zeitabhängige Umsatzrate des Katalysators ist, wobei die relative Beladung mit dem Symbol θ bezeichnet wird, wobei θ_ln(t) eine verkürzte Schreibweise für den natürlichen Logarithmus der zeitabhängigen relativen Beladung θ ist, und wobei p₀ , p₁ , p₂ und p₃ Modellparameter des phänomenologischen Modells sind, die heuristisch angenommen, oder - was im Folgenden noch näher beschrieben wird - bestimmt werden.Alternatively or additionally, the logarithmic relative loading of the catalyst is used as the variable of the polynomial function. If one looks at the conversion rate plotted against the logarithm of the relative loading, a functional relationship is shown that can be represented with good approximation as a third-order polynomial or can be adapted by a third-order polynomial. The natural logarithm of the relative loading is particularly preferably used as a variable. A function according to the following equation is preferably used as the phenomenological model: $$\eta \left(t\right)=p_0+{\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="DE102018216094B3\_0007.png"\; state="\{\{state\}\}">p</figure-callout>}}_1{\theta }_{ln}\left(t\right)+p_2{\left({\theta }_{ln}\left(t\right)\right)}^2+p_3{\left({\theta }_{ln}\left(t\right)\right)}^3.$$

where η (t) is the time-dependent conversion rate of the catalyst, the relative loading with the symbol θ is referred to, where θ _ln (t) is an abbreviation for the natural logarithm of the time-dependent relative load θ, and where p ₀ . p ₁ . p ₂ and p ₃ Model parameters of the phenomenological model are that heuristically assumed or - which will be described in more detail below - are determined.

According to a preferred embodiment, the reducing agent in the narrower sense is to be understood to mean ammonia, a reducing agent precursor product, in particular a urea-water solution, being introduced into the exhaust gas stream by the metering device, the reducing agent product being broken down or decomposing into the reducing agent in the exhaust gas stream. In particular, urea introduced into the exhaust gas stream decomposes into ammonia and water under the conditions present there.

wobei θ̇ die Ableitung der relativen Beladung nach der Zeit, α(t) die Dosierrate in Abhängigkeit von der Zeit, η(t) die Umsatzrate in Abhängigkeit von der Zeit, und δ(t) die Austragsrate in Abhängigkeit von der Zeit ist. K(t) ist ein zeitabhängiger Umrechenfaktor, in den insbesondere der zeitabhängige Abgasmassenstrom, die Stickoxidkonzentration im Abgas stromaufwärts des Katalysators in Abhängigkeit von der Zeit, das Katalysatorvolumen, die molare Masse des Abgases und die maximal mögliche Belegung des Katalysators mit Reduktionsmittel in Einheiten von Stoffmenge pro Volumeneinheit eingehen.According to a further development of the invention, the physical memory balance calculation includes at least one variable that is selected from a group consisting of a metering rate with which the reducing agent is metered into the exhaust gas stream via the metering device, the conversion rate of the catalyst, and a discharge rate , with which reducing agent stored in the catalyst is released from the catalyst. By taking into account at least one of these variables, in particular all of the variables mentioned, an exact, preferably complete, representation of the memory balance on a physical basis can be obtained. In particular, the memory balance, which is preferred for physical memory balance calculation, is considered as a function of the relative loading on the at least one size selected from the group mentioned above. The memory balance is particularly preferably described by the following equation: $$\dot{\theta }\left(t\right)=K\left(t\right)\left(\alpha \left(t\right)-\eta \left(t\right)-\delta \left(t\right)\right).$$

where θ̇ is the derivative of the relative load over time, α (t) the dosing rate as a function of time, η (t) the turnover rate as a function of time, and δ (t) the discharge rate is dependent on the time. K (t) is a time-dependent conversion factor that includes in particular the time-dependent exhaust gas mass flow, the nitrogen oxide concentration in the exhaust gas upstream of the catalytic converter as a function of time, the catalyst volume, the molar mass of the exhaust gas and the maximum possible use of reducing agent in units of the amount of substance per unit volume.

wobei hier M_A die Molmasse des Abgases, M_R die Molmasse des Reduktionsmittels, insbesondere Ammoniak, [NO_x]_v die Stickoxidkonzentration im Abgas stromaufwärts des Katalysators, ṁ_R der eindosierte Massenstrom des Reduktionsmittels, und ṁ_A der Massenstrom des Abgases (kurz: Abgasmassenstrom) ist. Dabei ist der Massenstrom des Reduktionsmittels durch eine Steuereinrichtung vorgegeben beziehungsweise eingestellt und daher bekannt, der Abgasmassenstrom kann gemessen oder anhand von Betriebsparametern des Abgassystems und/oder einer das Abgassystem aufweisenden Brennkraftmaschine berechnet werden, und die Stickoxidkonzentration stromaufwärts des Katalysators kann ebenfalls gemessen werden.The metering rate is defined as the rate based on the amount of nitrogen oxide in the exhaust gas, in particular as the mass of reducing agent based on the nitrogen oxide mass flow in the exhaust gas: $$\alpha \left(t\right)=\frac{{\dot{m}}_R{M}_A}{{\dot{m}}_A{\left[N{O}_x\right]}_v{M}_R}.$$

being here M _A the molar mass of the exhaust gas, M _R the molar mass of the reducing agent, in particular ammonia, [NO _x ] _v the nitrogen oxide concentration in the exhaust gas upstream of the catalytic converter, ṁ _R the metered mass flow of the reducing agent, and ṁ _A the mass flow of the exhaust gas (in short: exhaust gas mass flow). The mass flow of the reducing agent is predetermined or set by a control device and is therefore known, the exhaust gas mass flow can be measured or calculated on the basis of operating parameters of the exhaust system and / or an internal combustion engine having the exhaust system, and the nitrogen oxide concentration upstream of the catalytic converter can also be measured.

wobei hier [NO_x]_n die Stickoxidkonzentration im Abgas stromabwärts des Katalysators ist. Die Stickoxidkonzentrationen stromaufwärts und stromabwärts des Katalysators können durch geeignete Sensoren, insbesondere Stickoxidsensoren, gemessen werden.The turnover rate can also be determined by measurement according to the following equation: $$\eta =1-\frac{{\left[N{O}_x\right]}_n}{{\left[N{O}_x\right]}_v}.$$

where [NO _x ] _{n is} the nitrogen oxide concentration in the exhaust gas downstream of the catalyst. The nitrogen oxide concentrations upstream and downstream of the catalyst can be measured by suitable sensors, in particular nitrogen oxide sensors.

wobei hier mit θ_ln,opt eine (logarithmierte) Speicherbeladung bezeichnet wird, bei der die Umsatzrate η maximal ist, und die im Folgenden auch kurz als optimale Speicherbeladung bezeichnet wird, wobei die Umsatzrate η insbesondere bei höherer Speicherbeladung abnimmt. η_max ist die entsprechende, maximale Umsatzrate, die der optimalen Speicherbeladung θ_ln,opt zugeordnet ist. η(t) ist die momentane Umsatzrate des Katalysators, und d_R ist ein Parameter, der betriebspunktabhängig und abhängig von dem Material des Katalysators gewählt wird.The discharge rate δ (t) is preferably defined separately for two different loading areas of the catalyst, preferably according to the following equation $$\delta \left(t\right)=\{\begin{array}{cc}0& \text{For}{\theta }_{ln}\le {\theta }_{ln.Opt}\\ \frac{1}{d_R}\left(\eta \left(t\right)-{\eta }_{max}\right)& \text{For}{\theta }_{ln}>{\theta }_{ln.Opt}\end{array}.$$

being here with θ _{ln, opt} a (logarithmic) memory load is called, in which the turnover rate η is maximum, and which is also briefly referred to below as the optimal storage load, the conversion rate η decreasing, in particular, with a higher storage load. η _max is the corresponding, maximum turnover rate, that of the optimal memory loading θ _{ln, opt} assigned. η (t) is the current conversion rate of the catalyst, and d _R is a parameter that is selected depending on the operating point and the material of the catalyst.

According to a further development of the invention, it is provided that an instantaneous relative load - preferably as an actual load quantity - is determined using the phenomenological model by means of an estimation method. This allows the current relative loading during operation of the catalytic converter to be determined quickly and with little effort. A Kalman filter is preferably used as the estimation method, particularly preferably an extended Kalman filter (Extended Kalman Filter - EKF). The instantaneous relative loading is particularly preferably determined on the basis of the phenomenological model and the physical memory balance.

wobei Gleichung (6) erhalten wird, indem die Logarithmusfunktion der relativen Speicherbeladung gemäß der Kettenregel abgeleitet wird - wobei exp(-θ_ln(t)) schlicht dem Kehrwert der relativen Beladung entspricht - wobei für die zeitliche Ableitung der relativen Beladung die physikalische Speicherbilanz gemäß Gleichung (2) eingesetzt wird, wobei dann schließlich in diese physikalische Speicherbilanz für die Umsatzrate η(t) das phänomenologische Modell gemäß Gleichung (1) eingesetzt wird.In particular, the estimation method is preferably based on a system of equations which comprises the phenomenological model according to equation (1) as the initial equation and the following equation as the equation of state: $${\dot{\theta }}_{ln}\left(t\right)=\text{exp}\left\{-{\theta }_{ln}\left(t\right)\right\}K\left(t\right)\left(\alpha \left(t\right)-\left(p_0+{\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="DE102018216094B3\_0007.png"\; state="\{\{state\}\}">p</figure-callout>}}_1{\theta }_{ln}\left(t\right)+p_2{\left({\theta }_{ln}\left(t\right)\right)}^2+p_3{\left({\theta }_{ln}\left(t\right)\right)}^3\right)-\delta \left(t\right)\right).$$

where equation (6) is obtained by deriving the logarithm function of the relative storage load according to the chain rule - where exp (-θ _ln (t)) simply corresponds to the reciprocal of the relative loading - and for the time derivative of the relative loading the physical storage balance according to Equation (2) is used, and then finally into this physical memory balance for the turnover rate η (t) the phenomenological model according to equation (1) is used.

The estimation method can then be based in particular on the known metering rate α (t) and the measured turnover rate η (t) be performed.

Different estimation methods, in particular Kalman filters, are preferably used, depending on whether the current relative load is smaller or larger than the optimal memory load θ _{ln, opt} at which the maximum sales rate η _max of the system is achieved. This takes into account in an advantageous manner that the conversion rate as a function of the relative loading is by definition maximum in the area of the optimal storage loading, so that the corresponding conversion rate curve is only bijective in some areas, namely on the one hand in the area of smaller relative loads than the optimal storage loading, and on the other hand in the area Larger relative loads than it corresponds to the optimal storage load. In order to compensate for this ambiguity, different estimation methods are used in the different areas, or at least different starting values are used for the relative loading, each of which - depending on the area - is smaller or larger than the optimal memory loading θ _{ln, opt} , Both methods can also be used on a trial basis, in which case it can be checked which value from the estimation method for the target load size, in this case preferably for the turnover rate, better matches a current measured value.

According to a development of the invention, it is provided that the metering device is controlled as an actuator of a control system, with this control regulating the conversion rate of the catalyst or a pollutant concentration, in particular a nitrogen oxide concentration, in the exhaust gas downstream of the catalyst as a control variable. The target load size is preferably used as a reference value. The target conversion rate of the catalyst is particularly preferably used as the target loading variable. Alternatively, it is possible that a target nitrogen oxide concentration downstream of the catalytic converter is used as the reference variable and target load variable. However, it is also possible for the relative loading itself to be used as the target loading variable, controlled variable and reference variable.

The target conversion rate as the target load variable and thus the reference variable is preferably determined from an operating point-dependent emission target value for the target nitrogen oxide concentration downstream of the catalyst, in particular depending on an instantaneous raw nitrogen oxide concentration upstream of the catalyst and preferably an instantaneous output of an internal combustion engine having the exhaust gas aftertreatment system and / or an instantaneous one Exhaust mass flow calculated. The relative target load As a guide, it can be determined from the target turnover rate determined in this way by back calculation - in particular using the Cardanic formulas.

The target nitrogen oxide concentration can be determined either directly from the emission target value or by recalculation from the relative target load using the phenomenological model of the target conversion rate.

The regulation is preferably carried out by a controller based on the physical memory balance, preferably on the basis of the physical memory balance and the phenomenological model. The controller can preferably be a PI controller (proportional-integral controller).

As an alternative or in addition, the metering device is preferably controlled by a pilot control, a relative target loading of the catalytic converter and an instantaneous relative loading of the catalytic converter being included in the pilot control. The relative target loading is preferably determined from the phenomenological model, in particular in the same way as was previously described in connection with the control. The instantaneous relative loading of the catalyst is preferably obtained from the estimation method described above. The metering rate to be set by means of the pilot control is preferably determined on the basis of the physical memory balance. In a particularly preferred manner, a reduced storage balance is used to simplify the method, it being assumed that the discharge rate and preferably also the conversion rate are constant.

The metering device is preferably controlled both as an actuator of the control described above and by the pilot control described above. In this case, an output variable of the controller and an output variable of the precontrol are preferably linked to one another, in particular offset, preferably added, the metering device being controlled on the basis of the result of this operation.

According to a development of the invention, it is provided that at least one model parameter of the phenomenological model is determined by means of the estimation method. The estimation process allows the model parameters, in particular, in a simple and economical manner p ₀ . p ₁ . p ₂ and p ₃ , appreciate. In particular, it is possible for all model parameters to be estimated. However, it is also possible for only a part of the model parameters to be estimated, while another part of the model parameters is obtained in another way or is fixed constantly.

Alternatively, it is also possible for all model parameters to be fixed constantly.

The model parameters are in particular dependent on a current operating point and / or an aging state of the exhaust gas aftertreatment system.

If the model parameters are obtained from the estimation process, this is preferably carried out using a combined state and parameter estimation. The state estimate, i.e. the estimate of the current relative load, must be carried out on a different time scale than the parameter estimate. In particular, the state is estimated using the dynamics of the load balance, the model parameters being estimated using the dynamics of the thermal behavior of the exhaust gas aftertreatment system.

According to a preferred embodiment of the method, the current relative loading is estimated using a Kalman filter.

Alternatively or additionally, the at least one model parameter is preferably determined using the least squares method. This is possible in a simple and precise manner at the same time, since the phenomenological model according to equation (1) in the model parameters p ₀ . p ₁ . p ₂ . p ₃ , is linear. In particular, a linear least squares method can be used. In the course of the method, values for the model parameters, which originate from a simulation of the exhaust gas aftertreatment system or from test bench tests, are preferably initially taken, in particular for a first state estimation. These are preferably stored permanently. The instantaneous relative loading resulting from the state estimation is considered below in connection with the measured instantaneous turnover rate; in particular, related points of the instantaneously measured turnover rate and the estimated instantaneous relative loading are assigned to one another, so that pairs of values from the instantaneous Sales rate and the current relative loading are available. The phenomenological model according to equation (1) is then adapted to these pairs of values using the least squares method. In order for this to work with good results, the exhaust gas aftertreatment system should have a suitable excitation, i.e. there should be sufficiently distributed points in both sizes, i.e. both for the turnover rate and for the current relative loading. At the same time, the method improves if the respective pairs of values are measured in stationary operating states of the exhaust gas aftertreatment system.

In order to further improve the reliability of the least squares method, at least two points, that is to say at least two pairs of values, are preferably predetermined and are included in the adaptation of the phenomenological model. This is preferably a first pair of values with a minimal turnover rate and minimal relative loading, it being possible in particular for both values to be chosen ad hoc to zero. A pair of values at the slip limit is preferably chosen as a second fixed pair of values, that is to say where reducing agent is released from the catalyst for the first time. This pair of values can be determined, in particular, in test bench tests, in particular the slip conversion rate and the slip loading being determined where a predetermined reducing agent concentration in the exhaust gas upstream of the catalytic converter is detected for the first time by means of suitable sensors. As an alternative or in addition to this second pair of values, a pair of values which is above the slip limit can also be included, this pair of values being able to be arbitrarily determined, for example, from a simulation of the exhaust gas aftertreatment system.

The at least two defined pairs of values in particular form limitations for the least squares method. On the basis of the state estimate and / or by means of the model parameters determined on the basis of the least squares method, the pair of values from which the turnover rate and the relative loading are at which the turnover rate is maximum is then preferably determined. Since this is the extremum of the phenomenological model according to equation (1), the additional condition applies that the derivation of the phenomenological model is zero. The pair of values determined in this way from the maximum conversion rate and the associated relative loading - namely the optimal storage loading, see above - can then be used in the context of the regulation of the exhaust gas aftertreatment system, in particular to decide whether the current relative loading is greater or less than that of the maximum conversion rate assigned optimal memory loading.

The defined value pairs are also dependent on the operating point, in particular dependent on an instantaneous exhaust gas mass flow and an instantaneous temperature of the catalytic converter. They are preferably stored depending on the operating point and are not changed during the lifetime of the exhaust gas aftertreatment system.

Only with regard to the slip limit, that is to say the value of the relative load, from which the reducing agent is discharged from the catalytic converter, can an alternative to an adjustment be made during operation in accordance with a predetermined rule, the slip load being determined, for example, by a fixed one Factor greater than 1 multiplied by the optimal memory load, which is assigned to the maximum conversion rate, the factor being 110%, for example.

According to a development of the invention, it is provided that the at least one model parameter is determined and stored depending on the operating point. As already stated, the at least one model parameter is preferably determined as a function of a temperature of the catalytic converter, an exhaust gas mass flow and / or a raw nitrogen oxide concentration in the exhaust gas upstream of the catalytic converter and is stored as a function of at least one of these parameters, in particular of all of these parameters , preferably stored in a map. To use the phenomenological model according to equation (1), the corresponding model parameters are then read out depending on the operating point.

When estimating the state or in the combined state and parameter estimation, physically meaningful limitations of the state or the parameters can also be taken into account. The relative loading of the catalyst can be defined, for example, so that it takes on values between 0 and 1. If this limitation is taken into account in the estimation process, a better quality of the estimation of the condition can be achieved and / or implausible values can be prevented from being given as the result of the estimation. Sensible limitations of the model parameters, for example to certain sign areas, can be taken into account. This limits the degrees of freedom of the estimation method to physically meaningful value ranges. A bad or inaccurate estimate is prevented from resulting.

According to a development of the invention, it is provided that the relative target loading is dependent on the operating point, that is to say in particular depending on an operating point of an internal combustion engine with which the exhaust gas aftertreatment system is operated, in particular depending on a speed and a torque of the internal combustion engine, based on the phenomenological model from a target Sales rate is determined. In an advantageous manner, the target conversion rate is initially determined as a function of the operating point, in particular on the basis of an emission target value, with the target conversion rate being based on the phenomenological model, in particular by means of the so-called Cardanic formulas, which allow the algebraic resolution of the model relative target load is calculated back.

The emission target value for the amount of nitrogen oxide emitted, in particular in units of mass per energy, for example grams per kilowatt hour, is preferably read from a characteristic diagram on the basis of the current operating point of the internal combustion engine, the emission target value based on the current output of the internal combustion engine, the current exhaust gas mass flow, and the current raw nitrogen oxide concentration upstream of the catalytic converter is converted into the target conversion rate.

A stationary relative loading is preferably calculated using the phenomenological model, the relative target loading being obtained from the stationary relative loading as a limited relative target loading by limitation using a dynamic temperature model. This limitation of the relative target loading serves in particular to avoid slippage of reducing agents. In this respect, it has been shown that the storage behavior of the catalytic converter is strongly temperature-dependent. The dynamic temperature model is preferably used to carry out a time calculation for the expected temperature of the catalytic converter or in the vicinity of the catalytic converter. A current slip limit for the current operating state is then determined, in particular on the basis of the current temperature in the vicinity of the catalytic converter, and an expected slip limit for the temperature calculated in advance, which is to be expected in the future according to the predicted temperature development at a specified point in time. In order to safely avoid reducing agent emissions, the minimum as the minimum slip limit is selected from the current slip limit and the expected slip limit and used to limit the relative target load. This means in particular that the relative target load is equal to the stationary relative load as long as the stationary relative load is smaller than the minimum slip limit, the relative target load being limited to the minimum slip limit if the stationary relative load becomes larger than the minimum slip limit ,

Preferably, based on the phenomenological model, a limited target conversion rate and / or a limited target pollutant concentration in the exhaust gas is calculated from the — in particular limited — relative target loading. Again, this can be done based on equation (1). The limited target conversion rate and / or the limited target pollutant concentration is / are used as a reference variable (s) for the controller by which the metering device is controlled as an actuator.

According to a further development of the invention, it is provided that a dosing error for the dosing device is determined on the basis of a comparison of a model calculation of the relative loading and / or the storage balance with an estimate of the relative loading and / or the storage balance. This is based on the idea that the storage balance - taking into account the phenomenological model for the turnover rate - can be calculated algebraically on the one hand and on the other hand determined using the estimation method from the (faulty) measured values of the real system. If the algebraically obtained ideal value is related to the real value obtained from the estimation method, it can be concluded that there is a deviation that represents an error that can be assigned to the metering device and / or the measurement sensor system used, in particular the nitrogen oxide sensors. If it is assumed that the nitrogen oxide sensors function without errors or at most with one known error, this can in turn be used to infer the metering error of the metering device, in particular an injector error.

An advantage of the method proposed here is in particular the fundamental consideration of the physical system behavior in the form of the memory balance (equation (2)) and the consideration of the combination of relative loading and associated rate of conversion of the catalyst (equation (1)). By knowing the storage status and the model relationship between the relative loading and the conversion rate, the relative loading of the catalytic converter can be influenced in a targeted manner as part of a regulation of the exhaust gas aftertreatment system. It is thus possible to achieve a desired target value of the emissions downstream of the catalytic converter particularly quickly without generating undesirable secondary emissions, in particular reducing agent emissions.

The result is a physically correct description of the system via the memory balance and the phenomenological model. In contrast to an inaccurate parallel model calculation, the estimation of the current loading state of the catalyst is corrected using the estimation method. By determining the model parameters, the phenomenological model can adapt to the current system behavior without complex relationships, such as aging processes, having to be depicted in detail. A particularly fast emission control is possible, which can compensate transient processes of the system particularly well and at the same time minimizes secondary emissions.

Finally, the object is also achieved by creating an internal combustion engine which has at least one combustion chamber and an exhaust gas aftertreatment system for aftertreatment of exhaust gas from the at least one combustion chamber. The internal combustion engine has a control device that is set up to regulate the exhaust gas aftertreatment system according to the method according to the invention or a method according to one of the previously described embodiments. The advantages that have already been explained in connection with the method are realized in connection with the internal combustion engine.

The exhaust gas aftertreatment system of the internal combustion engine has, in particular, a metering device for metering a reducing agent into an exhaust gas stream upstream of a catalytic converter of the exhaust gas aftertreatment system, as well as the catalyst, which is preferably designed as a catalyst for reducing nitrogen oxides, in particular for selective reduction of nitrogen oxides, in particular as an SCR catalyst ,

It is possible for the exhaust gas aftertreatment system to have a blocking catalytic converter downstream of the catalytic converter, which is configured to oxidize reducing agent discharged from the catalytic converter. The phenomenological model described describes in particular the overall behavior of the combination of SCR catalytic converter and blocking catalytic converter.

The exhaust gas aftertreatment system preferably has at least one temperature sensor, in particular a first temperature sensor upstream of the catalytic converter and a second temperature sensor downstream of the catalytic converter.

The exhaust gas aftertreatment system preferably has at least one nitrogen oxide sensor, preferably a first nitrogen oxide sensor upstream of the catalytic converter and a second nitrogen oxide sensor downstream of the catalytic converter. The first nitrogen oxide sensor is preferably arranged upstream of the metering device. Typical nitrogen oxide sensors have a cross sensitivity, for example for NH ₃ emissions. This behavior can preferably also be mapped with the chosen phenomenological model.

The control device is operatively connected to the metering device for controlling it. The control device is preferably also operatively connected to the at least one temperature sensor and / or to the at least one nitrogen oxide sensor in order to obtain measured values from these sensors.

• 1 eine schematische Darstellung eines Ausführungsbeispiels einer Brennkraftmaschine mit einem Abgasnachbehandlungssystem;
• 2 eine schematische Darstellung einer Ausführungsform eines Verfahrens zum Regeln des Abgasnachbehandlungssystems, und
• 3 eine diagrammatische Darstellung eines phänomenologischen Modells zur Regelung des Abgasnachbehandlungssystems.

The invention is explained in more detail below with reference to the drawing. Show:

• 1 a schematic representation of an embodiment of an internal combustion engine with an exhaust gas aftertreatment system;
• 2 a schematic representation of an embodiment of a method for controlling the exhaust gas aftertreatment system, and
• 3 a diagrammatic representation of a phenomenological model for controlling the exhaust gas aftertreatment system.

1 shows a schematic representation of an embodiment of an internal combustion engine 1 who have an engine block 3 with at least one combustion chamber 5 as well as an exhaust gas aftertreatment system 7 has, which is set up for the after-treatment of exhaust gas from the at least one combustion chamber 5 ,

The internal combustion engine 1 is preferably designed as a reciprocating piston engine. In particular, it can have a plurality of preferably identical combustion chambers 5 have, in particular four, six, eight, ten, twelve, fourteen, sixteen, eighteen or twenty combustion chambers. Another, smaller or larger number of combustion chambers 5 is possible.

The exhaust aftertreatment system 7 exhibits a catalyst 9 that is set up to reduce nitrogen oxides. The catalyst is preferred 9 set up for the selective catalytic reduction of nitrogen oxides, in particular it is preferably designed as an SCR catalyst. In the exemplary embodiment shown here, the exhaust gas aftertreatment system has 7 downstream of the catalyst 9 - towards one of the combustion chambers 5 upcoming exhaust gas mass flow - a blocking catalyst 11 on, which is set up to oxidize one from the catalyst 9 discharged or through the catalyst 9 passing through reducing agent.

Upstream of the catalyst 9 is a dosing device 13 arranged, which is set up for metering a reducing agent in the exhaust gas stream. The catalyst 9 is set up to at least partially store the reducing agent, the reducing agent being in particular on a surface of the catalyst 9 can be added.

The internal combustion engine 1 also has a control device 15 on that with the dosing device 13 is connected to control them. In particular, the control device 15 in this way specify the amount of reducing agent - in particular reducing agent mass per unit of time - metered into the exhaust gas stream.

The control device 15 is particularly designed to carry out a method for regulating the exhaust gas aftertreatment system described in more detail below 7 , the dosing device 13 in particular as a function of a target loading size of the catalyst 9 is controlled, the target load size - in particular depending on the operating point - using a phenomenological model of a relationship between a relative loading of the catalyst 9 and a conversion rate of the catalyst 9 is determined, and being used to control the metering device 13 a physical memory balance calculation is used.

The control device 15 is preferably used to determine an operating point of the engine block 3 operatively connected to it, in particular at an instantaneous speed and an instantaneous load of the engine block 3 to investigate.

Furthermore, the embodiment of the internal combustion engine shown here has 1 a first nitrogen oxide sensor 17 upstream of the catalyst 9 , in particular upstream of the metering device 13 , and a second nitrogen oxide sensor 19 downstream of the catalyst 9 , in particular downstream of the blocking catalytic converter 11 , on. The internal combustion engine also points 1 a first temperature sensor 21 upstream of the catalyst 9 , in particular upstream of the metering device 13 , and a second temperature sensor 23 downstream of the catalyst 9 , in particular downstream of the blocking catalytic converter 11 on. The nitrogen oxide sensors 17 . 19 and the temperature sensors 21 . 23 are each with the control device 15 operatively connected, so that these measured values of the sensors are available.

The control device 15 is preferably also set up to - if necessary by means of a suitable sensor - an instantaneous exhaust gas mass flow in the exhaust gas aftertreatment system 7 to detect or - in particular on the basis of the current operating state of the engine block 3 - to calculate.

Within the scope of the method, a polynomial function of the turnover rate depending on the relative loading is used in particular as a phenomenological model. A third-order polynomial is preferably used as the polynomial function. Alternatively or additionally, the logarithmic relative load is used as the variable of the polynomial function. In particular, the above equation (1) is used as the phenomenological model.

The physical memory balance preferably includes at least one variable which is selected from a group consisting of a metering rate, the conversion rate of the catalyst, and a discharge rate of reducing agent from the catalyst 9 , The storage balance is preferably applied as a derivative of the relative loading of the catalyst 9 depending on the at least one size. In particular, the above equation (2) is preferably used as the physical memory balance.

The instantaneous relative loading of the catalyst is preferred 9 with reducing agent - in particular as the actual load size - based on the phenomenological model - and preferably also the storage balance - determined by means of an estimation method. In particular, this can be done in the control device 15 Known metering amount, in particular the current metering rate, and a means of nitrogen oxide sensors 17 . 19 certain conversion rate of the catalyst 9 be used. A Kalman filter is preferably used as the estimation method, particularly preferably an extended Kalman filter (Extended Kalman Filter - EKF).

Preferably the dosing device 13 controlled as an actuator of a control system, wherein the control rate of conversion of the catalyst 9 or a pollutant concentration, preferably a nitrogen oxide concentration, is regulated in the exhaust gas as a controlled variable. The target load variable is used in particular as a reference variable, the target load variable being, in particular, the relative load, the conversion rate or a target pollutant concentration, in particular target nitrogen oxide concentration, in the exhaust gas downstream of the catalytic converter 9 is used.

As an alternative or in addition, the metering device is controlled by a pilot control, with a relative target loading of the catalytic converter and an instantaneous relative loading of the catalyst in the pilot control 9 received. The target loading is preferably determined from the phenomenological model, the current relative loading being obtained from the estimation method. The pre-control takes place in particular on the basis of the physical memory balance. For this purpose, a reduced storage balance is preferably used, in which only the metering rate is variable, but not the conversion rate and the discharge rate, which are rather kept constant.

At least one model parameter of the phenomenological model is preferably determined using the estimation method. In particular, all model parameters can be obtained from the estimation process. The current relative load is preferably estimated using a Kalman filter. Alternatively or additionally, the at least one model parameter is determined by means of the least squares method, in particular on the basis of consideration or plotting of the estimated values for the relative loading on the one hand and the measured values for the turnover rate on the other hand.

The at least one model parameter is preferably dependent on the operating point, in particular depending on an exhaust gas temperature, an exhaust gas mass flow, and / or a raw nitrogen oxide concentration upstream of the catalytic converter 9 , determined and saved.

The relative target loading is preferably dependent on the operating point, in particular dependent on an instantaneous speed and an instantaneous load point of the engine block 3 , determined from the target sales rate using the phenomenological model.

A stationary relative loading is preferably calculated using the phenomenological model, the relative target loading being obtained from the stationary relative loading as a limited relative target loading by limitation using a dynamic temperature model, preferably using the phenomenological model from the limited relative target loading a limited target turnover rate and / or a limited target pollutant concentration in the exhaust gas is calculated, in particular as a control variable (s).

A comparison of a model calculation of the relative loading with an estimate of the relative loading of the catalyst is preferred 9 a dosing error for the dosing device 13 determined.

2 shows a schematic representation of an embodiment of a method for controlling the exhaust gas aftertreatment system 7 , This is done in one calculation step 25 based on an instantaneous speed n and an instantaneous torque M the internal combustion engine 1 , especially the engine block 3 , a stationary relative load θ _stat calculated. This is done in a limitation step 27 based on a thermal model 29 limited. Thereby, using the thermal model 29 a prediction for a future expected slip limit of the catalyst 9 taken, the expected slip limit is compared with a calculated current slip limit, and the stationary relative load is limited to the smaller value, selected from the current slip limit and the expected slip limit. In the limitation step 27 preferably go in an adaptation step 31 Adapted models obtained for the exhaust gas aftertreatment system 7 , in particular characteristics for the thermal behavior of the exhaust gas aftertreatment system 7 , especially taking into account the exhaust gas mass flow. From the limitation step 27 this results in a limited relative target load θ _s , This is done in a retroactive accounting step 33 based on the phenomenological model according to equation (1), calculated back to a target nitrogen oxide concentration in the exhaust gas and / or a target conversion rate of the catalytic converter 9 , which is used as a reference variable (s) for regulating the relative loading of the catalyst 9 serve, at least one size selected from these sizes in a comparison step 35 with a measured actual Nitrogen oxide concentration downstream of the catalyst 9 and / or a measured or from measured values of the nitrogen oxide sensors 17 . 19 calculated actual sales rate is / are compared, the comparison result in a controller 37 incoming, which is preferably designed as a PI controller and is based on the one hand on the physical memory balance and on the other hand on the phenomenological model.

The regulator 37 controls the dosing device 13 as an actuator.

At the same time, the limited relative target load goes θ _s also in a feedforward control 39 a, which also - in particular on the basis of a reduced memory balance while keeping the turnover rate and the discharge rate constant - a signal for controlling the metering device 13 generated. The control signals of the controller 37 and the pilot control 39 are preferred in a clearing element 41 offset against one another to form a calculated control signal, in particular added, and a controlled system 43 supplied, the metering device by means of the calculated control signal 13 is controlled, the controlled system 43 also the catalyst 9 and optionally the blocking catalyst 11 includes. From the controlled system 43 in turn result the actual conversion rate and / or the actual nitrogen oxide concentration downstream of the catalyst 9 ,

This and the current dosing rate go into the estimation process 45 one that is preferably based on the Kalman filter, in particular the physical memory balance and the phenomenological model for the estimation method 45 be used. The estimation method in turn results in an instantaneous relative actual loading θ _i that of the pilot control 39 is fed, the pilot control 39 their signal in particular on the basis of a comparison of the limited relative target load θ _s and the current relative actual load θ _i generated.

Furthermore result from the estimation procedure 45 the model parameters p ₀ . p ₁ . p ₂ . p ₃ for the phenomenological model, which in particular the adaptation step 31 be fed for adjustment.

3 shows a diagrammatic representation of a plot of the turnover rate η against the logarithm of the relative loading θ _ln of the catalyst 9 , the model curve shown here satisfying equation (1) and thus being a third-order polynomial.

Various excellent points are marked. In particular, points are marked here that can be used as fixed points when determining the at least one model parameter according to the least squares method, namely a first point P1 with a minimal turnover rate η _min and a minimal logarithmic relative load θ _{ln, min} , There is also a second possible fixed point P2 drawn in, which marks a slip limit, namely a logarithmic relative load θ _{ln, slip} as well as a slip turnover rate assigned to this η _slip , for the first time reducing agent from the catalyst 9 is carried out. This pair of values can be determined, in particular, depending on the operating point from test bench tests with suitable reducing agent measuring sensors.

Additionally or alternatively, a third point can be used as a further fixed point P3 Be determined beyond the slip limit, especially arbitrarily based on a simulation of the exhaust gas aftertreatment system 7 ,

After adapting the model according to equation (1) to pairs of values obtained by measurement of the conversion rate η on the one hand and the logarithmic relative loading θ _ln on the other hand - taking into account the specified points - the pair of values can finally be determined, which is the maximum turnover rate η _max as well as the logarithmic relative load assigned to it θ _{ln, opt} - also known as optimal memory loading - represented.

This optimal storage load θ _{ln, opt} can be part of the regulation of the exhaust gas aftertreatment system 7 are used in particular to decide on which branch of the curve according to the diagram of 3 the exhaust aftertreatment system 7 is currently operated, since the corresponding curve according to equation (1) does not cover its entire definition range, but only on the one hand to the left and on the other hand to the right of the optimal memory loading θ _{ln, opt} is bijective.

LIST OF REFERENCE NUMBERS

1

Internal combustion engine

3

block

5

combustion chamber

7

aftertreatment system

9

catalyst

11

blocking catalytic converter

13

metering

15

control device

17

first nitrogen oxide sensor

19

second nitrogen oxide sensor

21

first temperature sensor

23

second temperature sensor

25

calculation step

27

limiting step

29

thermal model

31

adaptation step

33

Recalculation step

35

comparison step

37

regulator

39

feedforward

41

clearing member

43

controlled system

45

estimation methods

Claims (10)

Method for regulating an exhaust gas aftertreatment system (7) for reducing nitrogen oxides in an exhaust gas stream treated with the exhaust gas aftertreatment system (7), wherein - A metering device (13) for metering a reducing agent in the exhaust gas stream upstream of a catalyst (9) is controlled, wherein - The metering device (13) is controlled as a function of a target loading size of the catalyst (9), wherein - The target load size is determined using a phenomenological model of a relationship between a relative loading of the catalyst (9) and a conversion rate of the catalyst (9), and wherein - A physical memory balance is used to control the metering device (13). Procedure according to Claim 1 , characterized in that a polynomial function of the conversion rate depending on the relative loading of the catalyst (9) is used as the phenomenological model, preferably - the polynomial function being a third-order polynomial, and / or - a logarithmic logarithmic function as a variable of the polynomial function relative loading of the catalyst (9) is used. Method according to one of the preceding claims, characterized in that at least one variable is included in the physical storage balance, which is selected from a group consisting of a metering rate of the metering device (13), a conversion rate of the catalyst (9), and an output rate of reducing agent the catalyst (9). Method according to one of the preceding claims, characterized in that an instantaneous relative loading is determined on the basis of the phenomenological model by means of an estimation method. Method according to one of the preceding claims, characterized in that the metering device (13) a) is controlled as an actuator of a control system, the control regulating the conversion rate of the catalytic converter (9) or a pollutant concentration in the exhaust gas, or the relative loading of the catalytic converter, as a controlled variable, and / or b) controlling it by a pilot control, wherein a relative target loading of the catalytic converter (9) and an instantaneous relative loading of the catalytic converter (9) go into the pilot control. Method according to one of the preceding claims, characterized in that at least one model parameter of the phenomenological model is determined by means of the estimation method, preferably a) the current relative loading being estimated using a Kalman filter, and / or b) the at least one model parameter using the Least squares method is determined. Method according to one of the preceding claims, characterized in that the at least one model parameter is determined and stored depending on the operating point. Method according to one of the preceding claims, characterized in that the relative target loading is determined as a function of the operating point and using the phenomenological model from a target conversion rate, preferably using the phenomenological model to calculate a stationary relative loading, the relative target loading being calculated from the stationary relative loading is obtained as a limited relative target loading by limiting using a dynamic temperature model, a limited target conversion rate and / or a predetermined target pollutant concentration in the exhaust gas preferably being calculated from the limited relative target loading using the phenomenological model. Method according to one of the preceding claims, characterized in that a dosing error for the dosing device (13) is determined on the basis of a comparison of a model calculation of the relative loading with an estimate of the relative loading. Internal combustion engine (1), with an exhaust gas aftertreatment system (7) and a control device (15), the control device (15) being set up to carry out a method according to one of the Claims 1 to 9 ,

Images