DE102011077251B3 - Diagnostic method for ammonia filter of nitrogen oxide sensor for exhaust system of diesel engine for motor car, involves deactivating filter when value of error is below threshold value - Google Patents

Abstract

A diagnostic method for a filter (5) of a NOx sensor (4) of an exhaust system (1) with an SCR catalytic converter (3) comprises the following steps:
- operating (11) the filter (5) including calculation (12) of a NOx estimate downstream of the catalyst (3);
- modeling (13) a linear NOx model based on the NOx estimate;
Comparing (15) the NOx estimate and the linear NOx model;
- Disabling (18, 19) of the filter (5) when the difference of the comparison (15) is below a threshold.

Description

The invention relates to a diagnostic method and a diagnostic module for a filter of a NOx sensor of an exhaust system, in particular an exhaust system with SCR catalyst for an internal combustion engine of a motor vehicle.

Systems with SCR catalyst (Selective Catalytic Reduction) are well suited for the removal of NOx emissions, especially for exhaust gases of diesel engines. In an active SCR system, ammonia (NH3) is metered into the SCR catalyst, where it is adsorbed on the catalyst and reacts with NO and NO ₂ from the exhaust gas. NH3 is typically not dosed directly but usually in the form of a urea solution, which partially converts to NH3 after injection. In a passive SCR system, there is no active injection of NH3 or urea upstream of the catalyst. Instead, NH3 is produced by another component upstream of the catalyst, such as an LNT (Lean NOx Trap), after the engine is in rich operation.

If insufficient NH3 is being metered or stored in the catalyst, or if the temperature is not adequate for complete NOx conversion, then not all of the NOx will be converted and NOx will slip through the catalyst. However, if too much NH3 has been metered and stored in the catalyst, desorption may occur. Typically, NH3 desorption occurs after a rapid increase in temperature, for example as a result of an increasing engine load. However, it can also occur with an overdose of NH3 at steady temperature.

To precisely control the processes in the SCR catalyst, information about the NOx and NH3 concentration downstream or downstream of the catalyst is of interest. For this purpose there are two types of sensors, NH3 sensors and NOx sensors. While NH3 sensors measure only NH3 concentrations or quantities, NOx sensors are sensitive to both NOx and NH3. This results in difficulty in the measurement in that it can not be easy to determine if the sensor is currently measuring NOx or NH3 values.

A filter or filtering process, logged in the parallel application DE 10 2011 077 246.4 Determines if the signal measured by the NOx sensor is a NOx reading or NH3 reading. For this purpose, two different models or models are created and / or adapted. One model describes the NOx conversion in the catalyst and the second model describes the behavior of the NH3 slip through the catalyst.

DE 10 2010 026 373 A1 discloses a system for identifying ammonia slip conditions in a selective catalytic reduction application. A filter module of the system filters a first signal indicative of an amount of NOx upstream of a catalyst and a second signal indicative of amounts of NOx and NH3 downstream of the catalyst. A slip determination module determines whether NH3 is present in the exhaust downstream of the catalyst based on frequency response of the first and second signals.

The accuracy or efficiency of the filter depends on the properties of the NOx level or signal upstream of the catalyst. In order to be able to differentiate downstream between the NOx values and NH3 values in the NOx signal, the NOx signal upstream of the catalytic converter must have a certain dynamics.

In the case of a NOx breakthrough in the catalyst, the NOx signal will dynamically follow the input NOx signal, while the NH3 value typically will not. Therefore, the decision for a rather steady NOx signal upstream of the catalytic converter (for example, in steady state driving) can not be made, and the filtered NOx / NH3 signal should not be used for feedback control.

The invention is based on the object to improve the operation of a NOx sensor in an exhaust system with catalyst.

This object is achieved with the features of claims 1 and 10, respectively. Advantageous developments of the invention are defined in the dependent claims.

• – Betreiben des Filters einschließlich Berechnung eines NOx Schätzwertes stromabwärts des Katalysators;
• – Modellieren eines linearen NOx Modells basierend auf dem NOx Schätzwert;
• – Durchführen eines Vergleichs mittels einer Fehlerberechnung zwischen dem NOx Schätzwert und einem NOx Schätzwert des linearen NOx Modells;
• – Deaktivieren des Filters, wenn der dadurch ermittelte Fehler unter einem Schwellwert liegt.

According to a first aspect of the invention, a diagnostic method for a filter of a NOx sensor of an exhaust system comprising an SCR catalyst comprises the following steps:

• - operating the filter including calculating a NOx estimate downstream of the catalyst;
• Modeling a linear NOx model based on the NOx estimate;
• - making a comparison by means of an error calculation between the NOx estimated value and a NOx estimated value of the linear NOx model;
• - Deactivate the filter if the error thus detected is below a threshold value.

The method stabilizes the filter for the NOX sensor and determines when the decision of the filter can be trusted. One implementation of the filter is to continuously adapt two different models and compare their accuracy. A model represents a simplified kinetic model of NOx conversion. The second model is a simple linear time algorithm that can well describe NH3 slip behavior over a limited time window. In some cases, the filter can not distinguish between NOx and NH3. This typically happens when the NOx signal does not have sufficient dynamics, for example in steady state operation. In this case, the NOx downstream of the catalyst may also be described by a linear time algorithm. By determining this condition, it can be prevented that the filter makes wrong assumptions or makes decisions. The filter may be disabled or its decision may be discarded or not used if, for example, the absolute difference between the two signals of the comparison is below a threshold.

For the comparison, an error calculation between the NOx estimation value and a linear NOx estimation value of the linear NOx model can be used. This error calculation is easy to perform and provides fast and reliable results.

The filter can be activated if the error detected in comparison is above a threshold value. This threshold may be equal to the deactivation threshold, or two different thresholds may be used. Active or activate means that the decisions of the filter are used. If disabled, the filter will run in the background, but the results will not be used. Alternatively, the filter can be turned off and turned on again when activated.

The filter can remain active for a defined period of time if the error determined in comparison is above the threshold value. This mode of operation may be referred to as transient activation. In transient activation, it is assumed that the "old" filter decision, that is, at the beginning or before the defined period of time, is still valid if a decision has recently been made and the NOx sensor signal has not moved outside defined limits. Typically, NOx and NH3 downstream of the catalyst do not change from one moment to another, so a recent decision may still be valid even if no decision can be made from the current data. The transient activation assumes that no new decision has been made, the filter is still activated, and its results are used for feedback or further processing. The transient activation is deactivated after a predetermined period of time or when a NOx sensor signal exceeds or falls below predetermined limits.

With the filter deactivated, a change in the NOx value upstream of the catalyst can be initiated. Thus, in the case where the filter can not make a reliable decision, a NO x (engine out) fault flag or message may be set so as to re-excite or reliably start the base algorithm of the filter. When deactivated, the filter does not provide a reliable signal, neither NOx nor NH3. In this case, the metering or injection for the catalyst runs without feedback, ie in an open loop. As long as the NOx sensor signal remains within certain allowable limits, this may be acceptable. However, if it breaks these limits, it may be desirable to force a determination or calculation of the NOx sensor signal. This can be achieved by targeted disturbances or changes in the NOx leaving the engine in the exhaust gas, for example by complete or partial switching on and / or off the exhaust gas recirculation valve (EGR). This case can occur, for example, during a long stationary operation, for example by a cruise control system (cruise control). As soon as the operation and thus the NOx signal gains a certain dynamic, the filter automatically starts with normal operation.

For the modeling of NOx conversion, a kinetic model of NOx conversion can be used, this may preferably be a simplified kinetic model to reduce the computational effort and the subsequent processing of the data.

The steps or calculations may be performed at sampling instants with a sampling time of preferably one second. Too few sample times do not allow for consistent models, while too many increase the computational overhead unnecessarily.

In addition, for the calculation of the steps, values of past sampling times may preferably be used in a time frame of twenty seconds. This increases the reliability of the calculation and the determination of which measured values are being measured. For comparison, the average absolute difference between the two signals can then be used as a criterion.

According to a second aspect of the invention, a diagnostic module for a filter of a NOx sensor of an exhaust system with an SCR catalyst comprises a signal input for signals of the filter, a computing unit configured to carry out the diagnostic method described above, and a signal output for outputting a result of the diagnostic method. The advantages and modifications described above apply.

In the following the invention will be described in more detail with reference to the drawing, in which:

1 a schematic block diagram of an exhaust system with diagnostic module according to the invention.

2 a block diagram of the diagnostic method for a filter according to the invention.

3 and 4 in each case a diagram of an application of the diagnostic method according to the invention.

The drawings are merely illustrative of the invention and do not limit it. The drawings and the individual parts are not necessarily to scale. Like reference numerals designate like or similar parts.

1 shows a part of an exhaust aftertreatment system 1 for an internal combustion engine 2 , For example, a diesel engine of a motor vehicle. The exhaust aftertreatment system or exhaust system 1 includes a downstream of the engine 2 arranged catalyst 3 such as an SCR (Selective Catalytic Reduction) catalyst.

Downstream of the catalyst 3 is a NOx sensor 4 arranged to measure the NOx concentration and the amount of NOx in the exhaust gas. The NOx sensor 4 does not have to be directly behind the catalyst 3 be arranged, he may also be located downstream.

Measured values of the NOx sensor 4 arrive at a filter or NH3 filter 5 for the NOx sensor 4 , This filter 5 processes the readings or prepares them to be able to detect if the of the NOx sensor 4 measured values are NOx values or NH3 values. The filter 5 outputs appropriate signals, for example to an engine control or exhaust aftertreatment control.

The filter 5 needed next to that of the NOx sensor 4 supplied measured value downstream of the catalyst 3 the NOx concentration or NOx amount emitted by the engine 2 is produced and this in the direction of the catalyst 3 leaves. Either there is an additional NOx sensor 6 upstream of the catalyst 3 present or the NOx concentration upstream of the catalyst 3 comes with a model 6 That determines the engine 2 generated amount and / or concentration of NOx determined.

Between the optional NOx sensor 6 upstream of the catalyst 3 and the catalyst 3 is also an optional injection point 7 intended for urea solution or NH3. The NOx sensor 6 or the modeling is upstream of the injection point 7 arranged to avoid NH3 cross-sensitivity.

The exit 5b of the filter 5 is with a diagnostic module 8th connected. The diagnostic module 8th can be a stand-alone unit with a signal input 8a for signals of the filter 5 , a computing unit for performing the diagnostic method and a signal output 8b to output a result of the method. The diagnostic module 8th may also be part of a control, such as an engine control or exhaust aftertreatment control, or the filter 5 be. This can be realized in the form of a software routine or a software module. The signal output 8b can also be realized in software. That via the signal output 8b output can be an independent assessment or validation of the result of the filter 5 and / or the approved or not released result of the filter 5 be.

Based on 2 Now the functioning of the diagnostic module 8th or the diagnostic procedure described. The diagnostic procedure starts from the NOx sensor 4 behind the catalyst 3 out.

The in 2 schematically represented process has a first step or step 10 the modeling of a linear NOx model. In a first step or sub-step 11 becomes the filter 5 operated, in a second step 11 as a result of the filter 5 the estimated NOx value downstream of the catalyst 3 NOxPostEstV is determined.

For NOX modeling, a 0D CSTR (Continuous Stirred-Tank Reactor) model with first-order kinetics is used:

wobei

The signals can be either scalar (a sample or sample) or vectorial (time window with elements of wise operation). For the time window, the coefficient of the reaction rate (k _R ) is assumed to be constant. The optimum reaction rate coefficient can be determined analytically by the least squares descent method:

in which

The optimization criterion min (ε ^T ε)
leads to:

In a further step 13 Based on this NOX estimated value NOxPostEstV, the linear NOx model is formed. As a result of the linear NOx model stands in step 14 a linear NOx estimate downstream of the catalyst 3 NOxLinPostEstV available.

The linear time algorithm for the linear NOx model based on the NOX model signal (NOxPostEstV-N vector (eg twenty samples)) is calculated as follows: NO _x LinPostEstV = an + b where a and b are constants to be fitted and n is the number of samples from 0 to the time horizon -1.

wobei

A and b are calculated by minimizing the least-square criterion:

in which

The optimization criterion min (ε ^T ε)
leads to:

In a second stage 15 the NOx estimated value and the linear NOx model or the linear NOx estimated value are compared with each other.

This is _done via an error _calculation ε _NOxLin between NOxLinPostEstV and NOxPostEstV in one step 16 achieved according to the following formula:

The smaller the error, the more linear the NOx estimation and thus it is more difficult to determine if the sensor 4 NOx or NH3 measures.

In a fourth stage 17 is determined or determined whether the signals of the NOx sensor 4 NOx or NH3 values are. This is done in one step 18 the error signal ε _{NOxLin compared} with a predefined limit or threshold ε _min . If the error signal ε _{NOxLin is} smaller than the limit, a flag or signal is set that the decision of the filter 5 can not be made or taken into account, or in other words the filter 5 is deactivated. Since a new error signal is calculated at each sampling _instant , the last or some of the last values for the error signal ε _{NOxLin can also be} taken into account for this decision.

If the error signal ε _{NOxLin is} greater than the limit, the filter becomes 5 according to step 19 operated or activated. If this is not the case, the filter can 5 according to step 20 remain activated for a certain predefined period as part of a so-called transient activation. In transient activation, it is assumed that the "old" filter decision, that is, at the beginning or before the defined period of time, is still valid if a decision has recently been made and the NOx sensor signal has not moved outside defined limits.

As long as the transient activation is in operation, according to step 19 the filter 5 activated, ie its results taken into account. After a predetermined period of time or when a NOx sensor signal exceeds or falls below preset limits, the transient activation is deactivated and it becomes step 21 branched.

In step 21 becomes the filter 5 deactivated, ie either switched off or its results are not considered.

If the filter 5 is deactivated, it does not provide a reliable signal, neither NOx nor NH3. In this case, the dosage or injection for the catalyst is in progress 3 without feedback, ie in an open loop. As long as the NOx sensor signal remains within certain allowable limits, this may be acceptable.

If in one step 22 it is determined that the NOx sensor signal is outside certain allowable limits, this may be a positive or negative deviation, in one step 23 Forcing a determination or calculation of the NOx sensor signal. This can be achieved by targeted disturbances or changes in the NOx leaving the engine in the exhaust gas, for example by complete or partial switching on and / or off the exhaust gas recirculation valve (EGR).

This procedure runs in the diagnostic module 8th from. The process either outputs a valid NOx or NH3 value or outputs a signal indicating the reliability of the filter 5 indicates. The diagnostic module 8th is in the signal chain between NOx sensor 4 , Filters 5 and further processing control (not shown here) switched. Accordingly, the diagnostic module 8th also in the control or the NOx sensor 4 or the filter 5 be integrated.

The 3 and 4 show diagrams showing the operation of the method or the diagnostic module 8th illustrate. It shows how robust the operation of the exhaust system 1 or the filter 5 through the diagnostic method or the diagnostic module 8th becomes.

3 shows a dynamic driving cycle in which the NOx and NH3 values can be well separated. The error of the linear NOx model ε _NOxLin remains relatively high most of the time. The signal filter_enabled, indicating the activation of the filter 5 indicates that most of the time is active. In the two lower diagrams of 3 are the determined NOx and NH3 values downstream of the catalyst 3 and actual values obtained with separate meters or sensors, and it can be seen how well these values match. The method works so reliable.

4 shows a more stationary operation where the diagnostic procedure determines that no reliable decision can be made and the filter 5 is deactivated. Therefore, the filter_enabled signal is inactive or low in the second temporal portion of the graphs. As previously described, the feedback is disabled until the filter 5 is activated again.

It should be noted that the filter 5 during the first approximately 260 seconds remains active, although the error of the linear NOx model ε _{NOxLin is} low. This follows the fact that the NOx sensor measures very low values and therefore the result of the filter 5 anyway corrected.

The procedure or the diagnostic module 8th can be used in any exhaust system with catalyst in particular SCR catalyst (passive and active).

Claims (10)

Diagnostic procedure for a filter ( 5 ) of a NOx sensor ( 4 ) of an exhaust system ( 1 ) comprising an SCR catalyst ( 3 ), with the following steps: - operating ( 11 ) of the filter ( 5 ) including calculation ( 12 ) of a NOx estimate downstream of the catalyst ( 3 ); - Modeling ( 13 ) a linear NOx model based on the NOx estimate; - carrying out a comparison ( 15 ) by means of an error calculation ( 16 ) between the NOx estimated value and a NOx estimated value of the linear NOx model; - Deactivate ( 18 . 19 ) of the filter ( 5 ), if the error thus determined is below a threshold value. Diagnostic method according to claim 1, wherein for the comparison ( 15 ) an error calculation ( 16 ) is used between the NOx estimate and a linear NOx estimate of the linear NOx model. Diagnostic method according to claim 1 or 2, wherein the filter ( 5 ) is activated when compared ( 15 ) detected error is above a threshold. Diagnostic method according to claim 1 to 3, wherein the filter ( 5 ) remains active for a defined period of time when compared ( 15 ) detected error is above the threshold. Diagnostic method according to claims 1 to 4, wherein when the filter is deactivated ( 5 ) a change ( 23 ) of the NOx value is introduced upstream of the catalyst. Diagnostic method according to claims 1 to 5, wherein for the calculation ( 12 ) of the NOx estimate a kinetic model of NOx conversion is used. Diagnostic method according to claims 1 to 6, wherein for the modeling ( 13 ) of the linear NOx model a linear time algorithm is used. Diagnostic method according to claim 1 to 7, wherein the steps are performed at sampling times with a sampling time of preferably one second. The diagnostic method according to claim 8, wherein the steps use values of past sampling times preferably in a time frame of twenty seconds. Diagnostic module for a filter ( 5 ) of a NOx sensor ( 4 ) of an exhaust system ( 1 ) with an SCR catalyst ( 3 ), having a signal input ( 8a ) for signals of the filter ( 5 ), a computing unit designed to carry out the diagnostic method according to claims 1 to 9 and a signal output ( 8b ) for outputting a result of the diagnostic procedure.

Images