JP6435730B2 - Control device for internal combustion engine - Google Patents

Description

  The present invention relates to a control device for an internal combustion engine, and more particularly, to learning correction for an injector.

  Conventionally, a NOx occlusion reduction type catalyst is known as a catalyst for reducing and purifying nitrogen compounds (NOx) in exhaust gas discharged from an internal combustion engine. This NOx occlusion reduction type catalyst occludes NOx contained in the exhaust when the exhaust is in a lean atmosphere, and reduces and purifies NOx occluded by hydrocarbons contained in the exhaust when the exhaust is in a rich atmosphere. Detoxify and release. For this reason, when the NOx occlusion amount of the catalyst reaches a predetermined amount, so-called NOx purge that makes the exhaust gas rich by exhaust pipe injection needs to be performed periodically in order to recover the NOx occlusion capacity (for example, patents) Reference 1).

The NOx occlusion reduction type catalyst also occludes sulfur oxide (hereinafter referred to as SOx) contained in the exhaust gas. When this SOx occlusion amount increases, there is a problem that the NOx purification ability of the NOx occlusion reduction type catalyst is lowered. For this reason, when the SOx occlusion amount reaches a predetermined amount, unburnt fuel is supplied to the upstream oxidation catalyst by exhaust pipe injection in order to remove SOx from the NOx occlusion reduction catalyst and recover from S poisoning. Therefore, it is necessary to periodically perform a so-called SOx purge for raising the exhaust temperature to the SOx separation temperature (see, for example, Patent Document 2). In Patent Document 3, a sub-feedback correction coefficient is calculated based on the output of a downstream side exhaust gas sensor arranged downstream of the catalyst, an average value of the sub-feedback correction coefficient is learned as a sub-feedback learning coefficient, It is described that the feedback learning coefficient is reflected in the fuel injection amount.

JP 2008-202425 A JP 2009-47086 A JP 2005-048711 A

  By the way, in order to maintain the exhaust air-fuel ratio in a desired rich state during NOx purge or SOx purge, it is preferable to improve the injection amount accuracy of post injection. However, an injector may have an error between the commanded injection amount and the actual injection amount due to the effects of aging, characteristic changes, individual differences, etc .. Need to be eliminated.

  An object of the disclosed system is to effectively reduce an error between the instructed injection amount of the injector and the actual injection amount.

The disclosed system includes a lambda sensor provided in an exhaust passage of an internal combustion engine for acquiring an actual exhaust lambda value, an estimated exhaust lambda value estimated from an operating state of the internal combustion engine, and an actual acquired by the lambda sensor. Learning means for learning a correction coefficient for the command injection amount to be output to the injector of the internal combustion engine based on a difference from the lambda value, and command injection to be output to the injector based on the correction coefficient learned by the learning means Correction means for correcting the amount , wherein the learning means calculates the correction coefficient based on a value obtained by multiplying a difference between the estimated lambda value and the actual lambda value by a correction sensitivity coefficient, and the learning means The correction sensitivity coefficient is calculated based on the actual lambda value, and the correction sensitivity coefficient increases as the actual lambda value increases.

  According to the disclosed system, an error between the instructed injection amount of the injector and the actual injection amount can be effectively reduced.

1 is an overall configuration diagram showing an exhaust purification system according to an embodiment. It is a timing chart explaining SOx purge control concerning this embodiment. It is a block diagram which shows the setting process of the MAF target value at the time of SOx purge lean control which concerns on this embodiment. It is a block diagram which shows the setting process of the target injection amount at the time of SOx purge rich control which concerns on this embodiment. It is a timing chart explaining catalyst temperature adjustment control of SOx purge control concerning this embodiment. It is a timing chart figure explaining NOx purge control concerning this embodiment. It is a block diagram which shows the setting process of the MAF target value at the time of NOx purge lean control which concerns on this embodiment. It is a block diagram which shows the setting process of the target injection quantity at the time of NOx purge rich control which concerns on this embodiment. It is a block diagram which shows the process of the injection amount learning correction | amendment of the injector which concerns on this embodiment. It is a flowchart explaining the calculation process of the learning correction coefficient which concerns on this embodiment. It is a block diagram which shows the setting process of the MAF correction coefficient which concerns on this embodiment.

  Hereinafter, an exhaust purification system according to an embodiment of the present invention will be described with reference to the accompanying drawings.

  As shown in FIG. 1, each cylinder of a diesel engine (hereinafter simply referred to as an engine) 10 is provided with an injector 11 that directly injects high-pressure fuel stored in a common rail (not shown) into each cylinder. The fuel injection amount and fuel injection timing of each injector 11 are controlled in accordance with an instruction signal input from an electronic control unit (hereinafter referred to as ECU) 50.

  An intake passage 12 for introducing fresh air is connected to the intake manifold 10A of the engine 10, and an exhaust passage 13 for leading the exhaust to the outside is connected to the exhaust manifold 10B. In the intake passage 12, an air cleaner 14, an intake air amount sensor (hereinafter referred to as MAF sensor) 40, a compressor 20A of the variable displacement supercharger 20, an intercooler 15, an intake throttle valve 16 and the like are provided in order from the intake upstream side. ing. The exhaust passage 13 is provided with a turbine 20B of the variable displacement supercharger 20, an exhaust aftertreatment device 30 and the like in order from the exhaust upstream side. In FIG. 1, reference numeral 41 denotes an engine speed sensor, reference numeral 42 denotes an accelerator opening sensor, and reference numeral 46 denotes a boost pressure sensor.

  The EGR device 21 includes an EGR passage 22 that connects the exhaust manifold 10B and the intake manifold 10A, an EGR cooler 23 that cools EGR gas, and an EGR valve 24 that adjusts the EGR amount.

  The exhaust aftertreatment device 30 is configured by arranging an oxidation catalyst 31, a NOx occlusion reduction type catalyst 32, and a particulate filter (hereinafter simply referred to as a filter) 33 in order from the exhaust upstream side in a case 30A. The exhaust passage 13 upstream of the oxidation catalyst 31 is provided with an exhaust pipe injection device 34 that injects unburned fuel (mainly HC) into the exhaust passage 13 in accordance with an instruction signal input from the ECU 50. It has been.

  The oxidation catalyst 31 is formed, for example, by carrying an oxidation catalyst component on the surface of a ceramic carrier such as a honeycomb structure. When the unburned fuel is supplied by the post-injection of the exhaust pipe injector 34 or the injector 11, the oxidation catalyst 31 oxidizes this and raises the exhaust temperature.

  The NOx storage reduction catalyst 32 is formed, for example, by supporting an alkali metal or the like on the surface of a ceramic carrier such as a honeycomb structure. The NOx occlusion reduction type catalyst 32 occludes NOx in the exhaust when the exhaust air-fuel ratio is in a lean state, and occludes with a reducing agent (HC or the like) contained in the exhaust when the exhaust air-fuel ratio is in a rich state. NOx is reduced and purified.

  The filter 33 is formed, for example, by arranging a large number of cells partitioned by porous partition walls along the flow direction of the exhaust gas and alternately plugging the upstream side and the downstream side of these cells. . The filter 33 collects PM in the exhaust gas in the pores and surfaces of the partition walls, and when the estimated amount of PM deposition reaches a predetermined amount, so-called filter forced regeneration is performed in which the PM is burned and removed. Filter forced regeneration is performed by supplying unburned fuel to the upstream side oxidation catalyst 31 by exhaust pipe injection or post injection, and raising the exhaust temperature flowing into the filter 33 to the PM combustion temperature.

  The first exhaust temperature sensor 43 is provided on the upstream side of the oxidation catalyst 31 and detects the exhaust temperature flowing into the oxidation catalyst 31. The second exhaust temperature sensor 44 is provided between the oxidation catalyst 31 and the NOx storage reduction catalyst 32 and detects the exhaust temperature flowing into the NOx storage reduction catalyst 32. The NOx / lambda sensor 45 is provided on the downstream side of the filter 33, and detects the NOx value and lambda value (hereinafter also referred to as excess air ratio) of the exhaust gas that has passed through the NOx storage reduction catalyst 32.

  The ECU 50 performs various controls of the engine 10 and the like, and includes a known CPU, ROM, RAM, input port, output port, and the like. In order to perform these various controls, sensor values of the sensors 40 to 45 are input to the ECU 50. In addition, the ECU 50 includes a filter forced regeneration control unit 51, a SOx separation processing unit 60, a NOx separation processing unit 70, a MAF follow-up control unit 80, an injection amount learning correction unit 90, and a MAF correction coefficient calculation unit 95. As a functional element. Each of these functional elements will be described as being included in the ECU 50 which is an integral hardware, but any one of these may be provided in separate hardware.

[Filter forced regeneration control]
The filter forced regeneration control unit 51 estimates the PM accumulation amount of the filter 33 from the travel distance of the vehicle or the differential pressure across the filter detected by a differential pressure sensor (not shown), and the estimated PM accumulation amount has a predetermined upper limit threshold. If it exceeds, the forced regeneration flag F _DPF is turned on (see time t _{1 in} FIG. 2). When the forced regeneration flag F _DPF is turned on, an instruction signal for causing the exhaust pipe injection device 34 to execute exhaust pipe injection is transmitted, or an instruction signal for causing each injector 11 to perform post injection is transmitted. The exhaust temperature is raised to the PM combustion temperature (for example, about 550 ° C.). This forced regeneration flag F _DPF is turned off when the PM accumulation estimated amount falls to a predetermined lower threshold (determination threshold) indicating combustion removal (see time t _{2 in} FIG. 2). Note that the determination threshold value for turning off the forced regeneration flag F _DPF may be based on, for example, the upper limit elapsed time or the upper limit cumulative injection amount from the start of filter forced regeneration (F _DPF = 1).

[SOx purge control]
The SOx desorption processing unit 60 makes the exhaust rich and raises the exhaust temperature to a sulfur desorption temperature (for example, about 600 ° C.) to recover the NOx occlusion reduction type catalyst 32 from SOx poisoning (hereinafter, this control). (Referred to as SOx purge control).

FIG. 2 shows a timing chart of the SOx purge control of this embodiment. As shown in FIG. 2, SOx purge flag F _SP to start SOx purge control is turned off and on at the same time forced regeneration flag F _DPF (see time t ₂ in FIG. 2). As a result, it is possible to efficiently shift to the SOx purge control from the state in which the exhaust gas temperature has been raised by the forced regeneration of the filter 33, and the fuel consumption can be effectively reduced.

  In the present embodiment, the enrichment by the SOx purge control is performed by adjusting the excess air ratio to the lean side from the theoretical air-fuel ratio equivalent value (about 1.0) from the steady operation (for example, about 1.5) by the air system control. SOx purge lean control for reducing to 1 target excess air ratio (for example, about 1.3) and injection system control to reduce the excess air ratio from the first target excess air ratio to the second target excess air ratio on the rich side (for example, about 0) This is realized by using together with the SOx purge rich control that lowers to .9). Details of the SOx purge lean control and the SOx purge rich control will be described below.

[Air system control for SOx purge lean control]
FIG. 3 is a block diagram showing a process for setting the MAF target value MAF _{SPL_Trgt} during SOx purge lean control. The first target excess air ratio setting map 61 is a map that is referred to based on the engine speed Ne and the accelerator opening Q (the fuel injection amount of the engine 10), and the engine speed Ne, the accelerator opening Q, A target value λ _{SPL_Trgt} (first target air excess rate) at the time of SOx purge lean control corresponding to is preset based on experiments or the like.

First, the excess air ratio target value λ _{SPL_Trgt} at the time of SOx purge lean control is read from the first target excess air ratio setting map 61 using the engine speed Ne and the accelerator opening Q as input signals, and the MAF target value calculation unit 62 Entered. Further, the MAF target value calculation unit 62 calculates the MAF target value MAF _{SPL_Trgt} at the time of SOx purge lean control based on the following formula (1).
MAF _{SPL_Trgt} = λ _{SPL_Trgt} × Q _{fnl_corrd} × Ro _Fuel × AFR _sto / _{Maf_corr} (1)

In Equation (1), Q _{fnl_corrd} is a learning-corrected fuel injection amount (excluding post-injection) described later, Ro _Fuel is a fuel specific gravity, AFR _sto is a stoichiometric air-fuel ratio, and _{Maf_corr} is a MAF correction coefficient described later. Yes.

MAF target value MAF _{SPL_Trgt} calculated by the MAF target value calculation unit 62, when the SOx purge flag F _SP is turned on (see time t ₂ in FIG. 2) is input to the lamp unit 63. The ramp processing unit 63 reads the ramp coefficient from the ramp coefficient maps 63A and _63B using the engine speed Ne and the accelerator opening Q as input signals, and outputs the MAF target ramp value MAF _{SPL_Trgt_Ramp} to which the ramp coefficient is added to the valve control unit 64. To enter.

The valve control unit 64 throttles the intake throttle valve 16 to the _close side and feeds back the EGR valve 24 to the open side so that the actual MAF value MAF _Act input from the MAF sensor 40 becomes the MAF target ramp value MAF _{SPL_Trgt_Ramp.} Execute control.

Thus, in the present embodiment, the MAF target value MAF _{SPL_Trgt} is set based on the excess air ratio target value λ _{SPL_Trgt} read from the first target excess air ratio setting map 61 and the fuel injection amount of each injector 11. The air system operation is feedback controlled based on the MAF target value MAF _{SPL_Trgt} . Thus, without providing a lambda sensor upstream of the NOx storage reduction catalyst 32, or even when a lambda sensor is provided upstream of the NOx storage reduction catalyst 32, the sensor value of the lambda sensor is not used. The exhaust can be effectively reduced to a desired excess air ratio required for SOx purge lean control.

_Further, by using the fuel injection amount Q _{fnl_corrd} after learning correction as the fuel injection amount of each injector 11, the MAF target value MAF _{SPL_Trgt} can be set by feedforward control, and the aging deterioration and characteristic change of each injector 11 can be _achieved. The influence of individual differences can be effectively eliminated.

Further, by adding a ramp coefficient that is set according to the operating state of the engine 10 to the MAF target value MAF _{SPL_Trgt} , it is possible to prevent misfire of the engine 10 due to a sudden change in the intake air amount, deterioration of drivability due to torque fluctuation, and the like. It can be effectively prevented.

[Fuel injection amount setting for SOx purge rich control]
FIG. 4 is a block diagram showing processing for setting the target injection amount Q _{SPR_Trgt} (injection amount per unit time) of exhaust pipe injection or post injection in SOx purge rich control. The second target excess air ratio setting map 65 is a map that is referred to based on the engine speed Ne and the accelerator opening Q, and at the time of SOx purge rich control corresponding to the engine speed Ne and the accelerator opening Q. The air excess rate target value λ _{SPR_Trgt} (second target air excess rate) is set in advance based on experiments or the like.

First, the excess air ratio target value λ _{SPR_Trgt} at the time of SOx purge rich control is read from the second target excess air ratio setting map 65 using the engine speed Ne and the accelerator opening Q as input signals, and an injection quantity target value calculation unit 66. Further, the injection amount target value calculation unit 66 calculates the target injection amount Q _{SPR_Trgt} at the time of SOx purge rich control based on the following formula (2).
Q _{SPR_Trgt} = MAF _{SPL_Trgt} × _{Maf_corr} / (λ _{SPR_Target} × Ro _Fuel × AFR _sto ) −Q _{fnl_corrd} (2)

In Formula (2), MAF _{SPL_Trgt} is the MAF target value at the time of SOx purge _{lean, and} is input from the MAF target value calculation unit 62 described above. Q _{fnlRaw_corrd} is a fuel injection amount (excluding post-injection) before application of learning corrected MAF tracking control described later, Ro _Fuel is fuel specific gravity, AFR _sto is a theoretical air-fuel ratio, and _{Maf_cor} is a MAF correction coefficient described later. Show.

The target injection amount Q _{SPR_Trgt} calculated by the injection amount target value calculation unit 66 is transmitted as an injection instruction signal to the exhaust pipe injector 34 or each injector 11 when a SOx purge rich flag F _SPR described later is turned on.

As described above, in this embodiment, the target injection amount Q _{SPR_Trgt} is set based on the air excess rate target value λ _{SPR_Trgt} read from the second target air excess rate setting map 65 and the fuel injection amount of each injector 11. It has become. Thus, without providing a lambda sensor upstream of the NOx storage reduction catalyst 32, or even when a lambda sensor is provided upstream of the NOx storage reduction catalyst 32, the sensor value of the lambda sensor is not used. The exhaust can be effectively reduced to a desired excess air ratio required for SOx purge rich control.

_Further, by using the fuel injection amount Q _{fnl_corrd} after learning correction as the fuel injection amount of each injector 11, it becomes possible to set the target injection amount Q _{SPR_Trgt} by feedforward control, and the aging deterioration and characteristic change of each injector 11. Etc. can be effectively eliminated.

[Catalyst temperature adjustment control for SOx purge control]
The exhaust temperature (hereinafter also referred to as catalyst temperature) flowing into the NOx occlusion reduction type catalyst 32 during the SOx purge control is the SOx for performing exhaust pipe injection or post injection as shown at times t _{2 to} t ₄ in FIG. The purge rich flag F _SPR is controlled by alternately switching on / off (rich / lean). When the SOx purge rich flag F _SPR is turned on (F _SPR = 1), the catalyst temperature rises by exhaust pipe injection or post injection (hereinafter, this period is referred to as an injection period T _{F_INJ} ). On the other hand, when the SOx purge rich flag F _SPR is turned off, the catalyst temperature is lowered by stopping the exhaust pipe injection or the post injection (hereinafter, this period is referred to as an interval _{TF_INT} ).

In the present embodiment, the injection period _{TF_INJ} is set by reading values corresponding to the engine speed Ne and the accelerator opening Q from an injection period setting map (not shown) created in advance by experiments or the like. In this injection time setting map, an injection period required to reliably reduce the excess air ratio of exhaust gas obtained in advance through experiments or the like to the second target excess air ratio is set according to the operating state of the engine 10. ing.

The interval _{TF_INT} is set by feedback control when the SOx purge rich flag F _{SPR at} which the catalyst temperature is highest is switched from on to off. Specifically, the proportional control for changing the input signal in proportion to the deviation ΔT between the target catalyst temperature and the estimated catalyst temperature when the SOx purge rich flag F _SPR is turned off, and the time integral value of the deviation ΔT are proportional. This is processed by PID control constituted by integral control for changing the input signal and differential control for changing the input signal in proportion to the time differential value of the deviation ΔT. The target catalyst temperature is set at a temperature at which SOx can be removed from the NOx storage reduction catalyst 32. The estimated catalyst temperature is, for example, the inlet temperature of the oxidation catalyst 31 detected by the first exhaust temperature sensor 43, and the oxidation catalyst 31. It may be estimated based on the exothermic reaction in the NOx occlusion reduction type catalyst 32 or the like.

As shown at time t ₁ in FIG. 5, when the SOx purge flag F _SP is turned on at the end of forced filter regeneration (F _DPF = 0), the SOx purge rich flag F _SPR is also turned on, and the previous SOx purge control is performed. The interval _{TF_INT} that is sometimes feedback calculated is also reset. That is, for the first time immediately after the forced filter regeneration, exhaust pipe injection or post injection is executed in accordance with the injection period _{TF_INJ_1} set in the injection period setting map (see times t _{1 to} t _{2 in} FIG. 5). As described above, since the SOx purge control is started from the SOx purge rich control without performing the SOx purge lean control, the fuel gas consumption is promptly shifted to the SOx purge control without lowering the exhaust temperature that has been raised by the forced filter regeneration. Can be reduced.

Then, when the SOx purge rich flag F _SPR is turned off with the passage of the injection period T _{F_INJ_1,} until interval T _{F_INT_1} set by PID control has elapsed, SOx purge rich flag F _SPR is turned off (time in FIG. 5 t see ₂ ~t _3). Further, when the SOx purge rich flag F _SPR is turned on by the lapse of the interval T _{F_INT_1,} injection period T _{F_INJ_2} exhaust pipe injection or post injection according to is performed again (see time t ₃ ~t ₄ of 5 ). Thereafter, the switching on and off of these SOx purge rich flag F _SPR is repeatedly executed until the SOx purge flag F _SP is turned off (see time t _n in FIG. 5) by the completion judgment of the SOx purge control described later.

As described above, in the present embodiment, the injection period T _{F_INJ} for raising the catalyst temperature and lowering the excess air ratio to the second target excess air ratio is set from a map referred to based on the operating state of the engine 10, The interval _{TF_INT} for lowering the catalyst temperature is processed by PID control. This makes it possible to reliably reduce the excess air ratio to the target excess ratio while effectively maintaining the catalyst temperature during the SOx purge control within a desired temperature range necessary for the purge.

[Determining completion of SOx purge control]
SOx purge control, (1) SOx purge flag F from on the _SP injection quantity of the exhaust pipe injection or post injection accumulated, when the amount of the cumulative injected has reached the predetermined upper limit threshold amount, of (2) SOx purge control When the elapsed time counted from the start reaches a predetermined upper threshold time, (3) calculation is performed based on a predetermined model formula including the operating state of the engine 10 and the sensor value of the NOx / lambda sensor 45 as input signals. If any of the conditions in the case of SOx adsorption amount of NOx occlusion-reduction catalyst 32 has decreased to a predetermined threshold value indicating a SOx removal success is established, SOx purge flag F _SP is terminated by turning off the (time t ₄ in FIG. 2 , reference time t _n in FIG. 5).

  As described above, in this embodiment, when the SOx purge control end condition is provided with the upper limit of the cumulative injection amount and the elapsed time, the fuel consumption amount when the SOx purge does not progress due to a decrease in the exhaust temperature or the like. Can be effectively prevented from becoming excessive.

[NOx purge control]
The NOx detachment processing unit 70 restores the NOx storage capacity of the NOx storage reduction catalyst 32 by making the exhaust atmosphere rich and detoxifying and releasing NOx stored in the NOx storage reduction catalyst 32 by reduction purification. Control (hereinafter, this control is referred to as NOx purge control) is executed.

The NOx purge flag F _NP for starting the NOx purge control is turned on when the NOx emission amount per unit time is estimated from the operating state of the engine 10 and the estimated cumulative value ΣNOx obtained by accumulating this exceeds a predetermined threshold value ( (See time t _{1 in} FIG. 6). Alternatively, the NOx purification rate by the NOx occlusion reduction type catalyst 32 is calculated from the NOx emission amount upstream of the catalyst estimated from the operating state of the engine 10 and the NOx amount downstream of the catalyst detected by the NOx / lambda sensor 45. , if the NOx purification rate becomes lower than a predetermined judgment threshold, NOx purge flag F _NP is turned on.

  In the present embodiment, the enrichment by the NOx purge control is performed on the lean side of the excess air ratio from the stoichiometric air-fuel ratio equivalent value (about 1.0) from the time of steady operation (for example, about 1.5) by the air system control. NOx purge lean control for reducing to 3 target excess air ratio (for example, about 1.3) and injection system control to reduce the excess air ratio from the fourth target excess air ratio to the second target excess air ratio on the rich side (for example, about 0) .9) and NOx purge rich control for reducing the pressure to 9). The details of the NOx purge lean control and the NOx purge rich control will be described below.

[NOF purge lean control MAF target value setting]
FIG. 7 is a block diagram showing a setting process of the MAF target value MAF _{NPL_Trgt} at the time of NOx purge lean control. The third target excess air ratio setting map 71 is a map that is referred to based on the engine speed Ne and the accelerator opening Q, and during NOx purge lean control corresponding to the engine speed Ne and the accelerator opening Q. The excess air ratio target value λ _{NPL_Trgt} (third excess air ratio) is set in advance based on experiments or the like.

First, the excess air ratio target value λ _{NPL_Trgt} at the time of NOx purge lean control is read from the third target excess air ratio setting map 71 using the engine speed Ne and the accelerator opening Q as input signals, and is sent to the MAF target value calculation unit 72. Entered. Further, the MAF target value calculation unit 72 calculates the MAF target value MAF _{NPL_Trgt} at the time of NOx purge lean control based on the following formula (3).
MAF _{NPL_Trgt} = λ _{NPL_Trgt} × Q _{fnl_corrd} × Ro _Fuel × AFR _sto / _{Maf_corr} (3)

In Equation (3), Q _{fnl_corrd} represents a learning-corrected fuel injection amount (excluding post-injection) described later, Ro _Fuel represents fuel specific gravity, AFR _sto represents a stoichiometric air-fuel ratio, and _{Maf_corr} represents a MAF correction coefficient described later. Yes.

MAF target value MAF _{NPL_Trgt} calculated by the MAF target value calculation unit 72 is input the NOx purge flag F _SP is turned on (see the time t ₁ in FIG. 6) to the lamp unit 73. The ramp processing unit 73 reads the ramp coefficient from the ramp coefficient maps 73A and _73B using the engine speed Ne and the accelerator opening Q as input signals, and _{sets the} MAF target ramp value MAF _{NPL_Trgt_Ramp} to which the ramp coefficient is added as a valve control unit 74. To enter.

The valve controller 74 throttles the intake throttle valve 16 to the _close side and feeds back the EGR valve 24 to the open side so that the actual MAF value MAF _Act input from the MAF sensor 40 becomes the MAF target ramp value MAF _{NPL_Trgt_Ramp.} Execute control.

As described above, in the present embodiment, the MAF target value MAF _{NPL_Trgt} is set based on the excess air ratio target value λ _{NPL_Trgt} read from the third target excess air ratio setting map 71 and the fuel injection amount of each injector 11. The air system operation is feedback-controlled based on the MAF target value MAF _{NPL_Trgt} . Thus, without providing a lambda sensor upstream of the NOx storage reduction catalyst 32, or even when a lambda sensor is provided upstream of the NOx storage reduction catalyst 32, the sensor value of the lambda sensor is not used. It is possible to effectively reduce the exhaust gas to a desired excess air ratio required for NOx purge lean control.

_Further, by using the fuel injection amount Q _{fnl_corrd} after learning correction as the fuel injection amount of each injector 11, the MAF target value MAF _{NPL_Trgt} can be set by feedforward control, and the aging deterioration and characteristic change of each injector 11 are possible. Etc. can be effectively eliminated.

Further, by adding a ramp coefficient that is set according to the operating state of the engine 10 to the MAF target value MAF _{NPL_Trgt} , it is possible to prevent misfire of the engine 10 due to a sudden change in the intake air amount, deterioration of drivability due to torque fluctuation, and the like. It can be effectively prevented.

[NOx purge rich control fuel injection amount setting]
FIG. 8 is a block diagram showing processing for setting a target injection amount Q _{NPR_Trgt} (injection amount per unit time) for exhaust pipe injection or post injection in NOx purge rich control. The fourth target excess air ratio setting map 75 is a map that is referred to based on the engine speed Ne and the accelerator opening Q, and during NOx purge rich control corresponding to the engine speed Ne and the accelerator opening Q. The air excess rate target value λN _{PR — Trgt} (fourth target air excess rate) is set in advance based on experiments or the like.

First, from the fourth target excess air ratio setting map 75, the excess air ratio target value λ _{NPR_Trgt} at the time of NOx purge rich control is read using the engine speed Ne and the accelerator opening Q as input signals, and an injection amount target value calculation unit 76 is performed. Is input. Further, the injection amount target value calculation unit 76 calculates a target injection amount Q _{NPR_Trgt} at the time of NOx purge rich control based on the following formula (4).
Q _{NPR_Trgt} = MAF _{NPL_Trgt} × _{Maf_corr} / (λ _{NPR_Target} × Ro _Fuel × AFR _sto ) −Q _{fnl_corrd} (4)

In Formula (4), MAF _{NPL_Trgt} is a NOx purge lean MAF target value, and is input from the MAF target value calculation unit 72 described above. Q _{fnlRaw_corrd} is a fuel injection amount (excluding post-injection) before application of learning corrected MAF tracking control described later, Ro _Fuel is fuel specific gravity, AFR _sto is a theoretical air-fuel ratio, and _{Maf_cor} is a MAF correction coefficient described later. Show.

The target injection amount Q _{NPR_Trgt} that is calculated by the injection amount target value computing unit 76, NOx purge flag F _SP When turned on, is sent as the injection instruction signal to the exhaust pipe injector 34 or the injectors 11 (time of FIG. 6 t ₁ ). Transmission of the injection instruction signal is continued until the NOx purge flag F _NP is turned off (time t ₂ in FIG. 6) by the completion judgment of the NOx purge control described later.

As described above, in this embodiment, the target injection amount Q _{NPR_Trgt} is set based on the air excess rate target value λ _{NPR_Trgt} read from the fourth target air excess rate setting map 75 and the fuel injection amount of each injector 11. It has become. Thus, without providing a lambda sensor upstream of the NOx storage reduction catalyst 32, or even when a lambda sensor is provided upstream of the NOx storage reduction catalyst 32, the sensor value of the lambda sensor is not used. It is possible to effectively reduce the exhaust gas to a desired excess air ratio required for NOx purge rich control.

_Further, by using the fuel injection amount Q _{fnl_corrd} after learning correction as the fuel injection amount of each injector 11, the target injection amount Q _{NPR_Trgt} can be set by feedforward control, and the aging deterioration and characteristic change of each injector 11 are possible. Etc. can be effectively eliminated.

[No air system control for NOx purge control]
The ECU 50 feedback-controls the opening degree of the intake throttle valve 16 and the EGR valve 24 based on the sensor value of the MAF sensor 40 in the region where the operating state of the engine 10 is on the low load side. On the other hand, in the region where the operating state of the engine 10 is on the high load side, the ECU 50 feedback-controls the supercharging pressure by the variable displacement supercharger 20 based on the sensor value of the boost pressure sensor 46 (hereinafter, this region is referred to as “high”). Booth pressure FB control area).

In such a booth pressure FB control region, a phenomenon occurs in which the control of the intake throttle valve 16 and the EGR valve 24 interferes with the control of the variable displacement supercharger 20. For this reason, there is a problem that the intake air amount cannot be maintained at the MAF target value _{NPL_Trgt} even if the NOx purge lean control for performing feedback control of the air system based on the MAF target value MAF _{NPL_Trgt} set by the above equation (3) is executed. . As a result, even if the NOx purge rich control for executing the post injection or the exhaust pipe injection is started, the excess air ratio can be reduced to the fourth target excess air ratio (the excess air ratio target value λ _{NPR_Trgt} ) necessary for the NOx purge. There is no possibility.

In order to avoid such a phenomenon, the NOx detachment processing unit 70 of this embodiment prohibits NOx purge lean control for adjusting the opening degree of the intake throttle valve 16 and the EGR valve 24 in the booth pressure FB control region, and exhaust pipe The excess air ratio is reduced to the fourth target excess air ratio (the excess air ratio target value λ _{NPR_Trgt} ) only by injection or post injection. As a result, the NOx purge can be reliably performed even in the booth pressure FB control region. In this case, the MAF target value set based on the operating state of the engine 10 may be applied to the MAF target value MAF _{NPL_Trgt} of the above-described equation (4).

[Determining completion of NOx purge control]
In the NOx purge control, (1) when the NOx purge flag F _NP is turned on, the injection amount of exhaust pipe injection or post injection is accumulated, and when this cumulative injection amount reaches a predetermined upper limit threshold amount, (2) NOx purge control When the elapsed time counted from the start reaches a predetermined upper threshold time, (3) calculation is performed based on a predetermined model formula including the operating state of the engine 10 and the sensor value of the NOx / lambda sensor 45 as input signals. If any of the conditions in the case of the NOx occlusion amount of the NOx occlusion-reduction catalyst 32 has decreased to a predetermined threshold value indicating a successful removal of NOx is satisfied, is terminated by turning off the NOx purge flag F _NP (6 time t ₂ reference).

  As described above, in the present embodiment, the cumulative injection amount and the upper limit of the elapsed time are provided in the end condition of the NOx purge control, so that the fuel consumption amount is reduced when the NOx purge is not successful due to a decrease in the exhaust temperature or the like. It is possible to reliably prevent the excess.

[MAF tracking control]
The MAF follow-up control unit 80 includes (1) a period for switching from a lean state in normal operation to a rich state by SOx purge control or NOx purge control, and (2) lean in normal operation from a rich state by SOx purge control or NOx purge control. During the switching period to the state, control for correcting the fuel injection timing and the fuel injection amount of each injector 11 in accordance with the MAF change (hereinafter, this control is referred to as MAF follow-up control) is executed.

[Injection amount learning correction]
As shown in FIG. 9, the injection amount learning correction unit 90 includes a learning correction coefficient calculation unit 91 and an injection amount correction unit 92. The learning correction coefficient calculation unit 91 and the injection amount correction unit 92 constitute a learning unit and a correction unit of the present invention.

The learning correction coefficient calculation unit 91 is based on the error Δλ between the actual lambda value λ _Act detected by the NOx / lambda sensor 45 during the lean operation of the engine 10 and the estimated lambda value λ _Est and the learning correction coefficient F for the fuel injection amount. Calculate _Corr . When the exhaust is in a lean state, the oxidation reaction of HC does not occur in the oxidation catalyst 31, so that the actual lambda value λ _Act in the exhaust that passes through the oxidation catalyst 31 and is detected by the downstream NOx / lambda sensor 45, and It is considered that the estimated lambda value λ _Est in the exhaust discharged from the engine 10 coincides. Therefore, if an error Δλ occurs between the actual lambda value λ _Act and the estimated lambda value λ _Est , it can be assumed that the difference is between the commanded injection amount and the actual injection amount for each injector 11. Hereinafter, the learning correction coefficient calculation processing by the learning correction coefficient calculation unit 91 using the error Δλ will be described based on the flow of FIG.

  In step S300, based on the engine speed Ne and the accelerator opening Q, it is determined whether or not the engine 10 is in a lean operation state. If it is in the lean operation state, the process proceeds to step S310 to start the calculation of the learning correction coefficient.

In step S310, an error Δλ obtained by subtracting the actual lambda value λ _Act detected by the NOx / lambda sensor 45 from the estimated lambda value λ _Est is multiplied by the learning value gain K ₁ and the correction sensitivity coefficient K ₂ to thereby obtain the learning value F _CorrAdpt is calculated (F _CorrAdpt = (λ _Est −λ _Act ) × K ₁ × K ₂ ). The estimated lambda value λ _Est is estimated and calculated from the operating state of the engine 10 according to the engine speed Ne and the accelerator opening Q. Further, the correction sensitivity coefficient K ₂ is read the actual lambda value lambda _Act detected by the NOx / lambda sensor 45 from the correction sensitivity coefficient map 91A shown in FIG. 9 as an input signal.

In step S320, it is determined whether or not the absolute value | F _CorrAdpt | of the learning value F _CorrAdpt is within the range of the predetermined correction limit value A. If the absolute value | F _CorrAdpt | exceeds the correction limit value A, the control is returned to stop the current learning.

In step S330, it is determined whether the learning prohibition flag _FPro is off. The learning prohibition flag F _Pro corresponds to, for example, transient operation of the engine 10, SOx purge control (F _SP = 1), NOx purge control (F _NP = 1), and the like. This is because, when these conditions are satisfied, the error Δλ increases due to a change in the actual lambda value λ _Act , and accurate learning cannot be performed. Whether or not the engine 10 is in a transient operation state is determined when, for example, the time change amount is larger than a predetermined threshold based on the time change amount of the actual lambda value λ _Act detected by the NOx / lambda sensor 45. What is necessary is just to determine with a transient operation state.

In step S340, the learning value map 91B (see FIG. 9) referred to based on the engine speed Ne and the accelerator opening Q is updated to the learning value F _CorrAdpt calculated in step S310. More specifically, on the learning value map 91B, a plurality of learning areas divided according to the engine speed Ne and the accelerator opening Q are set. These learning regions are preferably set to have a narrower range as the region is used more frequently and to be wider as a region is used less frequently. As a result, learning accuracy is improved in regions where the usage frequency is high, and unlearning can be effectively prevented in regions where the usage frequency is low.

In step S350, the learning correction coefficient F _Corr is calculated by adding “1” to the learning value read from the learning value map 91B using the engine speed Ne and the accelerator opening Q as input signals (F _Corr = 1 + F). _CorrAdpt ). This learning correction coefficient F _Corr is input to the injection amount correction unit 92 shown in FIG.

Injection amount correcting unit 92, by multiplying the pilot injection Q _Pilot, pre-injection Q _Pre, main injection Q _M ain, after injection QAfter, the learning correction coefficient FCorr each basic injection quantity of post injection QPOST, the amount of these fuel injection Perform the correction.

In this way, by correcting the fuel injection amount to each injector 11 with the learning value corresponding to the error Δλ between the estimated lambda value λ _Est and the actual lambda value λ _Act , the aging deterioration, characteristic change, individual difference of each injector 11 is corrected. It is possible to effectively eliminate such variations.

[MAF correction coefficient]
The MAF correction coefficient calculation unit 95 sets the MAF target value MAFSPL_Trgt and target injection amount QSPR_Trgt during SOx purge control, and the MAF correction coefficient used for setting the MAF target value MAF _{NPL_Trgt} and target injection amount Q _{NPR_Trgt} during NOx purge control. _{Maf_corr} is calculated.

In the present embodiment, the fuel injection amount of each injector 11 is corrected based on the error Δλ between the actual lambda value λ _Act detected by the NOx / lambda sensor 45 and the estimated lambda value λ _Est . However, since lambda is the ratio of air to fuel, the factor of error Δλ is not necessarily only the effect of the difference between the commanded injection amount and the actual injection amount for each injector 11. That is, there is a possibility that the error of not only each injector 11 but also the MAF sensor 40 affects the lambda error Δλ.

FIG. 11 is a block diagram showing the setting process of the MAF correction coefficient _{Maf_corr} by the MAF correction coefficient calculation unit 95. The correction coefficient setting map 96 is a map that is referred to based on the engine speed Ne and the accelerator opening Q. The MAF indicating the sensor characteristics of the MAF sensor 40 corresponding to the engine speed Ne and the accelerator opening Q is shown in FIG. The correction coefficient _{Maf_corr} is set in advance based on experiments or the like.

The MAF correction coefficient calculation unit 95 reads the MAF correction coefficient _{Maf_corr} from the correction coefficient setting map 96 using the engine speed Ne and the accelerator opening Q as input signals, and outputs the MAF correction coefficient _{Maf_corr} to the MAF target value calculation unit 62, 72 and the injection amount target value calculation units 66 and 76. Thus, SOx purge control when the MAF target value MAF _{SPL_Trgt} and the target injection amount Q _{SPR_Trgt,} the setting of the MAF target value MAF _{NPL_Trgt} and the target injection amount Q _{NPR_Trgt} during NOx purge control effectively the sensor characteristics of the MAF sensor 40 It becomes possible to reflect.

[Others]
In addition, this invention is not limited to the above-mentioned embodiment, In the range which does not deviate from the meaning of this invention, it can change suitably and can implement.

DESCRIPTION OF SYMBOLS 10 Engine 11 Injector 12 Intake passage 13 Exhaust passage 16 Intake throttle valve 24 EGR valve 31 Oxidation catalyst 32 NOx occlusion reduction type catalyst 33 Filter 34 Exhaust pipe injection device 40 MAF sensor 45 NOx / lambda sensor 50 ECU

Claims (3)

A lambda sensor provided in an exhaust passage of the internal combustion engine for acquiring an actual lambda value of the exhaust;
Based on the difference between the estimated lambda value of the exhaust gas estimated from the operating state of the internal combustion engine and the actual lambda value acquired by the lambda sensor, a correction coefficient for the indicated injection amount output to the injector of the internal combustion engine is learned. Learning means,
Correcting means for correcting the command injection amount to be output to the injector based on the correction coefficient learned by the learning means ,
The learning means calculates the correction coefficient based on a value obtained by multiplying a difference between the estimated lambda value and the actual lambda value by a correction sensitivity coefficient,
The learning means calculates the correction sensitivity coefficient based on the actual lambda value,
The control device for an internal combustion engine, wherein the correction sensitivity coefficient increases as the actual lambda value increases . The control apparatus for an internal combustion engine according to claim 1, further comprising learning prohibiting means for prohibiting learning by the learning means during transient operation of the internal combustion engine. An exhaust purification device is provided in the exhaust passage upstream of the lambda sensor;
The learning means performs learning of the correction coefficient during lean operation of the internal combustion engine,
The control device, the transient operation of the internal combustion engine, or for an internal combustion engine according to claim 1, further comprising a learning prohibition means for prohibiting the learning of the learning means when reproduction processing to recover the purification ability of the exhaust gas purifier Control device.

Images