METHOD OF CONTROLLING A DIESEL PARTICULATE FILTER

Abstract

A method and apparatus for controlling a diesel particulate filter is disclosed. The control is carried out with the aid of a soot loading model selected from a plurality of available soot loading models including at least one soot loading physical model which makes use of signals from an exhaust gas pressure sensor. A reliability score is allocated to each soot loading model such that the soot loading model with the best reliability score is used for the diesel particulate filter control. If a DPF parking effect is identified based on an engine shut-off time a DPF inlet temperature either the reliability score of the soot loading physical model or the reliability scores of all remaining (non-physical) soot loading models are changed, and use of the soot loading physical model is avoided or disabled to help the control of the diesel particulate filter.

Description

CROSS REFERENCE TO RELATED APPLICATION

This application claims priority to British Patent Application No. 1307754.0 filed Apr. 30, 2013, which is incorporated herein by reference in its entirety.

TECHNICAL FIELD

The technical field relates to a method of controlling a diesel particulate filter (DPF), and more particularly a method to the identify the DPF parking effect in order not to worsening the soot loading evaluation.

BACKGROUND

It is known that modern engines are provided with one or more exhaust aftertreatment devices. The aftertreatment devices may be any device configured to change the composition of the exhaust gases, such as a diesel oxidation catalyst (DOC) located in the exhaust line for degrading residual hydrocarbons (HC) and carbon oxides (CO) contained in the exhaust gas, and a diesel particulate filter (DPF) located in the exhaust line downstream the DOC, for capturing and removing diesel particulate matter (soot) from the exhaust gas. The diesel particulate filter collects liquid and solid particles in a porous substrate structure, while allowing exhaust gases to flow through. As it reaches its nominal storage capacity, the DPF needs to be cleaned by a process called regeneration. A DPF physical model returns a soot loading evaluation starting from the differential pressure signal read across the DPF by an exhaust gas pressure (EGP) sensor. This is applied to the overall soot estimation strategy in order to optimize the DPF regeneration frequency.

A particular condition related to the estimation of the soot loading appears during a prolonged vehicle disuse, if the soot in the DPF reaches very low temperatures. This condition is well known as parking effect. As a result, when the vehicle is started again, even if the soot quantity is not changed in the trap, there is a significant pressure drop reduction across the filter, thus leading to a soot loading under-estimation by the DPF physical model, taking into account the information of the EGP sensor.

In DE 10 2006 062515, a method of controlling a DPF is shown, which attempts to consider the parking effect. The method involves measuring concentration of one of exhaust gas components such as nitric oxide and nitrogen dioxide in an exhaust gas mass flow downstream of a DPF. A functional condition and/or a loading condition of the particle filter with soot are determined from the measured concentration. A diagnostic signal is assigned to the functional condition and/or loading condition. However, there is a need in the art to provide a method of controlling the DPF, which is more reliable in identifying the parking condition and avoiding soot loading under-prediction.

SUMMARY

An embodiment of the present disclosure provides a method of controlling a diesel particulate filter, wherein the control is carried out with the aid of a soot loading model selected from a plurality of available soot loading models. The soot loading models include at least one soot loading physical model which makes use of signals from an exhaust gas pressure sensor, wherein a reliability score is allocated to each soot loading model and wherein the soot loading model with the best reliability score is used for the diesel particulate filter control.

The method includes identifying a DPF parking effect if an engine shut-off time is longer than a first time threshold and a DPF inlet temperature is less than a first temperature threshold. This step can be carried out before or shortly after cranking. If a parking effect is identified because the above conditions are met, either the reliability score of the soot loading physical model or the reliability scores of all remaining (non-physical) soot loading models are changed, and use of the soot loading physical model is avoided or disabled to help the control of the diesel particulate filter.

In the first alternative the reliability score of the soot loading physical model is changed, such that its new reliability score indicates that it is less reliable. The score may be lowered, such that a lower score indicates a reduced reliability. In the second alternative the reliability scores of all remaining soot loading models, i.e. of all non-physical soot loading models, are changed. This is done such that the new scores render all remaining soot loading models to be more reliable than before and in particular to be more reliable than the soot loading physical model. In both alternatives, which as a matter of fact can be combined, the likelihood that the soot loading physical model is used for DPF control is reduced.

An apparatus is also disclosed for controlling a diesel particulate filter. The apparatus includes a DPF controller configured to control the DPF using a soot loading model selected from a plurality of available soot loading models. Again, the soot loading models include at least one soot loading physical model which makes use of signals from an exhaust gas pressure sensor. The controller is configured to allocate a reliability score for each soot loading model and to identify the soot loading model with the best reliability score for controlling the diesel particulate filter control. The controller is configured to identify a DPF parking effect if an engine shut-off time is longer than a first time threshold and a DPF inlet temperature is less than a first temperature threshold. The controller is configured to change the reliability score of the soot loading physical model or the reliability scores of all remaining (non-physical) soot loading models are changed, and use of the soot loading physical model is avoided or disabled to help the control of the diesel particulate filter.

An advantage of this strategy is that all needed parameters for identifying the parking effect are defined and the related countermeasure, i.e. not to use a physical soot loading model during the parking effect, is established.

According to another embodiment, the method further includes the step of restoring the reliability score of the physical model or the reliability scores of all remaining soot loading models to their previous values if the DPF inlet temperature is higher than a second temperature threshold and an elapsed time is longer than a second time threshold. This serves to allow that the soot loading physical model is used or enabled to help control the diesel particulate filter. The previous values are the reliability scores before the detection of the parking effect, so this step neutralizes the prior change of the reliability score or scores.

Similarly, the controller is configured to restore the reliability score of the physical model or the reliability scores of all remaining soot loading to their previous value(s) if the DPF inlet temperature is higher than a second temperature threshold and an elapsed time is longer than a second time threshold, which serves to to allow that the soot loading physical model is used or enabled to help the control the diesel particulate filter.

An advantage of this strategy is an ability to define all parameters, establishing until when said parking effect strategy should be maintained.

According to a further embodiment, changing a reliability score is carried out by lowering the reliability score of the at least one soot loading physical model. Assuming again that a lowered score indicates a less reliable model, this helps to avoid that a soot loading physical model has the best reliability score out of the multitude of available soot loading models. Similarly, the controller is configured to lower the reliability score of the at least one soot loading physical model, to avoid that a soot loading physical model has the best reliability score out of the multitude of available soot loading models. This improves the likelihood that soot estimation is correct if a parking effect is identified.

According to a still further embodiment, a change of reliability scores is carried out by increasing the reliability scores of all remaining soot loading models, that is the non-physical models, such that a non-physical soot loading model has a better or the best reliability score out of the multitude of available soot loading models. Similarly, the controller is configured to increase the reliability scores of all available soot loading models, to allow that a non-physical soot loading model has the best reliability score out of the multitude of available soot loading models. This embodiment, in an alternative way, brings the advantage that the soot loading physical models not having anymore the best reliability score among the other soot loading models will not be chosen as the soot loading model helping the control of the DPF.

According to another embodiment, the multitude of available soot loading models includes a soot loading statistical model, and the soot loading statistical model is chosen for helping the control of the diesel particulate filter if it shows a reliability score better than the reliability score of said soot loading physical model. Similarly, the controller is configured to choose a soot loading model wherein the multitude of available soot loading models includes a soot loading statistical model, and the soot loading statistical model is chosen for helping the control of the diesel particulate filter, if it shows a reliability score better than the reliability score of said soot loading physical model. In this way, the soot loading estimation will be performed by a model, which will not suffer the problem of the soot loading under-estimation, since it is not related to the exhaust gas pressure sensor signals, but it is an available soot loading statistical model.

The above-described embodiments may be readily implemented in an automotive system including an internal combustion engine, for example a Diesel engine having a diesel particulate filter, an exhaust gas pressure sensor, and an electronic control unit configured to control the diesel particulate in the above-described manner. In particular, the electronic control unit may include program code for carrying out all the steps of the method described above when it is executed by a computer, microprocessor or controller. The computer program may be part of a transitory computer program product such as an electromagnetic or optical signal which carries, by means of modulation such as QPSK, binary data representing the computer program. In the alternative the computer program may be part of a non-transitory computer program product which is embodied in a conventional data carrier such as a flash memory, an Asic, a CD or the like. The computer program product can be a component of a control apparatus for an internal combustion engine, including an Electronic Control Unit (ECU), a data carrier associated to the ECU, and the computer program stored in a data carrier. In this case, when the control apparatus executes the computer program control of the DPF described above is carried out.

BRIEF DESCRIPTION OF THE DRAWINGS

The present disclosure will hereinafter be described in conjunction with the following drawing figures, wherein like numerals denote like elements, and:

FIG. 1 shows an automotive system;

FIG. 2 is a section of an internal combustion engine shown in FIG. 1;

FIG. 3 is a schematic view of an after-treatment system according to the present disclosure;

FIGS. 4a-4c are a schematic representation of the parking effect;

FIG. 5 is a graph depicting the soot level under-estimation due to the parking effect; and

FIG. 6 shows the flowchart of the method according to the present disclosure.

DETAILED DESCRIPTION OF THE DRAWINGS

The following detailed description is merely exemplary in nature and is not intended to limit the present disclosure, the application or uses of the present disclosure. Furthermore, there is no intention to be bound by any theory presented in the preceding background or the following detailed description.

Some embodiments may include an automotive system 100, as shown in FIGS. 1 and 2, that includes an internal combustion engine (ICE) 110 having an engine block 120 defining at least one cylinder 125 having a piston 140 coupled to rotate a crankshaft 145. A cylinder head 130 cooperates with the piston 140 to define a combustion chamber 150.

A fuel and air mixture (not shown) is disposed in the combustion chamber 150 and ignited, resulting in hot expanding exhaust gasses causing reciprocal movement of the piston 140. The fuel is provided by at least one fuel injector 160 and the air through at least one intake port 210. The fuel is provided at high pressure to the fuel injector 160 from a fuel rail 170 in fluid communication with a high pressure fuel pump 180 that increase the pressure of the fuel received from a fuel source 190.

Each of the cylinders 125 has at least two valves 215, actuated by a camshaft 135 rotating in time with the crankshaft 145. The valves 215 selectively allow air or an air-fuel mixture into the combustion chamber 150 from the port 210 and alternately allow exhaust gases to exit through a port 220. In some examples, a cam phaser 155 may selectively vary the timing between the camshaft 135 and the crankshaft 145.

Air may be distributed to the air intake port(s) 210 through an intake manifold 200. An air intake duct 205 may provide air from the ambient environment to the intake manifold 200. In other embodiments, a throttle body 330 may be provided to regulate the flow of air into the manifold 200. In still other embodiments, a forced air system such as a turbocharger 230, having a compressor 240 rotationally coupled to a turbine 250, may be provided. Rotation of the compressor 240 increases the pressure and temperature of the air in the duct 205 and manifold 200. An intercooler 260 disposed in the duct 205 may reduce the temperature of the air. The turbine 250 rotates by receiving exhaust gases from an exhaust manifold 225 that directs exhaust gases from the exhaust ports 220 and through a series of vanes prior to expansion through the turbine 250. The exhaust gases exit the turbine 250 and are directed into an exhaust system 270. This example shows a variable geometry turbine (VGT) with a VGT actuator 290 arranged to move the vanes to alter the flow of the exhaust gases through the turbine 250. In other embodiments, the turbocharger 230 may be fixed geometry and/or include a waste gate.

The exhaust system 270 may include an exhaust pipe 275 having one or more exhaust aftertreatment devices 280. The aftertreatment devices may be any device configured to change the composition of the exhaust gases. Some examples of aftertreatment devices 280 include, but are not limited to, catalytic converters (two and three way), oxidation catalysts, lean NOx traps, hydrocarbon adsorbers, selective catalytic reduction (SCR) systems, and particulate filters 282. In particular, a diesel particulate filter (or DPF) is a device designed to remove diesel particulate matter or soot from the exhaust gas of a diesel engine. Wall-flow diesel particulate filters usually remove 85% or more of the soot, and under certain conditions can attain soot removal efficiencies of close to 100%. Some filters are single-use, intended for disposal and replacement once full of accumulated ash. Others are designed to burn off the accumulated particulate either passively through the use of a catalyst or by active means such as a fuel burner which heats the filter to soot combustion temperatures; engine programming to run when the filter is full in a manner that elevates exhaust temperature or produces high amounts of NOx to oxidize the accumulated ash, or through other methods. The process of particulate burn off is known as “filter regeneration.”

Other embodiments may include an exhaust gas recirculation (EGR) system 300 coupled between the exhaust manifold 225 and the intake manifold 200. The EGR system 300 may include an EGR cooler 310 to reduce the temperature of the exhaust gases in the EGR system 300. An EGR valve 320 regulates a flow of exhaust gases in the EGR system 300.

The automotive system 100 may further include an electronic control unit (ECU) 450 in communication with one or more sensors and/or devices associated with the engine 110 and equipped with a data carrier 40. The ECU 450 may receive input signals from various sensors configured to generate the signals in proportion to various physical parameters associated with the engine 110. The sensors include, but are not limited to, a mass airflow and temperature sensor 340, a manifold pressure and temperature sensor 350, a combustion pressure sensor 360, coolant and oil temperature and level sensors 380, a fuel rail pressure sensor 400, a cam position sensor 410, a crank position sensor 420, exhaust pressure and temperature sensors 430, an EGR temperature sensor 440, and an accelerator pedal position sensor 445. Furthermore, the ECU 450 may generate output signals to various control devices that are arranged to control the operation of the engine 110 and its various sensors and devices, including, but not limited to, the fuel injectors 160, the throttle body 330, the EGR Valve 320, the VGT actuator 290, and the cam phaser 155. Note, dashed lines are used to indicate communication between the ECU 450 and the various sensors and devices, but some are omitted for clarity.

Turning now to the ECU 450, this apparatus may include a micro controller or digital central processing unit (CPU) in communication with a memory system and an interface bus. The CPU is configured to execute instructions stored as a program in the memory system, and send and receive signals to and from the interface bus. The memory system may include various storage types including optical storage, magnetic storage, solid state storage, and other non-volatile memory. The interface bus may be configured to send, receive, and modulate analog and/or digital signals to/from the various sensors and control devices. The program may embody the methods disclosed herein, allowing the CPU to carryout out the steps of such methods and control the ICE 110.

The diesel particulate filter (DPF) 282 collects liquid and solid particles in a porous substrate structure, while allowing exhaust gases to flow through. As it reaches its nominal storage capacity, needs to be cleaned by a process called regeneration, during which the exhaust gas temperature is increased substantially to create a condition, whereby the soot contained in the DPF is burned, that is to say oxidized. A DPF physical model returns a soot loading evaluation starting from the differential pressure signal read across the DPF by an exhaust gas pressure (EGP) sensor 283, as shown in FIG. 3. This basic information is corrected, considering the pressure drop on the DPF filter in clean conditions and divided by the volumetric flow rate of the exhaust gas, in order to obtain the so called “flow resistance”. The physical model soot estimation is directly correlated to the flow resistance information, that is returned to give an accurate indication of the soot storage level at different exhaust conditions (especially in terms of temperature), wherein the only EGP information could not be accurate. This physical model is applied to the overall soot estimation strategy in order to optimize the DPF regeneration frequency.

As mentioned, a particular condition related to the estimation of the soot loading is the so called DPF parking effect. The DPF parking effect takes place during a prolonged engine shut-off (an engine shut-off time threshold can be established by calibration) and if the soot in DPF reaches very low temperatures (such temperature can be evaluated at the DPF inlet and a DPF inlet temperature threshold can be established by calibration). During this period, the particulate properties change leading to a remarkable decrease of pressure signal evaluated across the DPF. It means that, when the vehicle is started again, even if the soot quantity is not changed in the trap, there is a significant pressure drop reduction across the filter, thus leading to a soot loading under-estimation by the DPF physical model, which takes into account the information of the EGP sensor 283.

In FIG. 4 is schematically shown the so called parking effect. In the upper part, (FIG. 4a) the soot layer 500 is deposited on the DPF substrate 510. This is a normal situation before the vehicle parking, for example in a case when there is a high pressure drop, due to the fact that the last regeneration process was performed a long time ago. In FIG. 4b, supposing a long parking time and low DPF inlet temperature, the increasing of the soot permeability leads to a lower pressure drop sensed across the filter, notwithstanding the soot layer 501 remains unchanged. Finally, as shown in FIG. 4c, after the parking the following soot 500 has the same properties of the soot before the vehicle parking. Therefore, the pressure drop will increase accordingly to the soot loading, but starting from a lower level and without any possibility to consider this offset until a regeneration process will take place.

As already mentioned, during the vehicle parking the soot quantity stored inside the DPF does not change. But, in terms of physical model behavior, this results in a lower drop of the differential pressure measured across the filter. This is due to the fact that the particulate permeability increases during this period, so that the particulate offers a lower resistance and consequently there is a lower pressure drop across the DPF. The pressure drop reduction leads to an under-estimation of the physical soot model based on it. In FIG. 5, the DPF loading and the coolant temperature behavior vs. time are shown. As can be seen, after the cold start, the apparent DPF loading is lowered by about 28%. As soon as new soot is stored inside the DPF, the physical model of the DPF loading will show an increase of the loading itself, but without taking into account the lost amount prediction. This potentially drastic under-evaluation by the soot physical model might lead to a not properly working DPF regeneration, due to a real amount of soot too high at the beginning of the regeneration process.

The present method starts from the following consideration: the current model, which estimates soot loading models, also known as ranked model, is a software that evaluates the reliability of different available soot loading models (e.g. the physical model) in every driving condition, for example, checking the actual value of signals, like ambient air temperature, exhaust temperature, exhaust mass flow rate, exhaust NOx concentration. In this way, knowing the strength and the drawbacks of each soot model, the software recognizes if the current engine condition allows each soot loading model to return a reliable evaluation of the soot trapped. Consequently, the reliability score, which is associated with a soot loading model that does not return the proper estimation is modified by means of calibration factors. In particular, a change of the reliability score can be more or less severe, depending on how bad is the soot model evaluation in every specific case. Then, case by case, the soot loading information, which comes from the best ranked model is chosen and contributes to build a new soot loading information that always takes the best estimation available.

Some relevant parameters have to be taken into account. First of all, the elapsed time with the engine not running (vehicle disuse); the reduction of the pressure drop signal is dependent on the length of the engine shut-off. For example, at a DPF inlet temperature of −10° C., after 2 h the reduction in pressure drop is negligible, while after 6 h, such reduction is about 16%. Therefore, a first time threshold t₁ for the engine shut-off time (condition 1) is requested to enable the strategy. Also relevant is the DPF inlet temperature, since the parking effect takes place only below a certain temperature. For example in a specific engine application, for an engine shut-off time of about 12 h, at 0° C. the reduction in pressure drop is negligible, while at −25° C., such reduction is about 22%. Therefore, a first temperature threshold T₁ for DPF inlet temperature (condition 2) at engine start is also requested to enable the strategy.

Once parking effect is recognized according to the previous conditions (shut-off time, DPF inlet temperature), the strategy must be kept active even if the DPF temperature condition is no longer satisfied, since the pressure drop takes place even if the DPF inlet temperature is above the previous threshold (condition 2) and until the same temperature has not overcome a different threshold for a certain time. In fact, by performing two different cold starts in the same conditions (shut-off time: 12 h, pre-conditioning temperature: −25° C.), it results that: the time after cranking corresponding to the end of parking effect is different (e.g. 330 s vs 290 s); the DPF inlet temperature corresponding to the end of parking effect is different (e.g. 148° C. vs 167° C.); the time interval between a DPF inlet temperature of 125° C. and the end of parking effect is the same (e.g. 150 s). Therefore, a second temperature threshold T₂ for DPF inlet temperature and a second time threshold t₂ (condition 3) are requested. When these thresholds are overcome, it is possible to disable the strategy, since the physical model can be again considered reliable.

In FIG. 6 a schematic flowchart of the new method is shown. After the engine disuse due to the vehicle parking, with the engine in the purely optional condition S20 “wait for cranking”, the present method identifies S21 the DPF parking effect, if the engine shut-off time is higher than a first calibratable time threshold t₁ (condition 1, for example, 6 h) and the DPF inlet temperature is lower than a first calibratable temperature threshold T₁ (condition 2, e.g. −20° C.).

After having identified the DPF parking effect, the method will change S22 the reliability score RS of either the soot loading physical model and/or of all remaining soot loading models. For example, the reliability score RS of all available physical models for soot loading estimation, based on the exhaust gas pressure sensor 283 can be lowered. As an alternative, the reliability score RS of all available non-physical model (for example, soot loading statistical models) can be increased. In any case, for helping the control of the DPF (i.e. performing the soot loading estimation) the best ranked model will not be any more a physical model but a different available one, for example a statistical model or however any non-physical model, which is not based on the EGP sensor signal. As known, a statistical model is a model that links a driving style to an average soot production. As for driving style can be intended a driving profile (for example, “urban”, “extra-urban” or “highway” that is caught by the model, checking the current values of some key vehicle parameters (for example, gear ratio, engine and vehicle speed, load). For each of these driving conditions an average soot emitted can be identified by means of experimental activities: this information, is then used to build the soot level evaluation of the model. When the method recognizes the DPF parking effect, therefore, a statistical model will have a better rank than a DPF physical model and will be used instead, thus avoiding the problem of the soot loading under-estimation.

The method considers the parking effect still valid until when the condition 3, related to the DPF temperature, will not be satisfied and the condition 3 about the related elapsed time will not be satisfied as well. In other words, the method will restore S25 the previous reliability score RS values (i.e. the soot loading physical model will be again the best ranked model), if the DPF inlet temperature is higher than S23 the second temperature threshold T₂ and, after this temperature condition has been satisfied, the elapsed time is higher than S24 the second time threshold t₂. When applying the method in the real world, the engine cranking can finally be performed S26, which however is out of scope of the present disclosure which concerns the DPF control only.

The benefit of this new method is easy to be understood. While, without any correction, in case of cold start at −25° C. after a shut-off time of 12 h, a large drop (about −28%) is observed in the ranked model evaluation, due to the error that affects the physical model, in case of detected parking effect, with the engine still off, the ranked model is frozen with the last available value of the soot loading. As soon as the engine is started, the method here proposed modifies the soot loading models reliability score RS and the ranked model considers the soot flow coming from other available models, for example statistical models. After the DPF inlet temperature is higher than a calibratable threshold and the debouncing time is elapsed, parking detection is switched off. Therefore, the physical model reliability score RS comes back to the previous value and the ranked model again considers the soot loading physical model as the most reliable: the under-estimation prevented is about of 40%.

Summarizing, the present strategy allows an improved accuracy in soot loading evaluation, that means an optimization of the regeneration frequency, so avoiding thermal stress for the components, due to soot loading under-estimation.

While at least one exemplary embodiment has been presented in the foregoing summary and detailed description, it should be appreciated that a vast number of variations exist. It should also be appreciated that the exemplary embodiment or exemplary embodiments are only examples, and are not intended to limit the scope, applicability, or configuration in any way. Rather, the foregoing summary and detailed description will provide those skilled in the art with a convenient road map for implementing at least one exemplary embodiment, it being understood that various changes may be made in the function and arrangement of elements described in an exemplary embodiment without departing from the scope as set forth in the appended claims and their legal equivalents.

Claims

1-9. (canceled)

10. A method for controlling a diesel particulate filter (DPF) comprising:

defining a plurality of soot loading models including a soot loading physical model and at least one soot loading non-physical model which makes use of signals from an exhaust gas pressure sensor;

allocating a reliability score to each of the plurality of soot loading models;

identifying a DPF parking effect when an engine shut-off time is longer than a first time threshold and an inlet temperature of the diesel particulate filter is less than a first temperature threshold; and

changing the reliability score for either the soot loading physical model or all of the soot loading non-physical models from a previous value for disabling the soot loading physical model in the control of the diesel particulate filter;

wherein the soot loading model with the highest reliability score is enabled for controlling the diesel particulate filter.

11. The method according to claim 10, further comprising restoring the reliability scores for either the soot loading physical model or of all of the soot loading non-physical models to the previous values when the inlet temperature is higher than a second temperature threshold and an elapsed time is longer than a second time threshold.

12. The method according to claim 10 wherein changing a reliability score is carried out by lowering the reliability score of the soot loading physical model.

13. The method according to claim 10 wherein changing the reliability score is carried out by increasing the reliability score for all of the soot loading non-physical models.

14. The method according to claim 10 wherein the at least one soot loading non-physical models comprises a soot loading statistical model.

15. A control apparatus for an exhaust system of an internal combustion of the type having a diesel particulate filter and an exhaust gas pressure sensor, the control apparatus comprising an electronic control unit having:

a memory storing an instruction set and a plurality of soot loading models including a soot loading physical model and at least one soot loading non-physical model, wherein the soot loading physical model makes use of signals from the exhaust gas pressure sensor; and

a microcontroller executing the instruction set and configured to: allocate a reliability score to each of the plurality of soot loading models; identify a DPF parking effect when an engine shut-off time is longer than a first time threshold and an inlet temperature of the diesel particulate filter is less than a first temperature threshold; and change the reliability score for either the soot loading physical model or all of the soot loading non-physical models from a previous value for disabling the soot loading physical model in the control of the diesel particulate filter; wherein the soot loading model with the highest reliability score is enabled for controlling the diesel particulate filter.

16. The control apparatus according to claim 15, wherein the microcontroller is configured to restore the reliability scores for either the soot loading physical model or of all of the soot loading non-physical models to the previous values when the inlet temperature is higher than a second temperature threshold and an elapsed time is longer than a second time threshold.

17. The control apparatus according to claim 15, wherein the microcontroller is configured to change a reliability score by lowering the reliability score of the soot loading physical model.

18. The control apparatus according to claim 15, wherein the microcontroller is configured to change the reliability score by increasing the reliability score for all of the soot loading non-physical models.

19. The control apparatus according to claim 15, wherein the at least one soot loading non-physical models comprises a soot loading statistical model.