DE102007007945A1 - Internal combustion engine's recirculated exhaust gas rate adjusting method, involves producing feed forward control signal based on data contained in characteristic field, where data depends on act of fluid physics - Google Patents

Abstract

The method involves producing a feed forward control signal (14) based on data contained in a characteristic field (37), where the data contained in the characteristic field depends on act of fluid physics. A pressure drop drifting and lying based on the recirculated exhaust gas is provided. A correction signal is overlaid from another characteristic field to the feed forward control signal.

Description

The The invention relates to a method for adjusting an exhaust gas recirculation rate an internal combustion engine according to the preamble of claim 1.

The Problem of previous solutions for controlling the exhaust gas recirculation (EGR) on internal combustion engines is that the control at the same time must be done quickly and accurately to both the requirements to the running characteristics of the engine (no dropouts due to temporary EGR overdose due to EGR rate overshoot) as well as its exhaust emission (on the one hand, no NOx potential Give away by passing too slowly to an EGR target value or by EGR rate undershoot, on the other hand no soot emission peaks Generate EGR rate overshoot).

specially with increasing tightening of emissions regulations, in particular As for nitrogen oxides, one is often forced to interpret the AGR rates close to the limits, in terms of misfiring and soot emissions are barely acceptable, d. H. the Requirements for speed and accuracy of the EGR control increase continue on.

at Today's regulators, however, are in the balancing act between speed and Accuracy compromises needed. Do you make the controller faster, If you reach the dynamic mode after a change the EGR setpoint faster this new EGR setpoint, this buys but with a fiercer overshoot Goal out. Conversely, one achieves by slowing down the Regulator over or Unterschwingerfreiheit, for but will be nominal EGR rate changes (eg at load or Speed jumps) implemented only slowly.

This is - a bit weakened, but in principle continue - also for the controller concepts that have underpinned the control a pilot control, so that the control of the EGR controller from a sum of feedforward (operating point dependent and without time delay available) and controller output is fed. An embodiment of such a concept is in 4 shown.

in this connection the controller only needs to take over part of the control, because you have the necessary AGR actuator control with appropriate Interpretation of the input tax already meets reasonably.

The German patent application DE 10257031 A1 shows a method for adjusting an exhaust gas recirculation rate (EGR) of an internal combustion engine, in which an exhaust gas recirculation valve (ARV) is adjusted by means of an actuator acting thereon in response to requirements of the internal combustion engine. The exhaust gas recirculation rate (EGR) is roughly regulated in steps in a known manner by means of external exhaust gas recirculation in stages to a desired value. After the coarse control, it is determined whether the exhaust gas recirculation rate (EGR) has reached a predetermined accuracy. If the specified accuracy has not yet been reached after the coarse control, a fine adjustment takes place by means of internal exhaust gas recirculation.

described is therefore a solution to the problem that a particular Exhaust gas recirculation rate by means of the outer Exhaust gas recirculation not set exactly enough can be, as it is z. B. then results when the EGR valve not continuously, but only in discrete steps can be. However, it must be noted that here is the coarse regulation and the fine control accomplished by means of various controllers and that both types of exhaust gas recirculation are not equivalent or affect, since the means of internal Exhaust gas recirculation recirculated Exhaust gas has a higher temperature than that by means of external Exhaust gas recirculation recirculated Exhaust. In particular, this is the case when the outer Exhaust gas recirculation one or more exhaust gas cooler which is the rule. Even without explicit exhaust gas cooler points the exhaust gas recirculation path of each outer EGR heat losses on their surfaces, so one Cooling effect, on. Thus changes dependent the proportion of external or internal exhaust gas recirculation the quality of the cylinder filling. In consequence of it emissions, such as soot formation, may deteriorate, also the power or torque output, which in applications with full load EGR is also significant. At least, however, must by means of further control effort the quality of the cylinder filling be maintained or restored.

Furthermore show the German patent application DE 2633617 A1 and the associated patent specification DE 2633617 C2 a method for determining adjustment variables in the region of an internal combustion engine, including the exhaust gas recirculation rate and the like, in which one or more operating state variables are evaluated and at least one further operating state variable dependent on the variation of the respective control variable is included as the feedback actual value for controlling the respective control variable.

In this case, a map control, which contains the end data of the respective setting variable, is supplied to at least one of the detected operating state variables for outputting a precontrol value which completely describes the respective setting variable. The precontrol value is only temporarily from a long evaluating the actual value superimposed time control to its quantitative adjustment, the transitions between control and control operation are continuous, long-term errors are eliminated and at the same time there is an immediate reaction of the feedforward control to changing operating state variables.

The Scripture shows no solutions for the specific Special features and problems of EGR control, such as For example the problem of dependence on the driving pressure gradient.

Of the present invention is based on the object, starting from an improved method for adjusting the Specify exhaust gas recirculation rate of an internal combustion engine.

Solved This object is achieved by a method with the specified in claim 1 characteristic features.

The invention shows a method for setting an exhaust gas recirculation rate of an internal combustion engine, wherein the adjustment by means of a controller ( 31 . 39 . 45 ) with superimposed control loop ( 2 ) or a regulation ( 2 ) with pilot control ( 31 . 39 . 45 ) and a pilot control signal ( 14 . 40 ) is generated on the basis of data which in characteristic curves or maps ( 37 ) are included.

Inventive features are that the data from the characteristic curves or characteristic diagrams ( 37 ) based on laws of fluid dynamics and in particular take into account a driving pressure gradient underlying the exhaust gas recirculation.

The inventive method has opposite The prior art has the advantages that even in the dynamic Motor operation a fast response and high precision the regulation are given. Here are the speed and accuracy the regulation according to the invention in particular attributed the absence of over- and undershorts. Furthermore, by the invention at the same time a better emission behavior (especially NOx and soot) and better runnability (eg Misfiring) of the engine guaranteed.

Further advantageous embodiments of the invention are in further claims, the description or the figures.

The Invention will now be based on an embodiment below With the help of the drawing explained.

there demonstrate:

1 A first embodiment of a novel controller concept for setting an exhaust gas recirculation rate (EGR rate),

2 a first exemplary embodiment or extension of the feedforward control 1 to a learning function,

3 a second exemplary embodiment or extension of the feedforward control 1 one opposite the 2 simplified learning function and

4 (Prior Art) An embodiment of a known controller concept for adjusting an exhaust gas recirculation rate (EGR rate), in which a control of a first and a second pilot control are subordinate.

The Invention is particularly suitable for controlling the exhaust gas recirculation an internal combustion engine, preferably a diesel engine with charging.

In 4 (Prior Art) is an embodiment of a known controller concept 1 for adjusting the exhaust gas recirculation rate (EGR rate), in which a control 2 a first feedforward control 3 (EGR pilot control) and a second pilot control 4 (EGR precontrol correction) are subordinate. A target EGR rate 5 is used as the setpoint or reference variable of an addition or summation point 6 supplied, as well as an actual EGR rate 7 , which can be determined for example from a simulation model and on the basis of measurement data (temperature, pressure, air flow), such. In DE 10229620 A1 and DE 10242234 A1 proposed.

A resulting control difference 8th becomes an EGR controller 9 , which is designed for example as a PID controller supplied. A control value 10 (Controller signal) as output signal of the controller 9 becomes another addition or summation point 11 supplied as well as a pilot control signal 14 , A sum 12 (Controller output signal) as an output signal of the addition or summation point 11 and the pilot control signal 14 becomes an EGR actuator 13 supplied as part of a not fully illustrated controlled system. As EGR actuator 13 For example, a stepping motor may be used which adjusts a controllable damper (eg, EGR valve) in an EGR passage.

The first pilot control 3 As input signals, for example, an engine speed 15 and a burden 16 (Torque, injection mass or quantity) supplied. With the help of a map 17 is based on the input signals 15 and 16 an operating point dependent EGR feedforward base signal 18 generated. In the simplest embodiment of the pilot control, if no further pilot control is added, are the Vorsteuerungsgesamtsignal 14 and the pre-control base signal 18 identical. In the embodiment illustrated here, the pilot control signal is set 14 from the pre-control base signal 18 and one of the second feedforward control 4 originating precontrol correction signal 19 together; the pilot signals 18 and 19 be in a further addition or summation 20 added.

The EGR pre-correction 4 As input signals, for example, turn the engine speed 15 and the load 16 supplied, furthermore the atmospheric pressure (height) 21 , the engine temperature 22 (eg cooling water, engine oil temperature or a characteristic combination thereof), the air temperature (outside, charge air or intake manifold temperature) 23 and possibly more signals 24 , From these input signals 15 . 16 and 21 to 24 is for example by means of one or more data fields 25 (Maps, curves) the Vorsteuerungskorrektursignal 19 generated and the addition or summation 20 fed.

Such a feedforward correction has z. For example, the purpose is: For the "normal conditions" for which the EGR pre-control base signal 18 from the map 17 is designed to deviate conditions z. At atmospheric pressure (altitude) 21 and temperatures 22 and 23 , as well as possibly further signals 24 Other engine EGR set rates are specified by the engine control unit, so other EGR flow rates must be set via the EGR valve. However, if one were to maintain the feedforward control, this would not match the modified desired EGR value and the EGR controller would be burdened with a disadvantageously high need for adjustment. Through the EGR pre-correction 4 this problem should already be prevented by more appropriate feedforward control.

Such rule structures 1 like the ones in 4 exemplified, allow in comparison to a mere control (without subordinate feedforward), a better design of the controller 9 in terms of the balance between speed and accuracy (minimized overshoot and undershoot), but even these are no longer optimal for future applications in this regard.

This is mainly because their feedforward 3 must be programmed depending on the operating point, but not the physical properties of the EGR line or the EGR valve accordingly.

Disadvantage of this operating point-dependent, but not the physics replicating procedure is that any change in the change in the EGR setpoints or their sizes 21 . 22 . 23 and 24 dependent corrections in the course of development again an adjustment of the feedforward control and the input tax corrections.

Currently, for example, you put in a map 17 The engine speed and load (the latter depending on the control unit philosophy as the injection mass, quantity or torque) are the control values for the EGR controller belonging to the respective operating points 13 , such as on a representative engine ("Average Engine" with "Average Parts") under normal operating conditions (engine warm, normal altitude operation, air cleaner clean, particulate filter clean, ...) to set the desired EGR rates required are.

Such an "average engine" would then, under these ideal boundary conditions, at least in steady-state operation, by the feedforward alone 3 Achieve exactly the desired EGR rates in each speed or load point, without the controller 9 would have to intervene.

• • Abweichung von den normalen Umgebungsbedingungen (Höhe, Umgebungslufttemperatur),
• • Abweichungen der Motortemperatur (61, Kühlwasser, Temperatur der Ladeluft- oder im Saugrohr, ...),
• • Abweichungen der Bauteile von der Mitte ihrer Streuung (z. B. Durchflusskennlinie des AGR-Stellers, Druckverluste der gasführenden Teile und/oder der AGR-Strecke, Wirkungsgrade eventueller Turbinen und Verdichter mit deren Auswirkungen auf den sich bei vorgegebenem Ladedruck p2 einstellenden Abgasgegendruck p3, Schluckvermögen des Motors, Verschmutzungsgrad gasführender Teile und/oder der AGR-Strecke, ...),
• • Beladung von Luft- oder Partikelfilter und
• • Instationären Betrieb des Motors.

The superimposed controller 9 So it still remains the task, the deviation of the required AGR actuator control 12 from the pilot control value 14 to adjust, z. B. can result from:

• Deviation from the normal ambient conditions (altitude, ambient air temperature),
• • deviations of the engine temperature ( 61 , Cooling water, temperature of the charge air or in the intake manifold, ...),
• • deviations of the components from the center of their dispersion (eg flow characteristic of the EGR valve, pressure losses of the gas-carrying parts and / or the EGR path, efficiencies of any turbines and compressors with their effects on the exhaust gas back pressure p3, which occurs at a given boost pressure p2 , The engine's ability to swallow, the degree of contamination of gas-carrying parts and / or the EGR route, ...),
• • Loading of air or particle filters and
• • Transient operation of the engine.

In a more complex design of the pilot control 3 you can get that from the "normal" feedforward 3 to a foreseeable extent different pre-taxation requirements for some of these deviations. For example, it would be conceivable to correct the pilot control data depending on input variables known to the engine control unit, such as atmospheric pressure (altitude), air and engine temperature and particle filter charge.

Nevertheless, the remaining sizes, the deviations do not remain for such a feedforward control 3 can be foreseen and fed. In addition, it is usually only possible to correctly describe the dependencies of individual deviating influences, ie store them in further corrections of the precontrol, but not their all interactions in the simultaneous occurrence of several or all of these deviating influences. This means that a correct "anticipation" of their effect is not possible even with sufficiently complex input tax corrections.

Of the deviations enumerated above, not only, but above all the transient operation of the engine, there are numerous significant disturbances for the control, which themselves are based on the regulator concept as described above 1 with complex subordinate feedforward control 3 . 4 working regulator 2 inevitably led to AGR overshoots and undershoots.

Such happens z. B. in engines with Aufladekonzepten that prepare inhomogeneous transitions by z. B. certain operating lines certain charging members, resonant volumes, Saugrohrlängen, exhaust valves, or the like on, off or switched. Examples would be z. As the transitions in multi-stage charges between their individual operating areas, or the hard switching on and off of individual superchargers (compressors, compressors, ...). This almost inevitably leads to a sudden jump in the driving pressure gradient across the EGR line, leaving the AGR controller 13 If the control was unchanged, suddenly a different EGR rate would occur.

BE REDUCED said switch always on the same speed / load lines in the operating map of the engine, so you could influence in a conventional, d. H. operating point-dependent Feedforward control even into it, but actually The operating lines of such switches are also subject to certain Shifts, z. B. by intentional hysteresis, corrective influences by altitude or certain temperatures, etc., or are even self-regulating (such as the circuit of the compressor bypass on the two-stage charged OM646-NCV3 by means of a check valve), so that the engine control unit does not know exactly when the switchover takes place.

A Another disturbance is the burning of the Particulate filter that runs faster than the loading. In this case, the exhaust back pressure falls during the burnup in a short time clearly, which is also the driving pressure gradient over the EGR route is affected. Also for vehicles with - for example for acoustic reasons - switching valves in multi - flow Exhaust systems happens similar, there even leaps and bounds.

In engines with bypassable EGR coolers, there are similar effects in switching the cooler bypass. The sudden bypassing of a radiator suddenly lowers the flow resistance of the EGR path and thus increases with initially constant control of the EGR valve 13 unintentionally the EGR rate (and vice versa).

If one wanted all these disturbing effects also in a Vorsteuerkororrektur 25 packing in according to an existing pattern, this would be almost infinitely complicated, unmanageable, only available at immense expense, and thus impractical.

To In any case, you would only be in able to predict the predictable disturbing effects (such as the transition from single-stage to two-tier loading area), but not the "classical" instationary effects such as the different rate of change Exhaust back pressure and boost pressure at a speed or load jump on a charged engine. This different pace has yes a variable course of the AGR-driving pressure gradient over the time after speed and / or load jump result, the one not Foreseeable and therefore not subject to input tax adjustments Tracking the actuator control by the controller necessary power.

The 1 shows a first embodiment of a novel controller concept 30 for adjusting an exhaust gas recirculation rate (EGR rate). Here is the scheme 2 (identical to the prior art according to 4 ) a pilot control according to the invention 31 underlain.

The pilot control according to the invention 31 has three input quantities 32 . 33 and 34 on, which is the input size 32 is a pressure p3 (exhaust backpressure before turbine), at the input 33 by a pressure p2 (boost pressure) and at the input value 34 by a quantity V_AGR_soll (target EGR volume flow to be set). The pressure p3 32 and the pressure p2 33 become an addition or summation point 35 fed to a differential pressure 36 from exhaust back pressure before turbine (p3) 32 and boost pressure (p2) 33 as the driving pressure gradient p3 - p2. This differential pressure (or this EGR-driving pressure gradient p3 - p2) 36 and the input size 34 become a simple map [z = f (x, y)] 37 which has the physically relevant and always correct relationship for a suitable pilot control variable, the pilot control signal 14 , contains.

The inventive design of the EGR pilot control 31 according to 1 can be used in a meaningful design to learn functions, as exemplified in 2 and 3 are shown to be extended.

The novel pilot control 31 is no longer based on operating-point-dependent characteristic curves or characteristic diagrams (supplemented, if necessary, by numerous precontrol corrections for numerous states deviating from the standard design case, see 4 ), but it takes into account the requirements of exhaust gas recirculation given by fluid physics, for example the pressure gradient p3 - p2.

behind it is the approach that the AGR-Steller so ultimately the task serves to dose the flow of EGR gas through an EGR path so that a certain, already known to the engine control unit, or at least the engine control unit calculable setpoint Exhaust gas flows through the EGR path.

This EGR gas flow is determined physically by the pressure gradient driving it, the flow resistance of the EGR path and the position of the EGR valve (controllable throttle element, eg EGR valve) therein:
EGR flow rate = f (pressure drop, flow resistance)

The EGR flow rate is the one to be set by the actuator Target size. The pressure gradient is the Control unit known, as explained below is, and the flow resistance is alone by the Position of the AGR controller modulated, ie by its control. (This at least applies to the short-term change the flow resistance of the EGR route as opposed to about to the gradual change due to pollution).

So the control of the EGR valve to be made by the engine control unit at the respective moment is given by pressure gradient and target EGR flow rate:
EGR actuator activation = f (pressure drop, target EGR flow rate).

Of the Flow resistance of the EGR route is determined by the geometry given the relevant components and so - apart of their diversification in manufacturing and pollution - only depends on the position of the EGR controller. The AGR controller is based exactly on the principle, in the EGR route over a variable opening cross section the flow resistance specifically designed to be variable.

at many types of AGR, at least those that are suitable for Applications with high exhaust demand, so high precision requirement recommend to the AGR, there is a fixed assignment between the signal of the engine control unit for actuator control and the setting (ie the released by the controller Flow cross-section). With these EGR controllers you can establish a fixed relationship between the actuator control from the engine control unit and the flow resistance the EGR route.

This is not the case when the control of the EGR controller by means of electro-pneumatic converter via a vacuum box on an EGR Tellerhubventil done. This is by the drive signal only the actuator force, but not an actuator position or an opening cross-section specified. With the same AGR actuator control, the Cross section then different, depending on others Influencing factors such as frictional forces in the AGR actuator guidance and gas forces on the Actuator.

The EGR driving pressure gradient across the EGR line corresponds to the usual design of the EGR route on turbocharged engines (removal before turbine, feeder after compressor or - if available - after Intercooler) the differential pressure from exhaust back pressure before turbine (p3) and boost pressure (p2), or is at least by this differential pressure characteristically described because you have certain, z. B. in the arrangement of the p2 and p3 sensors justified deviations from the Pressure values directly at the inlet or outlet of the EGR route as systematic may accept.

There the modern engine concepts already have p2 and p3 sensors anyway, is the information about the characteristic AGR-driving pressure gradient p3 - p2 so before or can in the engine control unit easily calculated.

The desired EGR flow rate is also calculable in the engine control unit. This is known at least for engines whose EGR control uses the EGR rate as a reference variable ( DE 10229620 A1 . DE 10242234 A1 ). In these cases, the target EGR rate is available at all times (ie also in the case of the currently existing system deviation).

• a) Berechnung aus der Füllung wie folgt: Gemäß der Offenlegungsschrift DE 10242233 B3 liegt zu jedem Betriebspunkt (Drehzahl n_Mot und Last) im Motorsteuergerät stets ein Wert für den Füllungsgrad des Motors ☐a vor, und zwar dank Lernfunktionen wie ILA (individuelle Luftaufwands-Adaption, siehe DE 10242233 B3 ) stets sehr genau.

From the desired EGR rate, the desired EGR mass flow can be calculated on the basis of the other information present in the engine control unit. There are two viable options for this calculation:

• a) Calculation from the filling as follows: According to the disclosure DE 10242233 B3 For each operating point (speed n_Mot and load) in the engine control unit is always a value for the degree of filling of the engine ☐a before, thanks to learning functions such as ILA (individual air effort adaptation, see DE 10242233 B3 ) always very accurate.

With knowledge of the boost pressure p2sr and the charge air temperature t2sr in the intake manifold can now dar from the target total mass air flow (fresh gas and EGR gas) mLges_soll be calculated by the engine in a known simple manner as:
mLges_soll = f (λa, n_Mot, p2sr, t2sr).

By means of the desired EGR rate rAGR_soll, the desired EGR mass flow mAGR_soll from the definition of the EGR rate is obtained therefrom: mAGR_soll = rAGR_soll · mLges_soll.

Now you have to consider the following small inaccuracy when using this method on engines with the most common sensor array today:
Although p2sr is known by the p2 sensor, the charge air temperature in intake manifold t2sr required for the calculation of the target total air mass flow (fresh gas and EGR gas) by the engine is a mixed temperature from the charge air temperature after charge air cooling, but before EGR admixture, and the temperature of the EGR gas at the admixing point, t2sr is thus itself dependent on the EGR rate.

In conventional sensor arrangement, the T2 is only before the EGR admixing point as a measured value. However, the temperature of the EGR gas at the admixing point - and thereby also the mixed gas temperature in the intake manifold t2sr according to the EGR rate - is present in the control unit when applying the invention to the EGR rate control only for the current actual EGR rate ( DE 10229620 A1 . DE 10242234 A1 ), but not for the EGR target rate.

this Ideally, the problem can be avoided by using either the T2 sensor instead before adding (or in addition to) a temperature sensor for the temperature T2SR placed in the suction tube.

Alternatively, it may easily be assumed that the temperature of the EGR gas before the admixing point is known by the change of the EGR rate from the actual EGR rate (for this, this temperature is known from the T3 and EGR cooler models, DE 10229620 A1 . DE 10242234 A1 ) changes to the desired EGR rate relatively little. Then one needs to calculate the influence of the EGR rate jump from rAGR_ist to rAGR_setpoint only with respect to the correspondingly changed mixing ratios according to the mixture equation, in which one simplifies assuming in the lower of the two following formulas that T_AGR_setpoint = T_AGR_act.

All quantities of the mixture equation for the current actual state with rAGR_ist are known: T2sr_ist = T2 + rAGR_ist (T_AGR_ist - T2)

The same applies to the target state with rAGR_soll: T2sr_soll = T2 + rAGR_soll (T_AGR_soll - T2)

Puts as explained above for T_AGR_soll approximately T_AGR_is on, so you are at least already for T2sr_soll reasonably good, especially considering that this calculation yes during the EGR control process (in which the EGR controller itself with its current rAGR_ist from the original rAGR_ist to rAGR_setpoint) with each computation cycle of the engine control unit is carried out again, that is by the simplifying Assumption T_AGR_soll = T_AGR_is conditional residual error with approximation of the rAGR_is getting smaller and smaller on rAGR_soll until finally becomes zero.

• b) Berechnung mittels annähernd Dreisatz wie folgt: Wenn man aus der obigen Überlegung, dass sich die durch vereinfachende Annahmen entstehenden Fehler in der Vorausberechnung des Soll-AGR-Massenstroms automatisch mit Annäherung der Ist-AGR-Rate an die Soll-AGR-Rate verringern, bereit ist, eine etwas größere Vereinfachung als im Fall a) zu treffen, dann kann man alternativ wie folgt vorgehen: Aus dem AGR-Kühlermodell ( DE 10229620 A1 , DE 10242234 A1 ) ist neben der Ist-AGR-Rate auch der AGR-Ist-Massenstrom bekannt. In der vereinfachenden Annahme gleichen Gesamtmassenstroms mL_ges durch den Motor (unabhängig von der AGR-Rate) lässt sich im Dreisatz mit der Soll-Rate der Soll-AGR-Massenstrom abschätzen wie folgt: mAGR_soll = mARG_ist·(rAGR_soll/rAGR_ist).

This can from the known computational relationships
mLges_soll = f (λa, n_Mot, p2sr, t2sr) and mAGR_soll = rAGR_soll · mLges_soll the EGR target mass flow mAGR_soll should be calculated sufficiently accurately.

• b) Calculation using approximately three sets as follows: If one considers from the above consideration that the errors in the prediction of the target EGR mass flow resulting from simplifying assumptions automatically decrease as the actual EGR rate approaches the target EGR rate , is willing to make a slightly greater simplification than in case a), then one can alternatively proceed as follows: From the EGR cooler model ( DE 10229620 A1 . DE 10242234 A1 ) In addition to the actual EGR rate and the actual EGR mass flow is known. In the simplistic assumption of the same total mass flow mL_ges by the engine (irrespective of the EGR rate), in the rule of three with the desired rate, the target EGR mass flow can be estimated as follows: mAGR_soll = mARG_ist · (rAGR_soll / rAGR_ist) ,

Across from the simplification in case a) (in the local subparagraph "no t2sr sensor available "), where it was only assumed the EGR rate difference between rAGR_set and rAGR_act does not affect the EGR gas temperature in front of the EGR admixing point, so here is also additional the influence of different sized depending on the EGR rate Amount of blended hot EGR gas to cool Fresh air T2 to the mixing temperature t2sr and thus their influence on the engine filling or the total gas mass flow through neglected the engine.

That is, in the temperature mixing equation T2sr = T2 + rAGR * (T_AGR-T2) in case a) only the influence of the EGR rate dependence of T_AGR on T2sr, in case b) additionally the influence of rAGR itself on T2sr is neglected - ie for the first moment, ie the first computation iteration at rAGR_soll not equal to rAGR_act , As rAGR_act approaches rAGR_setpoint, as a result of the precontrol and control process introduced thereby, the error made as a result, as stated, rapidly and finally shrinks to zero when the adjustment target is reached.

advantage This method is that no knowledge about the degrees of filling of the engine are needed, the computational effort accordingly is smaller.

Here, too, one can therefore use the known computational relationships:
mLges_soll = f (λa, n_Mot, p2sr, t2sr) and mAGR_soll = rAGR_soll · mLges_soll calculate the EGR target mass flow mAGR_soll sufficiently accurately.

In In both cases a) and b) the conversion is recommended the EGR mass flow into a characteristic EGR flow by means of characteristic temperatures and pressures in the EGR route. The conversion from mass to flow recommends because of the throttling characteristics of existing geometry a physical relationship between volume flow (and not Mass flow), pressure gradient and throttle geometry describe.

Of the Question which temperature and which pressure are used exactly to make the conversion, here comes a big Meaning too. If possible, these two sizes should be measured at the EGR valve, since this in the EGR control the Flow characteristic of the EGR route modulated. Because in practice however there are usually no pressure and temperature sensors may be arranged, may be a modeling these sizes will be helpful.

It So is for every single use case with his Edge sizes such as placement of the sensors and placement the EGR valve (before, after or between any existing EGR coolers) to decide pragmatically which of the sensors available as a sensor or calculable pressures and temperatures as adequate characteristic of this.

• • Bei AGR-Ventil vor AGR-Kühler: p3 und t3
• • Bei AGR-Ventil nach AGR-Kühler: p3 und t_AGR
• • Bei AGR-Ventil zwischen zwei AGR-Kühlern: p3 und (t3 + t_AGR)/2 oder – bei Verwendung von abschnittsweisen AGR-Kühlermodellen – eben dem für den Zustand zwischen beiden AGR-Kühlern im Modell berechneten t_AGR' (aus dem AGR-Kühlermodell der Erfindungen DE 10229620 A1 oder DE 10242234 A1 ist bei Aufteilung der AGR-Kühlstrecke in mehrere in Reihe geschaltete Abschnitte der Kühlstrecke die Temperatur am AGR-Ventil auch zwischen zwei Kühlern berechenbar, wenn die Grenzen der abschnittweise in Reihe berechneten Teilkühlstrecken passend gewählt werden).

It is conceivable z. B. in the presence of pressure and temperature sensor signals from the exhaust manifold (p3 and T3), a pressure sensor signal from the intake manifold (p2) and the knowledge of the EGR gas temperature at the end of the EGR path T_AGR (eg from the EGR Cooler model off DE 10229620 A1 or DE 10242234 A1 ) that, depending on the position of the EGR valve, the characteristic pressure p_charakt and the characteristic temperature T_charakt for the mass / volume flow conversion z. B. be determined as follows:

• • For EGR valve before EGR cooler: p3 and t3
• • For EGR valve after EGR cooler: p3 and t_AGR
• • For EGR valve between two EGR coolers: p3 and (t3 + t_AGR) / 2 or - when using sectional EGR cooler models - just the t_AGR 'calculated from the EGR cooler for the state between both EGR coolers (from the EGR cooler) Chiller model of the inventions DE 10229620 A1 or DE 10242234 A1 if the EGR cooling section is divided into several sections of the cooling section connected in series, the temperature at the EGR valve can also be calculated between two coolers, if the limits of the section cooling sections calculated in series are suitably selected).

The conversion of the desired EGR flow mass m_AGR_soll into the desired EGR flow rate V_AGR_soll takes place according to the gas equation: V_AGR_soll = m_AGR_soll · R · T_charact (in K) / p_charact

Herewith is then the EGR target volume flow V_AGR_soll before.

With knowledge of the relationship between the EGR volume flow V_AGR and the AGR-driving pressure gradient p3-p2 and the EGR actuator position (or control) determinable on a representative test engine with representative test parts (of medium value with respect to all EGR flow-relevant properties) now the pilot control map are bedatet as
AGR controller activation = f (p3 - p2, V_AGR_soll).

The Pilot control thus brings both stationary and dynamic very fast and accurate the EGR valve in the for adjusting the desired EGR rate correct position, and even in the presence all the disturbances, their potential for interference (for a conventional feedforward control) is that it interferes with the EGR-driving pressure gradient acts, so z. B. air filter contamination, particulate filter loading or Burnout, operation in height, switching in multi-level Charges, etc.

Also operating in conditions where the desired EGR rate is lower than normal operation gives way because they B. corrections dependent on ambient pressure or various temperatures is required for the feedforward no pilot controls more, since these influences are included here in the calculated and the novel feedforward control determining V_AGR_soll.

It remains superimposed for those of the EGR feedforward control EGR control therefore only the fine adjustment of such influences, itself the improved pilot control according to this Can not anticipate invention, because it does not affect the p3 - p2 still affect the V_AGR_soll, but z. B. on the flow characteristics the EGR route. So, in particular, the balance of copy scatters remains (eg in the characteristic "flow as a function of the actuator control" of the EGR actuator) or aging effects (such as polluting and thereby increasingly throttling EGR routes) a task of the superimposed EGR control.

Thereby, that the controller thus of many Störgrößenausgleichsaufgaben is relieved and relies on the said remaining tasks restrict the design of its control parameters much cheaper with regard to the goal, at the same time Speed and freedom from over or under-swingers to reach. It follows from the invention a structurally simple design of the pilot control with only two more Input variables (V_AGR_soll and p3 - p2).

It should be expressly pointed out that the pre-control proposed here after 1 itself can cover the abovementioned disturbance variables (scattering, soiling) remaining for the regulator if the feedforward control is extended by a learning function in another embodiment according to the invention (eg according to FIG 2 or in a simplified alternative form according to 3 ).

2 shows for this purpose a first exemplary embodiment or extension of the feedforward control 1 a learning function. With this controller concept 38 will the scheme 2 (identical to the prior art according to 4 ) a pilot control 39 underlain. In contrast to the pilot control 31 of the 1 be at the feedforward 39 the EGR-driving pressure gradient 36 and the target value of the EGR pilot amount V_AGR_soll 34 on the one hand the map 37 fed, the map 37 herein referred to as the EGR pre-tax basic map and an EGR pre-tax basic quantity 40 has as output.

On the other hand, the AGR-driving pressure gradient 36 and the target value of the EGR pilot amount V_AGR_soll 34 another map 41 fed. This map 41 contains as output variable so-called delta learning characteristic values 42 , which were learned during readjustments and allow the most accurate feedforward. These delta learning characteristics 42 be in a link 43 with the basic EGR input tax 40 linked, preferably added. However, a multiplicative link is also conceivable. From the EGR pre-tax basic size 40 together with an associated delta learning characteristic 42 this results in the total pilot signal 14 ,

Does it come despite the inventive, particularly advantageous embodiment of the pilot control according to 1 in sufficiently stationary, ie adjusted states, to a need of a controller intervention (ie the feedforward control does not meet the necessary control needs of the EGR controller correctly, but must be permanently supplemented by a controller), this is a sign of a systematic, so einlernbaren influence, and The Nachregelbedarf can be transferred by teaching for future operation in a suitable feedforward.

This happens in the example according to 2 by dropping this learning value 42 to the appropriate interpolation point of the EGR pre-tax delta learning map 41 , in the following example according to 3 by adding a corresponding "EGR pre-tax correction learning factor" 46 ,

The 3 shows a second exemplary embodiment or extension of the feedforward control 1 one - against the 2 simplified - learning function. In this regulator concept according to the invention 44 will the scheme 2 (identical to the prior art according to 4 ) a pilot control 45 underlain.

As in the pilot control according to the invention 31 according to 1 also has the feedforward control 45 three input variables, the pressure p3 (exhaust backpressure upstream of the turbine), the pressure p2 (boost pressure) 33 and the size V_AGR_soll 34 (Setpoint of EGR pilot control variable). The pressure p3 32 and the pressure p2 33 become the addition or summation point 35 fed to the differential pressure 36 from exhaust back pressure before turbine (p3) 32 and boost pressure (p2) 33 as the driving pressure gradient p3 - p2. This differential pressure (or this EGR-driving pressure gradient p3 - p2) 36 and the input size 34 be the map 37 fed, the map 37 here (as in 2 ) is referred to as the EGR pre-tax basic map and the EGR pre-tax basic variable 40 has as output.

Similar to the feedforward control 39 ( 2 ) is used for as exact a pilot control of the EGR pre-tax base size 40 by means of a linkage point 47 a correction factor 46 supplied preferably multiplicatively. However, an additive link is also conceivable. From the product of EGR pre-tax basic size 40 and associated correction factor 46 this results in the total pilot signal 14 ,

The correction factor 46 was learned during readjustments and enabled like the delta learning characteristics 42 ( 2 ) a precise precontrol.

In contrast to the delta learning characteristics 42 becomes the correction factor 46 but not taken from a map, as in the preferred multiplicative execution of the link 47 assumes that the pilot control map 37 only by factor must be scaled, so the relative learning needs at each operating point from the EGR flow rate and impelling pressure gradient is the same. From a sufficiently stationary learning process, it must be read, how large the ratio of the actual EGR actuator control (controller output signal 12 , so pilot signal 14 plus controller signal 10 ) to the mere pilot signal 14 is. (For additive execution of the link 47 must be read, which controller signal 10 is issued.)

If this ratio is 1, or if the control signal is zero, which are equivalent statements, is the feedforward control 45 alone just fine; There is then no need for correction. If, on the other hand, the ratio is not equal to 1 or the controller signal is not equal to zero (to avoid learning processes in the area of measuring accuracy, it would be sensible to select as a condition that this ratio must be different from zero by more than one threshold value to be established or there is a need for correction for the precontrol 45 , The to correct the pilot signal 40 necessary learning factor 46 then corresponds in the preferred multiplicative embodiment of the point of attachment 47 exactly the calculated ratio of the actual AGR actuator drive signal 12 (Feedforward control signal 14 plus controller signal 10 ) to the mere pre-tax value 14 , In the case of additive execution of the link 47 this corresponds to the correction of the pilot control signal 40 necessary learning factor 46 exactly the controller signal 10 ,

In order to avoid excessive, erratic learning steps (and that a singular event or a short-term disturbing signal causes a deception), not the full recognized correction need should be learned, but (similar to the implementation of DE 10242233 B3 ) a learning step width limitation are provided so that the feedforward only by repeated learning operations (eg permissible one per drive cycle) approaches the final learning needs.

In both cases ( 2 and 3 ) one would thus have the values of the "EGR pre-tax delta learning characteristic map" 41 ( 2 ) or "EGR pre-tax correction factor" 46 ( 3 ) from the motor operation by learning, and store them in the controller for later use.

The learned values of the EGR pre-tax delta learning map 41 or the EGR pre-tax correction factor 46 Thus, during operation, they always come from a memory of the control unit in which these learning values are stored with certainty.

At initial commissioning, these memories are the EGR pre-tax delta learning map 41 (when executed according to 2 ) or the EGR pre-tax correction factor 46 (when executed according to 3 ) in the case of additive links 43 respectively. 47 with zero or with multiplicative links 43 respectively. 47 1, so that initially no change in the pilot control values from the map 37 results until the first learning values are available.

The management of such memories and questions as to when they need to be cleared or reset to a certain value (eg ECU replacement or replacement of hardware components on the engine that affect the pilot control needs, eg EGR valve exchange) are known (eg from the implementation of DE 10242233 B3 ).

According to the embodiment 3 it is assumed that the characteristic in the formation of the input tax base characteristic field 37 about the sizes 34 (V_AGR_soll) and 36 (Differential pressure or AGR-driving pressure gradient p3 - p2) is correct, and the map 37 only by the factor 46 corrected (scaled in the multiplicative case, parallel raised in the additive case) in order to train in systematic deviations due to contamination, scattering, etc. The necessary correction factor 46 in a sufficiently stationary, ie adjusted operating point is then determined in the multiplicative case from the quotient of total control 12 of the EGR controller 13 (from pilot signal 14 plus controller signal 10 ) and the pure input tax 40 from the input VAT basic map 37 , In the additive case, it is the first learning process directly by the controller signal 10 described or must in further learning processes in each case to then possibly still available controller signal 10 be enlarged.

According to the embodiment 2 on the other hand, the pre-tax reason map 37 not only while maintaining its shaping (by a learning factor) scaled or raised in parallel, but also in its shaping by unterschiedli different learning values for different pairs of input variables 34 (V_AGR_soll) and 36 (Differential pressure or AGR-driving pressure gradient p3 - p2) are adjusted.

The embodiment according to 2 Thus, it allows, compared to the embodiment according to 3 to take even greater advantage of the learning processes. But it also requires more learning, namely with different pairs of input variables 34 (V_AGR_soll) and 36 (Differential pressure or AGR-driving pressure gradient p3 - p2) and a correspondingly more complex learning logic, but from the implementation of DE 10242233 B3 is known in principle.

The idea of being able to learn the aforementioned systematic errors at all is based on the knowledge that these are either invariable deviations from the design state of the pre-control basic map 37 with mean value parts is (for example, because of scattering of the component properties compared to the mean value parts), or at least by time only very slowly variable deviations from the design state of the pre-control basic map 37 (eg by gradually increasing pollution of the EGR route).

• • Lernfenstern,
• • Stationäritätsbedingungen für die Freigabe eines Lernvorgangs,
• • Lernschrittweiten,
• • Inter- und Extrapolation von Lernwerten auf mit Lernwerten zu befüllende Stützstellen in Lernwertekennfeldern,
• • Handling der Lernwertspeicherung für weitere Fahrzyklen,
• • Rücksetzbarkeit der Lernwerte bei Teiletausch,
• • Fehlerüberwachung und Stilllegen der Lernfunktion als Reaktion auf bestimmte, festzulegende, vom Motorsteuergerät erkannte Fehler,
• • etc.

ist bekannt (z. B. aus der Umsetzung von DE 10242233 B3 ) und wird deshalb hier nicht näher beschrieben.The design of such learning functions, eg. B. with

• • learning windows,
• Stationary conditions for releasing a learning process,
• • learning steps,
• • Interpolation and extrapolation of learning values on learning points to be filled with learning points in learning value maps,
• • Handling the learning value memory for further driving cycles,
• • resettable the learning values for parts exchange,
• • fault monitoring and disabling of the learn function in response to certain errors detected by the engine control unit,
• • Etc.

is known (eg from the implementation of DE 10242233 B3 ) and will therefore not be described here.

The control structure according to the invention 30 . 38 respectively. 44 is also suitable for an engine concept with so-called EGR cooler bypassing (not shown).

Increasingly many modern engines have the ability depending on the conditions to be determined by the applicator, the EGR cooler (or one of several EGR coolers) by bypass ( "EGR cooler Bypassierung").

Depending on the design of the EGR cooler, bypass section and bypass valve, it may happen that the flow resistance of the EGR path (for the same EGR valve position) for the cases Bypass on / are different. This means that, depending on the position of the bypass valve, there may then be two different relationships:
EGR flow = f (EGR actuator control, p3 - p2).

It would then be so for the cooler bypass operation to use a different from the EGR cooler operation pilot control map, ie it would then have two different EGR pre-control maps ( 37 ) give:
EGR pilot control_bypass = f (V_AGR_soll, p3 - p2) and
EGR pilot control_cooling = f (V_AGR_soll, p3 - p2).

In the 1 . 2 and 3 drawn input tax structures 31 . 39 respectively. 45 would therefore be duplicated.

When using the extension with learning functions ( 2 or 3 ) are then possibly also the learning maps 41 or the learning factor 46 since the need for correction by learning for the bypass section may be different than for the cooling section (eg due to differential progression of pollution in the two different EGR routes).

Depending on the control of the EGR cooler bypass valve, whose activation is also done by the engine control unit, then one must be switched to one or the other control map, and of course then also the learning result of the valid for the corresponding bypass position Vorsteuer- correction 41 respectively. 46 be assigned.

However, the problem can also be circumvented constructively by incorporating a flow resistance into the EGR bypass section, which raises the flow resistance of the bypass section to the same level as that of the EGR cooler to be bypassed. Then a single pilot control map will suffice 37 ,

If there is a double structure in the engine control unit for the calculation of the desired EGR rates, ie other EGR target rates for the bypass operation than for the cooler operation, this is for the inventive EGR pilot control 31 . 39 respectively. 45 to 1 . 2 and 3 (unlike the prior art according to the prior art 4 ) no problem, since the different target EGR rate in the calculation of the EGR target volume flow 34 (V_AGR_soll) received, so this Effect already at the input value of the pilot control map 37 is taken into account.

The control structure according to the invention 30 . 38 respectively. 44 is also suitable for an engine concept with intake air throttling (ALD). For such engine concepts, where the EGR-driving pressure gradient p3 - p2 36 is not sufficient to achieve the desired EGR rate with the existing EGR route, as a solution, the embodiments of the invention described below are possible.

In addition to opening the EGR valve, a throttle (not shown) arranged in the intake tract is completely or partially closed by means of a throttle actuator (not shown). With naturally aspirated engines, this lowers the pressure in the intake manifold (corresponds to the boost pressure in the supercharger engine in terms of its effect on the here significant AGR-driving pressure gradient) p2 33 clearly, while the pressure in the exhaust manifold (corresponds in its effect on the here significant AGR-driving pressure gradient the exhaust back pressure before turbine in the supercharger engine) p3 32 due to the lower gas flow through the engine only slightly falls. With the supercharger engine, the boost pressure control regulates the loader actuators to a constant boost pressure p2 (after throttle) 33 after, so that the loader has a higher boost pressure p2 (before throttle) 33 resulting in a higher adjusting exhaust backpressure before turbine p3 32 leads.

In both cases, ie when the engine is sucked and charged, the additional drop in pressure causes p3 - p2 36 increased to achieve the desired EGR rate.

Since the pressure gradient p3 - p2 36 but input of the EGR pilot control map according to the invention 37 is, the influence of throttling on the necessary control 12 of the EGR controller 13 Foreseen "correctly".

There are other systems in which the intake air throttling (ALD) is not controlled separately, but is also controlled by the EGR control loop. In such a case, there is a fixed coupling of the output signal 12 from the EGR control loop (meant here is the sum signal 12 from controller signal 10 and pilot signal 14 ) to control the two actuators (AGR controller 13 and throttle plate).

This can be in the way that first the AGR-Stelleller 13 fully open, with further increase of the output signal 12 then in addition the throttle plate is driven, or in such a way that the transition in a defined manner is made sliding or overlapping by z. B. already gradually begins with the approach of the EGR valve to its full opening throttling. The sliding or intersecting transition may, for. B. for reasons of better controllability be advantageous over the first mentioned successive onset of control.

From the perspective of the pilot control according to the invention 31 . 39 respectively. 45 Such a system with a defined coupling of EGR actuator and throttle actuator control can be treated as if it were one without throttling, ie the control structure according to the invention 30 . 38 respectively. 44 is also suitable for such systems with intake air throttling.

Consider then the combination of AGR-Steller 13 and throttle plate, both in the manner described above at the output of the EGR controller (controller output signal 12 ), as a "black-box actuator element" whose activation (output signal of the EGR controller, here is meant the sum signal 12 from controller signal 10 and pilot signal 14 ) together with the driving pressure gradient p3 - p2 36 determine the EGR flow.

In the pilot control map 37 then there is not the necessary control of the EGR controller 13 alone, but the necessary output signal 10 of the EGR controller 9 , which then by means of the specified allocation to the two operators involved (AGR-Steller 13 and throttle) in total to the desired EGR flow.

QUOTES INCLUDE IN THE DESCRIPTION

This list The documents listed by the applicant have been automated generated and is solely for better information recorded by the reader. The list is not part of the German Patent or utility model application. The DPMA takes over no liability for any errors or omissions.

Cited patent literature

• - DE 10257031 A1 [0007]
• - DE 2633617 A1 [0009]
• - DE 2633617 C2 [0009]
• - DE 10229620 A1 [0025, 0058, 0063, 0065, 0069, 0077, 0077]
• - DE 10242234 A1 [0025, 0058, 0063, 0065, 0069, 0077, 0077]
• - DE 10242233 B3 [0059, 0059, 0096, 0100, 0103, 0105]

Claims (3)

Method for setting an exhaust gas recirculation rate of an internal combustion engine, wherein the adjustment is effected by means of a controller ( 31 . 39 . 45 ) with superimposed control loop ( 2 ) or a regulation ( 2 ) with pilot control ( 31 . 39 . 45 ) and a pilot control signal ( 14 . 40 ) based on characteristic curves or characteristic diagrams ( 37 ) is generated, characterized in that the data from the characteristic curves or maps ( 37 ) based on laws of fluid dynamics and in particular take into account a driving pressure gradient underlying the exhaust gas recirculation. Method according to Claim 1, characterized in that the precontrol signal ( 14 . 40 ) a correction signal ( 42 ; 46 ) from another map ( 41 ) or from other characteristics is superimposed. Method according to Claim 2, characterized in that the correction signal ( 42 ; 46 ) Contains correction or learning values for learning to 2 ) to adapt to a changed situation, wherein the changed situation is due to a long-term drift or the like.