EP1413728A2 - Controller and method for controlling a NOX-sensor arranged in an exhaust gas channel of an internal combustion engine - Google Patents

Abstract

Operation of an exhaust gas sensor in an internal combustion engine exhaust involves control of the operating parameters of the sensor by use of model-based control and/or of exhaust gas-dependent pre-control. An Independent claim is also included for a device for controlling the operating parameters in the exhaust, the device being formed as a model-based control with structure determined by a model of the controlled system.

Description

The invention relates to a method and a controller for operating an exhaust gas sensor arranged in an exhaust gas duct of an internal combustion engine with the features mentioned in the preamble of claims 1 and 19.

It is known that, for example, NO _x sensors are usually used to determine the NO _x content in the exhaust gas of an internal combustion engine. Such NO _x sensors are known in a wide variety of designs, so that a detailed description can be dispensed with here. Especially with a lean executable combustion engine, it is to comply with the NO _x limit important to the NO _x content in the exhaust system behind a NO _x trap to know in order to, for example, determine the time at which _x is a storage of NO in the NO _x storage is no longer possible, that is, the NO _x storage is filled, so that a regeneration phase is necessary.

For this purpose, there is usually a measuring device behind the catalytic converter in the exhaust tract of an at least temporarily lean-burn internal combustion engine, which is capable of measuring the NO _x concentration in the exhaust gas, for example a NO _x sensor. The measurement accuracy of exhaust gas sensors such as NO _x sensors or lambda sensors generally depends on external parameters. It is therefore important to keep these parameters in the areas that are optimal for precise control and to adjust them if necessary. In the case of NO _x sensors, for example, the temperature must be kept constant in order to ensure a high accuracy of the signal. The smaller the deviations from the target temperature of the sensor, the more accurate the sensor signal.

Up to now, this exhaust gas sensor for temperature control has used a voltage-based PI controller without exhaust-gas-dependent feedforward control, which works with a resistance difference as the controller input variable and a voltage as the controller output variable. The weak point of this system is the great inaccuracy in controlling the sensor temperature in the event of fluctuations in the exhaust gas mass flow due to, among other things, a missing model of the route to be controlled.

The invention is therefore based on the object of providing a regulator and a method for regulating an exhaust gas sensor arranged in an exhaust gas duct of an internal combustion engine, which remedy the disadvantages mentioned, and in particular to provide a robust regulator or a robust method for regulating an exhaust gas sensor.

This object is achieved by a controller and a method with the features mentioned in claims 1 and 19.

A particular advantage of the method for operating an exhaust gas sensor arranged in an exhaust gas duct of an internal combustion engine is that the measurement accuracy of exhaust gas sensors is increased by regulating operating parameters of the exhaust gas sensor by means of a model-based controller and / or pre-controlling exhaust gas flow-dependent operating parameters of the exhaust gas sensor.

To regulate operating parameters of an exhaust gas sensor arranged in an exhaust gas duct of an internal combustion engine, a controller is therefore used which is characterized in that the controller is designed as a model-based controller and has a structure predetermined by a model of the controlled system.

In a preferred embodiment of the method according to the invention it is provided that the operating parameter temperature is regulated and / or controlled when operating a NO _x sensor. It is advantageous if the temperature is determined by measuring an electrical resistance value. It is also advantageous if the control system evaluates the difference between the current and the target temperature of the sensor as a controller input variable and the heating power for heating the sensor varies as a controller output variable. This creates a linear relationship between the controller input and output variable.

In a preferred embodiment of the method according to the invention it is provided that the exhaust gas flow-dependent pilot control is set up in such a way that the exhaust gas flow-dependent feedforward control becomes active when the exhaust gas flow decreases or increases very rapidly. In particular, this can be realized in that the exhaust gas flow-dependent precontrol becomes active when the exhaust gas flow changes faster than gradients matched to the sampling time.

In a further preferred embodiment of the invention it is provided that the current value of the measured sensor resistance and / or the current value of the measured sensor temperature is taken into account in the release condition of the exhaust flow-dependent precontrol.

It proves to be an advantage if the pilot control reacts briefly with a reduced heating power when the exhaust gas mass flow falls rapidly and the pilot control reacts briefly with an increased heating power when the exhaust gas mass flow increases rapidly. This compensates for undesired heating or cooling of the sensor.

In a further preferred embodiment of the method according to the invention, it is provided that, when the exhaust gas sensor is precontrolled, several successive differences in the measured exhaust gas mass flow values are evaluated. Good results are achieved in particular if the number of differences in time of the measured exhaust gas mass flow values to be evaluated for a pilot control is determined as a function of the sampling time, with a higher number being evaluated for short sampling times and a smaller number for long sampling times.

It has also proven to be advantageous that a precontrol with a low heating power value, which is matched to a sampling time of 50 ms, becomes active if three successive differences in the measured exhaust gas mass flow values are greater than the gradient to be applied. On the other hand, a pilot control with a high heating power value, which is matched to a sampling time of 50 ms, advantageously becomes active when four successive differences in the measured exhaust gas mass flow values are smaller than the gradient to be applied. Both of these approaches represent a compromise between avoiding incorrect control and a short response time.

In order to prevent undesired cooling of the exhaust gas sensor, a preferred embodiment of the method according to the invention provides that a high heating output pilot control takes place at least at the beginning of overrun phases.

It proves to be advantageous that the model-based controller is determined using a linear, continuous-time method. It turned out to be particularly practical when the model-based controller using the LQG / LTR process (LQG / LTR = Linear Quadratic Gaussian / Loop Transfer Recovery) is designed.

A particularly robust controller can be obtained if the model of the controlled system is based on the parameters time constant T, damping d and static gain k _SR .

lautet, wobei

hno0y

die aktuelle Ausgangsgröße,

hno1y

die im vorangegangenen Zeittakt berechnete Ausgangsgröße,

hno2y

die zwei Zeittakte zuvor berechnete Ausgangsgröße,

hno0u

die aktuelle Eingangsgröße,

hno1u

die im vorangegangenen Zeittakt gemessene Eingangsgröße,

hno2u

die zwei Zeittakte zuvor gemessene Eingangsgröße,

k_SR

die statische Verstärkung und

HNOij

die Regelparameter (i = 0, 1, 2 und j = U, YK)

darstellen.In a further preferred embodiment of the method according to the invention it is provided that the difference equation on which a time-discrete implementation is based $$\begin{gathered} {\text{hno0y = k}}_{\text{SR}}\text{· (HNO0U · hno0u + HNO1U · hno1u + HNO2U · hno2u)} \\ \text{- ((-HNO1YK) · hno1y + (-HNO2YK) · hno2y)} \end{gathered}$$

is where

hno0y

the current output variable,

hno1y

the output variable calculated in the previous time cycle,

hno2y

the two output cycles previously calculated,

hno0u

the current input variable,

hno1u

the input variable measured in the previous time cycle,

hno2u

the input variable measured two times before,

k _SR

the static gain and

HNOij

the control parameters (i = 0, 1, 2 and j = U, YK)

represent.

It proves to be particularly advantageous if the controller is designed as a linear controller. When using a linear controller, good control results are obtained if the controlled system also has a linear relationship, that is to say if there is a linear relationship between controller input and controller output variables.

A preferred embodiment of the controller according to the invention is characterized in that the controller comprises a Kalman filter and an optimal state feedback. To avoid undesired "run-up" of the integrator, it is advantageous if the controller has an integrator that avoids an integrator windup.

In a further preferred embodiment of the controller according to the invention It is provided that the controller includes a means for measuring the exhaust gas mass flow value, which enables a pre-control depending on the exhaust gas mass flow.

Further preferred embodiments of the invention result from the other features mentioned in the subclaims.

Figur 1

eine Darstellung der in einem Sprungversuch ermittelten Kennlinien und

Figur 2

eine Darstellung von Amplituden- und Phasengang (Bode-Diagramm) der offenen Kette bestehend aus dem erfindungsgemäßen Regler und einem ein Integrator-Windup vermeidenden Integrator.

The invention is explained in more detail below in exemplary embodiments with reference to the associated drawings. For example, a NO _x sensor is used as the exhaust gas sensor. Show it:

Figure 1

a representation of the characteristics determined in a jump test and

Figure 2

a representation of the amplitude and phase response (Bode diagram) of the open chain consisting of the controller according to the invention and an integrator avoiding an integrator windup.

The controller according to the invention is distinguished by the fact that it is based on a new controller structure which contains a model of the controlled system. Due to this additional information about the controlled system, a higher control accuracy is achieved. In addition to the new controller, the relationship between controller input and output variables has been linearized in order to achieve better control results. The controller now varies the heating power and thus regulates the heating of the sensor to its target temperature. Controller input variable is a temperature difference that results from the current and the target temperature of the sensor. The new controller structure includes a special integrator, the advantage of which is the avoidance of an "integrator windup", that is, an undesired "run-up" of the integrator when the manipulated variable has reached the saturation value of the controlled system input. In addition, the new controller structure can be combined with a pre-control depending on the exhaust gas mass flow in order to compensate for the dead time due to the position of the NO _x sensor far behind in the exhaust tract.

mit den oben definierten Größen.An exemplary embodiment of the model-based controller was designed according to the LQG / LTR method (LQG / LTR = Linear Quadratic Gaussian / Loop Transfer Recovery). The LQG / LTR process is a linear, continuous design process. The exemplary controller designed with this consists of a Kalman filter as an observer and an optimal state feedback. The underlying difference equation for a time-discrete implementation, such as is implemented in the software of an engine control unit, is: $$\begin{gathered} {\text{hno0y = k}}_{\text{SR}}\text{· (HNO0U · hno0u + HNO1U · hno1u + HNO2U · hno2u)} \\ \text{- ((-HNO1YK) · hno1y + (-HNO2YK) · hno2y)} \end{gathered}$$

with the sizes defined above.

Since the order of the controlled system model specifies the order of the model-based controller, it was necessary to determine a suitable, yet simple model of the controlled system. For this purpose, jump tests were carried out on the route, with jumps in performance being made on the route in such a way that the route output variable, the sensor temperature, jumps around its working point and the target temperature, as would be expected in normal operation. The result of a jump test carried out is shown in FIG. The voltage step size 10 U_sprung is converted into a power value via the heating resistor. The intermediate variable 11 R _PVS , that is the internal resistance of the sensor, and the desired step response 12 T_Tip, the temperature of the sensor element, are also plotted in the diagram. With the aid of a suitable aid, the measured step response 12 was overlaid with the simulated behavior of a PT ₂ element 13 T_model. As can be clearly seen in FIG. 1, it was found that a PT _{2 member} approximates the measured behavior well. The parameters of the exemplary model are: time constant T = 4.2s, damping d = 1.2s and static gain k _SR = 17 ° C / W.

Figure 2 shows the amplitude and phase response (Bode diagram) of the open chain 20 consisting of the controller according to the invention and the special integrator described above. With a phase reserve of around 73 °, the structure is extremely robust.

mit der Temperaturerhöhung ΔT in [K] und der Masse m in [kg] des Körpers sowie der ihm zugeführten Wärmemenge Q in [Ws]. Die Umrechnung vom bisher verwendeten Innenwiderstandswert des Sensors in seine Temperatur erfolgt mit Hilfe einer vom Sensorhersteller ermittelten nichtlinearen Kennlinie.Since it is a linear controller, a linear relationship between controller input and output variable is required for a good control result. This is the reason for the change in the controller input variable from the measured sensor resistance value to a sensor temperature value and from the heating voltage value of the controller output variable to a heating output value. The linear relationship between the temperature increase of a body and the amount of heat supplied to it is: $$\mathit{\text{.DELTA.T}}\text{~}\frac{\text{1}}{\mathit{\text{m}}}\mathit{\text{Q}}$$

with the temperature increase ΔT in [K] and the mass m in [kg] of the body and the amount of heat Q supplied to it in [Ws]. The conversion of the internal resistance value of the sensor used to its temperature is carried out with the help of a non-linear characteristic curve determined by the sensor manufacturer.

In order to better take fluctuations in the exhaust gas mass flow into account when regulating the temperature of the NO _x sensor, the proposed new controller structure can be combined with a feedforward control dependent on the exhaust gas mass flow. The pilot control becomes active when the exhaust gas mass flow increases or decreases very quickly. The idea behind this is that the exhaust gas mass flow provides information about an impending cooling or heating of the sensor located further back in the exhaust tract. If the mass flow changes faster than the gradients matched to the sampling time, the pilot control becomes active by either a strong heating up to anticipate the upcoming cooling down, or a lowering of the heating power to a low value due to an upcoming heating up. The current value of the measured sensor resistance is included in the enabling condition of the pilot control in order to avoid that the heating power pilot control is too low when the sensor temperature is too low and vice versa. As an alternative to the sensor resistance, the sensor temperature can also be measured, temperature and resistance being converted into one another using the characteristic curve mentioned above.

Measurements have shown that with a rapidly falling exhaust gas mass flow, an undesirable increase in the sensor temperature takes place at the first moment, to which the pilot control reacts briefly with a reduced heating power value in a compensating manner. Conversely, with a rapidly increasing exhaust gas mass flow, the sensor cools down at first, to which the pilot control in turn reacts with increased heating power.

The measurement of the exhaust gas mass flow is impaired by measurement noise. In order to differentiate yourself from this measurement noise, it is usually not appropriate to react to every (small) change in the exhaust gas mass flow by adjusting the heating output. Instead, it will make sense to consider a higher or lower number of consecutive differences in the measured exhaust gas mass flow values for a pilot control depending on the sampling time. The number of consecutive differences must be chosen so that a compromise is found between the delimitation against the measurement noise and an excessively long reaction time. For shorter sampling times, it can be advantageous to increase the number of increasing differences of the measured exhaust gas mass flow values taking into account in time; with longer sampling times, it may be necessary to reduce this number.

For example, for a pilot control with a low heating power value and the special, exemplary sampling time of 50 ms, the criterion "three consecutive differences in the measured exhaust gas mass flow values are greater than the gradient to be applied (corresponds to a reaction time of 3 * 50 ms)" can be a compromise between avoiding incorrect feedforward control and a short reaction time.

For a pilot control with a high heating power value, the compromise can be formed in a special embodiment with a sampling time of 50 ms, for example by the criterion "four successive differences in the measured exhaust gas mass flow values are smaller than the gradient to be applied (corresponds to a reaction time of 4 * 50 ms)" become.

At the beginning of a fuel cut-off phase, the sensor usually cools down somewhat. It is therefore possible to switch to maximum heating via the pilot control even for overrun phases, even before this occurs via the control deviation and the controller itself.

The invention is not restricted to the exemplary embodiments shown here. Rather, it is possible to implement further embodiment variants by combining and modifying the means and features mentioned, without leaving the scope of the invention.

LIST OF REFERENCE NUMBERS

10

Voltage step size

11

between size

12

Step response

13

simulated behavior of a PT ₂ link

20

open chain

Claims (24)

Method for operating an exhaust gas sensor arranged in an exhaust gas duct of an internal combustion engine, characterized in that operating parameters of the exhaust gas sensor are regulated by a model-based controller and / or pre-control of operating parameters of the exhaust gas sensor depending on the exhaust gas flow. A method according to claim 1, characterized in that the operating parameter temperature is regulated and / or controlled when operating a NO _x sensor. A method according to claim 2, characterized in that the temperature is determined by measuring an electrical resistance value. A method according to claim 2, characterized in that the control evaluates the difference between the current and the target temperature of the sensor as a controller input variable and the heating power for heating the sensor varies as a controller output variable. Method according to one of the preceding claims, characterized in that the pre-control dependent on the exhaust gas flow becomes active when the exhaust gas flow decreases or increases very rapidly. Method according to one of the preceding claims, characterized in that the exhaust gas flow-dependent precontrol becomes active when the exhaust gas flow changes faster than gradients matched to the sampling time. Method according to one of the preceding claims, characterized in that in a release condition of the exhaust gas flow-dependent pilot control - the current value of the measured sensor resistance and / or - the current value of the measured sensor temperature is taken into account. Method according to one of the preceding claims, characterized in that when the exhaust gas mass flow falls rapidly, the pilot control reacts briefly with a reduced heating power. Method according to one of the preceding claims, characterized in that when the exhaust gas mass flow increases rapidly, the pilot control reacts briefly with an increased heating power. Method according to one of the preceding claims, characterized in that when the exhaust gas sensor is precontrolled, several temporally successive differences of the measured exhaust gas mass flow values are evaluated. Method according to one of the preceding claims, characterized in that the number of temporally successive differences of the measured exhaust gas mass flow values to be evaluated for a pilot control is determined as a function of the sampling time, with a higher number being evaluated for short sampling times and a smaller number for long sampling times becomes. Method according to one of the preceding claims, characterized in that a precontrol, which is matched to a sampling time of 50 ms, becomes active with a low heating power value if three successive differences in the measured exhaust gas mass flow values are greater than the gradient to be applied. Method according to one of the preceding claims, characterized in that a feedforward control, tuned to a sampling time of 50 ms, becomes active when four successive differences in the measured exhaust gas mass flow values are smaller than the gradient to be applied. Method according to one of the preceding claims, characterized in that a high heating power pre-control takes place at least at the beginning of overrun phases. Method according to one of the preceding claims, characterized in that the model-based controller is designed with the aid of a linear continuous-time method. Method according to one of the preceding claims, characterized in that the model-based controller using the LQG / LTR method (LQG / LTR = linear Quadratic Gaussian / Loop Transfer Recovery). Method according to one of the preceding claims, characterized in that the parameters of the controlled system model - time constant (T), - Attenuation (d (and - static gain (k _SR ) be taken as a basis Method according to one of the preceding claims, characterized in that the difference equation on which a time-discrete implementation is based $$\begin{gathered} {\text{hno0y = k}}_{\text{SR}}\text{· (HNO0U · hno0u + HNO1U · hno1u + HNO2U · hno2u)} \\ \text{- ((-HNO1YK) · hno1y + (-HNO2YK) · hno2y)} \end{gathered}$$

is where hno0y the current output variable, hno1y the output variable calculated in the previous time cycle, hno2y the output variable previously calculated two time cycles, hno0u the current input variable, hno1u the input variable measured in the previous time cycle, hno2u the input variable measured two time cycles before, k _SR the static gain and HNOij the control parameters (i = 0, 1, 2 and j = U, YK) represent. Controller for regulating operating parameters of an exhaust gas sensor arranged in an exhaust gas duct of an internal combustion engine, characterized in that the controller is designed as a model-based controller and has a structure predetermined by a model of the controlled system. Controller according to claim 19, characterized in that the controller is designed as a linear controller. Controller according to one of claims 19 or 20, characterized in that there is a linear relationship between controller input and controller output variables. Controller according to one of claims 19 to 21, characterized in that the controller comprises a Kalman filter and an optimal state feedback. Controller according to one of claims 19 to 22, characterized in that the controller has an integrator avoiding an integrator windup. Controller according to one of claims 19 to 23, characterized in that the controller comprises a means for measuring the exhaust gas mass flow value.

Images