JP5902727B2 - Exhaust gas purification system for internal combustion engine - Google Patents

Description

  The present invention relates to an exhaust gas purification system for an internal combustion engine. More particularly, the present invention relates to a so-called exhaust fuel injection type exhaust purification system including an exhaust purification catalyst provided in an exhaust pipe and an exhaust fuel injector that injects fuel upstream of the exhaust purification catalyst.

  FIG. 33 is a diagram showing a configuration of an exhaust fuel injection type exhaust gas purification system 100. The exhaust purification system 100 stores an exhaust fuel injector 102 that injects fuel into the exhaust pipe 101, and NOx contained in the exhaust during lean operation. When fuel is injected from the injector 102, this is used as a reducing agent for NOx. An exhaust purification catalyst (hereinafter referred to as LNT (Lean NOx Trap)) 104 and a LAF sensor 105 that detects the air-fuel ratio of exhaust on the downstream side of the LNT 104.

  In recent years, when fuel is intermittently injected from the injector 102 at a period of 5 Hz or more, an intermediate product derived from hydrocarbons is generated in the LNT 104, and NOx can be continuously purified at a high purification rate by this intermediate product. (For example, see Patent Documents 1 and 2).

  FIG. 34 shows the exhaust fuel injection amount (the amount of fuel injected from the injector 102 per unit time by the intermittent injection), the NOx purification rate by the LNT 104, the HC slip amount from the LNT 104, and the intermediate product generation amount in the LNT 104. It is a figure which shows the relationship. As shown in FIG. 34, increasing the exhaust fuel injection amount also increases the NOx purification rate. However, since the production amount of the intermediate product in the LNT 104 is limited, if the exhaust fuel injection amount exceeds the amount where the increase in the production amount starts to slow down, it cannot become an intermediate product that contributes to NOx purification. Fuel is discharged to the downstream side of the LNT 104 as HC. For this reason, in order to suppress the HC slip amount as much as possible while increasing the NOx purification rate as much as possible, the exhaust fuel injection amount is controlled to the amount indicated by G2 at which the HC slip starts or slightly larger than G2. It is preferable.

  In FIG. 34, in order to control the exhaust fuel injection amount in a region slightly larger than G2 or G2, in order to prevent an excessively large HC slip from occurring, a sensor capable of detecting the HC slip is required. None of the existing in-vehicle sensors can detect HC in the exhaust. For this reason, conventionally, the exhaust fuel injection amount cannot be actively controlled to a region slightly larger than G2 or G2, and is suppressed to about G1, which is smaller than G2, so as to reliably prevent the occurrence of excessive HC slip. I had to do it.

  FIG. 35 is a diagram showing the relationship between the oxidation ability of LNT, the NOx purification rate, and the amount of intermediate products generated. The example of FIG. 35 shows a case where only the oxidation ability of LNT is changed while keeping the exhaust fuel injection amount constant. As shown in FIG. 35, the production amount of the intermediate product and the NOx purification rate show substantially the same change. That is, the NOx purification rate in the LNT increases as the amount of intermediate product generated increases.

When the oxidation ability of LNT becomes weaker than the region Ox_op indicated by hatching in FIG. 35, the amount of intermediate products generated in LNT decreases. Also, when the oxidation ability of LNT becomes stronger than the same region Ox_op, the amount of intermediate product produced in LNT is reduced. This is because when the oxidizing ability becomes strong, the supplied fuel is directly oxidized into CO ₂ and H ₂ O in the LNT. For this reason, if only the oxidation ability is changed while keeping the exhaust fuel injection amount constant, the amount of intermediate product produced shows upwardly convex characteristics as shown in FIG.

  Further, the position of the region Ox_op where the production amount of the intermediate product is close to the maximum changes according to the exhaust fuel injection amount. For this reason, in order to maintain a state where the NOx purification rate is high, it is necessary to control the oxidation capacity of the LNT within an optimum region Ox_op determined according to the exhaust fuel injection amount, but the oxidation capacity of the LNT is deteriorated. This is difficult because it varies depending on the degree and the carrier temperature.

  Further, since the region Ox_op where the production amount of the intermediate product is in the vicinity of the maximum is determined by the relationship between the oxidation performance of the LNT and the exhaust fuel injection amount as described above, the exhaust fuel injection amount G2 at which the HC slip starts (described above) 34 in FIG. 34) means that it changes depending on the oxidation performance of LNT. For this reason, the feedback control using the output of the LAF sensor provided on the downstream side of the LNT makes the exhaust fuel injection amount slightly larger than the G2 or G2 at which the HC slip starts, taking into account the oxidation capability of the LNT. It is possible to control.

JP 2013-15117 A JP 2012-62864 A

However, when the fuel is intermittently injected into the exhaust gas from the injector at an injection cycle of about several Hz, there are the following problems.
FIG. 36 is a diagram illustrating a change in the output value of the LAF sensor when intermittent injection is performed. When the fuel is intermittently injected from the injector in the mode shown in the uppermost stage of FIG. 36, the output value of the LAF sensor provided downstream from the injector also changes greatly as shown in the second stage from the top of FIG. End up. For this reason, when constructing feedback control using the output of the LAF sensor, it is necessary to suppress an extra output fluctuation caused by intermittent injection from the output value of the LAF sensor by using a filter.

  The third and fourth stages from the top of FIG. 36 are diagrams showing the outputs of the LAF sensors obtained through the band pass filter and the low pass filter, respectively. Even if the band-pass filter and the low-pass filter are adjusted so that the injection frequency of the intermittent injection is included in the stop band, the fluctuation component due to the intermittent injection cannot be removed as shown in FIG. This is considered to be because the output value of the LAF sensor becomes closer to a sawtooth wave rather than a sine wave corresponding to the injection frequency as a result of intermittent injection of fuel.

  The present invention can remove a fluctuation component caused by intermittent injection from an output value of an air-fuel ratio sensor provided downstream of an injector through which fuel is intermittently injected, and perform feedback control with a stable output. An object is to provide an exhaust gas purification system for an internal combustion engine.

  (1) An exhaust purification system (for example, exhaust purification systems 2 and 2B to be described later) of an internal combustion engine (for example, engine 1 to be described later) of the present invention is provided in an exhaust passage (for example, an exhaust passage 11 to be described later) of the internal combustion engine. An injector (for example, an exhaust fuel injector 452 to be described later) for intermittently injecting fuel upstream of the catalyst (for example, an LNT described later), and a signal corresponding to the air-fuel ratio of the exhaust downstream of the injector An air-fuel ratio sensor (for example, a post-catalyst LAF sensor 52 described later) and sampling means for acquiring an output value of the air-fuel ratio sensor at a predetermined sampling cycle (for example, control cycle ΔTex described later) (for example, ECU 3, 3B described later) ) And a moving average value calculating means for calculating an average value over a predetermined moving average section of the output value (AFact_ds) acquired by the sampling means ( For example, an ECU 3, 3B described later, an injection amount calculating means (for example, ECU 3, 3B described later) for calculating the fuel injection amount of the injector based on the average value (AFact_mav_ds), and an integral multiple of the sampling period Intermittent injection execution means (for example, ECU3, 3B described later) for driving the injector in accordance with the set injection cycle (Tfuel_ex) and the calculated fuel injection amount (Gfuel_ex).

  (2) In this case, the exhaust purification system calculates an injection cycle calculating means (for example, an intermittent injection cycle) obtained by multiplying the sampling cycle (ΔTex) by an injection cycle parameter (Nex) which is a predetermined integer. , ECU3, 3B, which will be described later, and the injection cycle parameter change stepwise according to the state of the vehicle equipped with the engine (for example, catalyst temperature, exhaust volume, catalyst deterioration degree, engine load, etc.) It is preferable to further include an injection cycle parameter setting unit (for example, ECUs 3 and 3B described later) for causing the moving average value calculating unit to calculate an average value using the injection cycle parameter as the number of taps corresponding to the moving average section. .

  (3) In this case, the exhaust purification system uses an estimation unit (for example, ECU3, 3B described later) for estimating the HC oxidation ability of the catalyst, and the injection cycle parameter setting unit uses an estimation result (for example, the estimation unit). It is preferable to change the injection cycle parameter in a step-like manner according to an oxidation capacity parameter (to be described later).

  (1) When fuel is intermittently injected from the injector at a predetermined injection frequency (or injection cycle), the raw output of the air-fuel ratio sensor provided downstream thereof is a rectangular wave rather than a simple sine wave of the injection frequency. It becomes close to a sawtooth wave (see FIG. 36 described above). That is, the raw output of the air-fuel ratio sensor under intermittent injection includes not only the injection frequency component but also various frequency components. In the present invention, the output value of the air-fuel ratio sensor is acquired at a predetermined sampling period, the average value of the acquired value over a predetermined moving average section is calculated, and the fuel injection amount of the injector is calculated using this average value. In addition to this, in the present invention, an integral multiple of the sampling period is set as the injection period of intermittent injection. That is, according to the present invention, by setting the control parameter relating to the moving average filter of the air-fuel ratio sensor and the control parameter relating to intermittent injection of the injector in combination, an extra fluctuation component included in the output of the air-fuel ratio sensor can be removed. . Further, by using the average value from which such fluctuating components are removed, the fuel injection amount can be stably calculated so that the NOx purification rate and the downstream HC emission amount in the catalyst are optimized.

  (2) In the present invention, the sampling cycle multiplied by the injection cycle parameter is used as the intermittent injection cycle, and the injection cycle parameter for designating the length of this injection cycle and the number of taps for calculating the moving average are used. . Thus, by associating the injection cycle and the number of taps with an integer injection cycle parameter, it is possible to remove excess fluctuation components of the air-fuel ratio sensor. In the present invention, the injection cycle parameter is changed in a step shape according to the state of the vehicle. This makes it possible to remove excess fluctuation components of the air-fuel ratio sensor at the same time while varying the injection cycle to a length corresponding to the state of the vehicle.

  (3) In the present invention, the injection cycle parameter is changed stepwise according to the HC oxidation ability of the catalyst. Thereby, according to the change of the HC oxidation ability of the catalyst, the injection cycle of the intermittent injection and the moving average section can be changed simultaneously in steps. Thereby, the injection cycle of intermittent injection and the fuel injection amount can be appropriately controlled according to the change in the HC oxidation ability of the catalyst, and the NOx purification rate and the downstream HC discharge amount in the catalyst can be optimized.

It is a figure which shows the structure of the engine which concerns on 1st Embodiment of this invention, and its exhaust gas purification system. It is a figure which shows the change of the filter value of the LAF sensor obtained by the filtering method of this invention. It is a figure which shows the relationship between the amount of exhaust fuel injection when purifying NOx by the exhaust fuel injection system, the NOx purification rate of LNT, the amount of HC slip from LNT, and the amount of intermediate products generated in LNT. . It is a figure which shows the relationship between the output value (vertical axis) of a post-catalyst LAF sensor, and its true value (horizontal axis). It is a figure for demonstrating the concept of the feedback control using a NOx purification parameter. It is a main flowchart which shows the specific procedure of the cylinder fuel injection control which determines the fuel-injection aspect by the fuel-injection valve of each cylinder. It is a flowchart which shows the specific procedure of stoichi operating condition judgment processing. It is an example of the map (for the ternary purification mode) for updating the stoichiometric mode flag. It is an example of the map (for combined mode) for updating a stoichiometric mode flag. It is a flowchart which shows the specific procedure of an additional injection fuel amount calculation process. It is an example of the map which determines the additional injection fuel amount. It is an example of the map which determines an additional injection ratio. It is a flowchart which shows the specific procedure of the target pre-catalyst air fuel ratio calculation process. It is an example of the map which determines the target pre-catalyst air fuel ratio in exhaust fuel injection purification operation. It is a flowchart which shows the specific procedure of the pre-catalyst air fuel ratio feedback calculation which determines an air fuel ratio correction coefficient. It is a flowchart which shows the specific procedure of the post-catalyst air-fuel ratio feedback calculation which determines a target post-catalyst air-fuel ratio. It is an example of the map which determines a target post-catalyst air fuel ratio under weak rich mode. It is an example of the map which determines a ternary purification target air fuel ratio. It is a flowchart which shows the specific procedure of the weak rich mode completion | finish determination process which updates a reduction process completion flag. It is a time chart which shows the specific example of weak rich mode completion | finish determination processing. It is a figure which shows the main flowchart which shows the specific procedure of the exhaust fuel injection control which determines the injection aspect of the exhaust fuel by an exhaust fuel injector. It is a continuation of the flowchart of FIG. It is an example of the map which determines the reference | standard exhaust fuel injection amount. It is a flowchart which shows the specific procedure of an intermittent injection parameter setting process. It is an example of the map which determines an injection period parameter. It is a flowchart which shows the specific procedure of an adaptive coefficient calculation process (for the time of HC slip feedback control). It is an example of the map for calculating the weight function value for NOx amount (upper stage) and the weight function value for LNT temperature (lower stage). It is a flowchart which shows the specific procedure of an adaptive coefficient calculation process (for HC slip suppression mode). It is a flowchart which shows the specific procedure of the failure determination process of an exhaust fuel injection system. It is a flowchart which shows the specific procedure of intermittent injection control. It is a time chart which shows the specific example of intermittent injection control. It is a figure which shows the structure of the engine which concerns on 2nd Embodiment of this invention, and its exhaust gas purification system. It is a figure which shows the structure of the conventional exhaust fuel injection type exhaust gas purification system. It is a figure which shows the relationship between the amount of exhaust fuel injection, the NOx purification rate, the amount of HC slip, and the production amount of an intermediate product. It is a figure which shows the relationship between the oxidation capability of LNT, the NOx purification rate, and the production amount of an intermediate product. It is a figure which shows the change of the output value of a LAF sensor at the time of performing intermittent injection.

Hereinafter, a first embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a diagram showing a configuration of an internal combustion engine (hereinafter referred to as “engine”) 1 and its exhaust purification system 2 according to the present embodiment. The engine 1 is based on so-called lean combustion in which the combustion air-fuel ratio is leaner than stoichiometric, more specifically, a diesel engine, a lean burn gasoline engine, or the like.

  The exhaust purification system 2 includes a lean NOx catalyst (hereinafter referred to as “LNT”) 41 and an exhaust purification filter 43 provided in the exhaust passage 11 of the engine 1, an exhaust fuel injection device 45 that injects fuel into the exhaust passage 11, And an electronic control unit (hereinafter referred to as “ECU”) 3 that controls the engine 1 and the exhaust fuel injection device 45.

  The engine 1 is provided with a fuel injection valve 13 for injecting fuel into each cylinder. These fuel injection valves 13 are connected to the ECU 3 via a driving device (not shown). The ECU 3 determines the fuel injection amount and fuel injection timing from the fuel injection valve 13 by in-cylinder fuel injection control which will be described later with reference to FIGS. 6 to 20, and the drive device realizes the determined fuel injection mode. The fuel injection valve 13 is driven as described above.

  The LNT 41 has at least three functions of an oxidation function, a DeNOx function, and a ternary purification function. Here, the oxidation function refers to a function of oxidizing HC and CO contained in exhaust gas during a lean operation in which the combustion air-fuel ratio is leaner than stoichiometric. The DeNOx function stores NOx contained in exhaust gas when the air-fuel ratio of the exhaust gas is leaner than stoichiometric, and enters the exhaust gas by exhaust fuel injection from the exhaust fuel injection device 45, post injection from the fuel injection valve 13, or the like. When fuel is supplied, it refers to the function of reducing NOx using this as a reducing agent. The three-way purification function refers to a function for purifying HC, CO, and NOx contained in exhaust gas together during a stoichiometric operation in which the combustion air-fuel ratio is stoichiometric.

  As described above, the NOx in the exhaust gas can be purified using the DeNOx function of the LNT 41 during the lean operation, and can be purified using the three-way purification function of the LNT 41 during the stoichiometric operation. Here, comparing the case where NOx is purified using the DeNOx function and the case where NOx is purified using the three-way purification function, the case where the three-way purification function is used is more efficient for NOx. It can be purified. Therefore, for example, when the amount of NOx discharged from the engine 1 increases during high-load operation, or when the DeNOx function cannot be fully exhibited because the LNT 41 has not reached its activity, the three-way purification function is used. In order to purify the exhaust gas, the lean operation is switched to the stoichiometric operation (see, for example, FIG. 7 described later).

  When the exhaust gas passes through fine holes in the filter wall, the exhaust gas purification filter 43 collects PM mainly composed of carbon in the exhaust gas by depositing it on the surface of the filter wall and the holes in the filter wall. As a constituent material of the filter wall, for example, a porous body made of aluminum titanate or cordierite is used. Further, in order to prevent HC slipped from the LNT 41 from being discharged out of the exhaust purification system 2, an HC oxidation catalyst that oxidizes HC in the exhaust is provided on the filter wall of the exhaust purification filter 43. Hereinafter, the abbreviation “CSF” is used for the exhaust purification filter 43 on which such a catalyst is supported.

  By the way, the exhaust passage 11 is divided into a section located in an engine room (not shown) (a section directly under the engine) and a section (under the floor section) located under the floor of a vehicle (not shown). The section immediately below is closer to the engine 1 than the section below the floor. Accordingly, the average temperature in the section immediately below is higher than that in the section under the floor, and the temperature rise after the start of the engine 1 is quicker. Therefore, the LNT 41 is provided in the section immediately below the exhaust passage 11 in order to exhibit the above-described oxidation function, three-way purification function, and DeNOx function as advantageously as possible.

  When PM is collected to the limit of the collection ability of CSF43, pressure loss becomes large. For this reason, the forced regeneration process which burns and removes the collected PM and regenerates the filter function of the CSF 43 is appropriately executed. In this forced regeneration process, for example, post injection or fuel injection from the exhaust fuel injection device 45 is executed, and the exhaust gas flowing into the CSF 43 is heated to burn and remove the accumulated PM in a short time.

  The exhaust fuel injection device 45 includes a fuel tank 451 in which fuel is stored, an exhaust fuel injector 452 provided on the upstream side of the LNT 11 in the exhaust passage 11, and pressurization for pressure-feeding the fuel in the fuel tank 451 to the injector 452. A pump 453. The exhaust fuel injector 452 is electromagnetically connected to the ECU 3 via a drive device (not shown). When the ECU 3 purifies exhaust using the DeNOx function of the LNT 41, the fuel injection amount per unit time from the exhaust fuel injector 452, the injection cycle, etc. by exhaust fuel injection control shown in FIGS. The exhaust fuel injection mode is determined, and the drive device drives the exhaust fuel injector 452 so that the determined exhaust fuel injection mode is realized.

By the way, in recent years, when fuel is injected from the exhaust fuel injector 452 and NOx is reduced using the DeNOx function of the LNT 41, the exhaust fuel injection amount from the exhaust fuel injector 452 is 5 Hz as shown in FIG. When the hydrocarbon concentration of the exhaust gas flowing into the LNT 41 is oscillated within the predetermined range at the above cycle, an intermediate product derived from hydrocarbons is generated on the LNT, and the exhaust gas is exhausted at a high purification rate by this intermediate product. It is known that NOx therein can be continuously purified. However, when fuel is injected in the above-described manner with the carrier temperature of LNT41 being about 350 ° C. or lower, unnecessary components (for example, N ₂ O) that do not contribute to NOx purification are generated, and downstream of LNT41 May be discharged to the side. Therefore, in the exhaust fuel injection control, the fuel is intermittently injected as described above only when the carrier temperature of the LNT 41 is about 350 ° C. or more and not more than the upper limit temperature of about 630 to 700 ° C.

  Hereinafter, a combination of the exhaust fuel injection device 45 and the LNT 41 is collectively referred to as an exhaust fuel injection system. Further, by intermittently injecting fuel from the exhaust fuel injector 452 as described above, it is possible to continuously purify NOx in the exhaust gas flowing into the LNT 41 while generating an intermediate product in the LNT 41. It is called purification operation.

  The ECU 3 includes a pre-catalyst LAF sensor 51, a post-catalyst LAF sensor 52, a pre-catalyst temperature sensor 53, a post-catalyst temperature sensor 54, a crank angle position as sensors for detecting the state in the exhaust passage 11 and the state of the engine 1. A sensor 55, an accelerator opening sensor 56, an air flow sensor 57, an atmospheric temperature sensor 58, and the like are connected.

  The pre-catalyst LAF sensor 51 is provided upstream of the LNT 41 and the exhaust fuel injector 452 in the exhaust passage 11. The pre-catalyst LAF sensor 51 detects the air-fuel ratio (ratio of fuel component to oxygen in the exhaust gas) upstream of the LNT 41 and before the fuel is injected from the exhaust fuel injector 452. A proportional signal is transmitted to the ECU 3. The post-catalyst LAF sensor 52 is provided between the LNT 41 and the CSF 43 in the exhaust passage 11. The post-catalyst LAF sensor 52 detects the air-fuel ratio of the exhaust gas between the LNT 41 and the CSF 43 and transmits a signal substantially proportional to the detected value to the ECU 3. The signals output from these LAF sensors 51 and 52 have a linear characteristic from a rich area to a lean area.

  The detection elements 51a and 52a of these LAF sensors 51 and 52 incorporate a heater (not shown) for heating the elements 51a and 52a. The ECU 3 independently sets the target temperature (Tcmd_laf_up) of the pre-catalyst LAF sensor 51 and the target temperature (Tcmd_laf_ds) of the post-catalyst LAF sensor 52 by exhaust gas fuel injection control (see FIG. 21), which will be described later. Each heater current value is controlled so that the set target temperature is realized by the control process. Further, an oxidizing material such as platinum having an HC oxidation function in exhaust gas is provided on the surface layer of the detection elements 51a and 52a.

  The pre-catalyst temperature sensor 53 is provided upstream of the LNT 41 in the exhaust passage 11, and the post-catalyst temperature sensor 54 is provided downstream of the LNT 41 in the exhaust passage 11. These temperature sensors 53 and 54 detect the temperatures of the exhaust gas flowing into the LNT 41 and the exhaust gas flowing out of the LNT 41, respectively, and transmit a signal substantially proportional to the detected value to the ECU 3. The estimated value of the carrier temperature of the LNT 41 provided between the sensors 54 and 54 is calculated by the ECU 3 as a weighted average value of the outputs of the temperature sensors 53 and 54, for example.

  The crank angle position sensor 55 detects the rotation angle of the crankshaft of the engine 1, generates a pulse for each predetermined crank angle, and transmits the pulse signal to the ECU 3. The rotational speed of the engine 1 is calculated by the ECU 3 based on this pulse signal. The accelerator opening sensor 56 detects a depression amount (hereinafter referred to as “accelerator opening”) of an accelerator pedal (not shown) of the vehicle, and transmits a detection signal substantially proportional to the detected value to the ECU 3. The ECU 3 calculates the driver request torque according to the accelerator opening and the engine speed. The air flow meter 57 is provided in the intake passage 12. The air flow meter 57 detects the amount of intake air flowing through the intake passage 12 and transmits a signal substantially proportional to the detected value to the ECU 3. The ECU 3 calculates the exhaust volume according to the intake air amount.

Next, problems unique to the exhaust fuel injection system that performs intermittent injection as described above and a method for solving this problem will be described with reference to FIGS. 36 and 2. FIG.
FIG. 36 is a diagram showing fluctuations in the output value of the post-catalyst LAF sensor when intermittent injection is performed with an injection period of 5 Hz or more by the exhaust fuel injector. 36 shows the exhaust fuel injection amount [mg / sec], which is the amount of fuel injected from the exhaust fuel injector per unit time, and the second and lower stages from the top indicate the air-fuel ratio [A / F] represents the output value of the post-catalyst LAF sensor. The second row from the top shows the output without the filter (AFact_ds), and the third row and the bottom row from the top show the output through the filter described later (AFact_flt_ds). In FIG. 36, the output value of the post-catalyst LAF sensor is indicated by a solid line, and the air-fuel ratio of the air-fuel mixture is indicated by a broken line. In the example shown in FIG. 36, the injection time (the amount of fuel injected in one injection cycle) is increased at time t1 while the air-fuel ratio of the air-fuel mixture is kept constant, and the injection cycle is shortened at times t2 and t3. Shows the case.

  When fuel is intermittently injected at a frequency as low as about 5 Hz so that an intermediate product that contributes to NOx purification is generated in the LNT, a fuel rich portion and a light portion are present in the exhaust downstream of the exhaust fuel injector. And appear alternately. The concentration of the fuel resulting from the intermittent injection also appears in the exhaust gas after passing through the LNT, and as a result, it is detected by the post-catalyst LAF sensor as an air-fuel ratio fluctuation as shown in the second stage from the top in FIG. The In order to continuously reduce NOx in the exhaust gas while generating an intermediate product in the LNT, it is necessary to perform intermittent injection with such a large air-fuel ratio fluctuation.

  However, if the fluctuation component resulting from intermittent injection remains in the output value of the LAF sensor, the HC feedback control described later cannot be performed accurately based on the output value of the LAF sensor, resulting in sufficient efficiency. NOx cannot be purified. Therefore, when performing this HC feedback control, it is necessary to remove the air-fuel ratio fluctuation caused by the intermittent injection from the output of the LAF sensor and obtain a stable output as shown by a one-dot chain line in FIG. 36 shows the air-fuel ratio of the exhaust gas downstream of the LNT when it is assumed that the fuel injected intermittently is sufficiently mixed in the exhaust pipe and on the LNT. Corresponds to the fuel ratio.

As means for suppressing such fluctuations in the output of the LAF sensor, it is conceivable to use a band-pass filter that blocks components near the injection frequency (Ffuel_ex) of intermittent injection or a low-pass filter that blocks components above the injection frequency. . These filters can be realized by performing the calculation shown in the following formula (1) on the output AFact_ds of the LAF sensor. In the following equation (1), AFact_flt_ds is a filter value, a1, a2, b1, and b2 are filter coefficients adjusted to obtain desired filter characteristics, and m is a control time of the exhaust fuel injection system. The control cycle of the exhaust fuel injection system, that is, the sampling cycle ΔTex of the output value of the LAF sensor by the ECU is, for example, 10 to 50 [msec], which is shorter than the injection cycle.

  The third and bottom stages from the top of FIG. 36 show the filter values obtained by the above equation (1). As shown in FIG. 36, even if a band-pass filter is used, periodic fluctuation components cannot be removed from the output of the LAF sensor. This is because the output AFact_ds of the LAF sensor is a sawtooth wave including various frequency components rather than a simple sine wave of the injection frequency, as shown in the second stage from the top in FIG. it is conceivable that.

  Further, when the low pass filter is used, higher frequency components can be removed than when the band pass filter is used. However, in order to use this filter value for feedback control, it is necessary to prevent a large phase delay, but considering this point, the cutoff frequency cannot be excessively lowered, and as a result, as shown in FIG. In addition, the fluctuation component of the injection frequency cannot be sufficiently removed.

Next, the filtering method of the present invention that solves such problems will be described.
First, in the filtering method of the present invention, two independent parameters, sampling period ΔTex and exhaust fuel injection period Tfuel_ex (m) or exhaust fuel injection frequency Ffuel_ex (m) (= 1 / Tfuel_ex (m)) The injection period parameter Nex (m), which is a positive integer greater than 1, is related. In other words, in the exhaust fuel injection system, the cycle ΔTex for obtaining the output value of the post-catalyst LAF sensor multiplied by the integer Nex (m) is set as the injection cycle Tfuel_ex (m) for intermittent injection.

  For example, when the control cycle ΔTex is 10 [msec] and the injection cycle parameter Nex is 20, the injection cycle Tfuel_ex is 200 [msec] and the injection frequency Ffuel_ex is 5 [Hz]. Further, when the control cycle ΔTex is 10 [msec] and the injection cycle parameter Nex is 18, the injection cycle Tfuel_ex is 180 [msec] and the injection frequency Ffuel_ex is 5.555.

Secondly, in the filtering method of the present invention, the output value AFact_ds of the LAF sensor is obtained by a moving average filter using the injection cycle parameter Nex (m) introduced to set the exhaust fuel injection cycle Tfuel_ex (m) as the number of taps. The filter value AFact_mav_ds (m) of (m) is calculated (see the following formula (3)).

  Thirdly, in the filtering method of the present invention, the injection cycle Tfuel_ex is also changed stepwise by changing the injection cycle parameter Nex, which is a positive integer, stepwise. Thus, the injection cycle parameter Nex and the injection cycle Tfuel_ex are changed so that the above formula (2) always holds.

  FIG. 2 is a diagram showing changes in the filter value AFact_mav_ds of the LAF sensor obtained by the filtering method of the present invention. The exhaust fuel injection amount was changed in the same manner as in the example shown in FIG. As shown in FIG. 2, according to the filtering method of the present invention, the fluctuation component caused by intermittent injection is completely removed, and substantially coincides with the ideal exhaust air / fuel ratio indicated by the broken line. In particular, the ideal exhaust air-fuel ratio changes at times t1 and t2, and the injection frequency changes at times t2 and t3, but the filter value AFact_mav_ds follows these changes without much delay. A specific example of feedback control using the output of the LAF sensor obtained by the filtering method as described above will be described later with reference to FIGS.

Next, the concept of the exhaust fuel injection control of the present invention will be described with reference to FIGS.
FIG. 3 shows the relationship between the exhaust fuel injection amount when purifying NOx by the exhaust fuel injection purification operation, the NOx purification rate of LNT, the HC slip amount from LNT, and the amount of intermediate product produced in LNT FIG.

  As described with reference to FIG. 34, conventionally, since there is no means for detecting or estimating the HC slip amount downstream of the LNT, the exhaust fuel injection amount is determined from the amount G2 at which HC slip starts to occur or from this G2. Even a large amount could not be actively controlled. In the exhaust fuel injection control of the present invention, a target value slightly larger than 0 is set with respect to the HC slip amount downstream of the LNT (see the broken line in FIG. 3), and later referring to FIGS. The amount of HC slip downstream of the LNT is estimated by the method described below, and the exhaust fuel injection amount is controlled so that the estimated amount of HC slip is maintained at the target value. In other words, in the exhaust fuel injection control of the present invention, the exhaust fuel injection amount is controlled to the amount G4 that is larger than the amount G2 and corresponds to the target value of the HC slip amount.

  Here, the effect of maintaining the HC slip amount at a target value slightly larger than 0 will be described. As described with reference to FIG. 35, the injection amounts G2 and G4 at which HC slip starts to occur vary according to the oxidation performance of the LNT at that time. However, when the HC slips slightly from the LNT, the amount of intermediate products generated in the LNT is almost maximized, and the NOx purification rate of the LNT at that time is almost maximized, which changes the oxidation performance of the LNT. It is thought that it will not change. Therefore, by controlling the exhaust fuel injection amount so that the HC slip amount is maintained at the target value, the optimal amount of the exhaust fuel injection amount is calculated in accordance with the oxidation capacity of the LNT that changes depending on the temperature and the degree of deterioration of the LNT. Therefore, the state in which the NOx purification rate is always maximized can be maintained.

Next, means for estimating the HC slip amount in the present invention will be described.
FIG. 4 is a diagram showing the relationship between the output value (vertical axis) of the post-catalyst LAF sensor and its true value (horizontal axis), that is, the actual air-fuel ratio of the exhaust gas at the detection location of the post-catalyst LAF sensor. FIG. 4 shows the change in the output value of the post-catalyst LAF sensor when the HC concentration in the exhaust gas changes with the air-fuel ratio of the exhaust gas kept constant.

  The exhaust gas contains HC as a fuel component left unburned during the combustion of the air-fuel mixture. The LAF sensor oxidizes HC around the detection element and outputs a signal substantially proportional to the amount of remaining oxygen as the air-fuel ratio of the exhaust. In general LAF sensors, in order to promote HC oxidation, an oxidizing material is provided on the surface layer of the detection element, and the temperature of the detection element is raised by a heater. Therefore, as shown on the left side of FIG. 4, when the oxidation performance of the LAF sensor is sufficiently high, the output value hardly changes even if the HC concentration in the exhaust gas changes.

  On the other hand, as shown on the right side of FIG. 4, when the oxidation performance of the LAF sensor is low, HC that has not been oxidized exists around the detection element. Shift in the direction. This is because the more HC that could not be oxidized by the LAF sensor, the more oxygen there is. Further, the magnitude of the offset of the output value of the LAF sensor with respect to the HC concentration in the exhaust gas increases as the oxidation performance of the LAF sensor decreases. As described above, the phenomenon in which the output value of the LAF sensor is shifted due to the presence of HC in the exhaust gas is described in, for example, Japanese Unexamined Patent Application Publication No. 2007-40130 and Japanese Unexamined Patent Application Publication No. 2011-58440. It is well known.

  Thus, since the difference between the output value of the LAF sensor provided downstream of the LNT and its true value has a correlation with the HC concentration in the exhaust, it corresponds to the true value of the air-fuel ratio downstream of the LNT. If the value to be obtained is obtained, the HC concentration of LNT (that is, the HC slip amount) can be estimated. In addition, when estimating the HC slip amount using the offset phenomenon of the output value of the LAF sensor, the means for reducing the oxidation performance of the downstream LAF sensor and the air-fuel ratio of the exhaust downstream of the LNT There are several possible modes for estimating the value corresponding to the true value.

Specific examples of means for reducing the oxidation performance of the LAF sensor include the following TYPE-A and B.
In TYPE-A, a sensor heater is used to control the detection element of the downstream LAF sensor to a temperature lower than the temperature of the detection element of the upstream LAF sensor.
In TYPE-B, the amount per unit area of the oxide material of the element surface layer of the detection element of the downstream LAF sensor is smaller than the amount per unit area of the oxide material of the element surface layer of the detection element of the upstream LAF sensor. To do. Alternatively, the oxidizing material is supported only on the detection element of the upstream LAF sensor and is not supported on the detection element of the downstream LAF sensor.

  Hereinafter, a case where the oxidation performance of the downstream LAF sensor is intentionally lowered by TYPE-A will be described, but the present invention is not limited to this. The oxidation performance may be lowered by TYPE-B or a combination of TYPE-A and B.

  Further, as means for estimating the air-fuel ratio of the exhaust gas downstream of the LNT, the following two types TYPE 1 and 2 can be roughly classified.

  In TYPE1, the exhaust air-fuel ratio on the downstream side of the LNT is estimated using the output value AFact_up of the pre-catalyst LAF sensor provided on the upstream side of the LNT and various calculated values. More specifically, the fuel injected into the cylinder between the discretized control times m-1 to m is Gfuel_tm (m), and the output value of the pre-catalyst LAF sensor at time m is AFact_up (m). Then, the new air amount Gair_ex (m) based on the output of the pre-catalyst LAF sensor is represented by the following equation (4-1). The time dc_i in the equation (4-1) corresponds to the time until the air-fuel mixture burned in the cylinder reaches the exhaust fuel injector. The cylinder-injector arrival time dc_i may be a predetermined fixed value, or a value variably set according to the exhaust volume, the engine load parameter, the engine speed, and the like. Further, when the LAF sensor-based fresh air amount Gair_ex and the exhaust fuel injection amount Gfuel_ex are used, theoretically, the exhaust air / fuel ratio AF_exh_id_up (m) upstream of the LNT at the time m (near the exhaust fuel injector) is It is represented by the following formula (4-2). Here, the fuel amount Gfuel_tm and the exhaust fuel injection amount Gfuel_ex are calculated using calculation values in the in-cylinder fuel injection control and the exhaust fuel injection control described later.

Note that the calculated exhaust air-fuel ratio calculated from the supplied fuel amount and the fresh air amount as in equation (4-2) is generically referred to as an ideal exhaust air-fuel ratio hereinafter. Except for errors such as fuel adhering to the LNT and oxygen storage, the ideal exhaust air / fuel ratio AF_exh_id_up on the upstream side of the LNT and the ideal exhaust air / fuel ratio AF_exh_id_ds on the downstream side are equal. Accordingly, assuming that the time until the exhaust reaches the downstream side of the LNT from the exhaust fuel injector is dLNT, the ideal exhaust air-fuel ratio AF_exh_id_ds (m) on the downstream side of the LNT at time m is expressed by the following equation (4-3). Is done. Note that the time dLNT in the equation (4-3) may be a fixed value or a variable set value, similar to the time dc_i. In TYPE1, the ideal exhaust air-fuel ratio AF_exh_id_ds calculated by the equations (4-1) to (4-3) is used as an estimated value of the exhaust air-fuel ratio downstream of the LNT, that is, the post-catalyst LAF provided downstream of the LNT. This is used as a true value for the sensor output value AFact_ds (or its filter value AFact_mav_ds).

In TYPE2, unlike the TYPE1, the air-fuel ratio of the exhaust gas downstream of the LNT is estimated based on the output values of various sensors and various calculation values without using the output value AFact_up of the pre-catalyst LAF sensor. More specifically, in TYPE2, the amount of fuel injected into the cylinder from time m-1 to m is Gfuel_tm (m), and is injected from the exhaust fuel injector from time m-1 to m. Let Gfuel_ex (m) be the amount of fuel consumed, and Gair_cyl_hat_tm (m) be the estimated value of the amount of fresh air drawn into the cylinder from time m-1 to m at time m according to the following equation (5-1). The ideal exhaust air / fuel ratio AF_ex_id_up (m) upstream of the LNT is calculated. Here, the estimated value Gair_cyl_hat_tm of the fresh air amount is calculated based on the output of the air flow meter, for example. Accordingly, assuming that the time required for the exhaust to reach the downstream side of the LNT from the exhaust fuel injector is dLNT, the ideal exhaust air-fuel ratio AF_exh_id_ds (m) on the downstream side of the LNT at time m is expressed by the following equation (5-2). Is done. In TYPE2, the ideal exhaust air / fuel ratio AF_exh_id_ds calculated by the equations (5-1) to (5-2) is used as an estimated value of the exhaust air / fuel ratio downstream of the LNT.

  In TYPE1, the new air amount is calculated using a pre-catalyst LAF sensor provided in the exhaust passage, whereas in TYPE2, the new air amount is calculated using an air flow meter provided in the intake passage. . For this reason, when TYPE1 and TYPE2 are compared, it can be said that TYPE1 has higher estimation accuracy to the extent that errors of various devices upstream from the pre-catalyst LAF sensor are removed. In the following, a case where the downstream air-fuel ratio is estimated by TYPE 1 using the pre-catalyst LAF sensor will be described, but the present invention is not limited to this. Even when the pre-catalyst LAF sensor is not provided, the downstream air-fuel ratio can be estimated by the TYPE2.

Returning to FIG. 4, in the present invention, the output value AFact_ds is true as shown on the right side of FIG. 4 by intentionally lowering the oxidation performance of the post-catalyst LAF sensor by any one of the above TYPEs 1 and 2. Deviation according to the HC concentration is caused from the value. Furthermore, in the present invention, the ideal exhaust air-fuel ratio AF_exh_id_ds of the exhaust on the downstream side of the LNT is calculated by any one of the above TYPEs 1 and 2, and this is used as a true value, which corresponds to the HC slip amount on the downstream side of the LNT. The NOx purification parameter P_LNT of the LNT to be calculated is calculated. More specifically, the purification parameter P_LNT is an ideal exhaust air space calculated from the output value AFact_ds of the post-catalyst LAF sensor whose output is offset according to the HC concentration by the equation (4-3) or (5-2). It is defined as an air-fuel ratio difference obtained by subtracting the fuel ratio AF_exh_id_ds (see the following formula (6)).

Since the actual output value AFact_ds of the post-catalyst LAF sensor has a saw-like shape due to the influence of intermittent injection, the purification parameter P_LNT is expressed by the following equation (7) obtained using the filtering method described with reference to FIG. It is preferable to use those defined by Here, since the pre-catalyst LAF sensor is provided on the upstream side of the exhaust fuel injector, the filtering method of the present invention is not necessarily applied to the exhaust air-fuel ratio AF_exh_id_up (m) upstream of the LNT, as shown in the following formula (7). There is no need to apply. However, in order to compensate for the phase lag caused by providing the moving average filter in the output of the post-catalyst LAF sensor, the exhaust air / fuel ratio AF_exh_id_up (m) is equivalent as shown in the following equation (7-3). It is preferable to provide a moving average filter.

FIG. 5 is a diagram for explaining the concept of feedback control of the present invention using the NOx purification parameter P_LNT.
As shown in FIG. 5, the NOx purification parameter P_LNT has a characteristic that is substantially proportional to the HC concentration (that is, the HC slip amount) on the downstream side of the LNT. Therefore, controlling the NOx purification parameter P_LNT to a predetermined target value P_LNT_cmd that is not 0 is equivalent to maintaining the HC slip amount that cannot be directly detected at the virtual target value HC_SLP_CMD. In the present invention, by controlling the NOx purification parameter P_LNT to the target value P_LNT_cmd, the NOx purification rate of the LNT is suppressed while suppressing the HC slip amount to an amount that does not exceed the amount that can be processed by the CSF downstream of the LNT. A substantially maximized state (that is, a state in which the amount of intermediate products generated is substantially maximized) is realized. In the following, the feedback control using the parameter P_LNT accompanied by a slight but aggressive HC slip to the downstream side of the LNT is referred to as HC slip feedback control.

Next, specific procedures for realizing in-cylinder fuel injection control and exhaust fuel injection control will be described with reference to FIGS.
FIG. 6 is a main flowchart showing a specific procedure of in-cylinder fuel injection control for determining the fuel injection mode by the fuel injection valve of each cylinder. The processing shown in FIG. 6 is executed in the ECU in synchronism with the TDC timing of each cylinder for each combustion cycle. In the following description, values that are updated or sampled in TDC synchronization in the ECU will be denoted by parenthesized symbols “k”.

  Before describing specific procedures with reference to FIG. 6 and the like, three main operation modes defined in in-cylinder fuel injection control and exhaust fuel injection control will be described. The operation mode is divided into a lean operation mode, a stoichiometric operation mode, and an exhaust fuel injection purification operation mode. In the following, in order to clearly indicate which of these three operation modes is being executed or which operation mode is being requested, the exhaust fuel injection purification operation mode flag F_ExINJ_mode, and the stoichiometric mode are described. Two types of flags, the flag F_Stoic_mode, are defined. The flag F_ExINJ_mode is updated by the exhaust fuel injection control shown in FIGS. 21 to 22 described later, and the flag F_Stoic_mode is updated by the process shown in FIG. 7 described later.

  The lean operation mode is an operation mode in which the air-fuel ratio of the air-fuel mixture is made leaner than stoichiometric. Note that when the lean operation mode is requested or when the lean operation mode is being executed, the two flags F_ExINJ_mode and F_Stoic_mode are both set to “0”. During the lean operation mode, the post-injection and fuel injection from the exhaust fuel injectors are periodically performed at a predetermined cycle, and the exhaust on the LNT is temporarily brought into a reducing atmosphere, so that the oxygen adsorbed or stored in the LNT NOx is reduced.

  The stoichiometric operation mode is an operation mode in which exhaust control is performed using the three-way purification function of the LNT by executing feedback control using the outputs of the pre-catalyst LAF sensor and the post-catalyst LAF sensor. Note that the flag F_Stoic_mode is set to “1” when the stoichiometric operation mode is being executed or when execution of the stoichiometric operation mode is requested.

  In the exhaust fuel injection purification operation mode, the exhaust fuel injection amount is determined by the HC slip feedback control described with reference to FIGS. 3 to 5, and the determined amount of fuel is intermittently supplied from the exhaust fuel injector as described above. This operation mode continuously purifies NOx in the exhaust gas flowing into the LNT while maximizing the amount of intermediate products generated in the LNT. Note that the flag F_ExINJ_mode is set to “1” when the exhaust fuel injection purification operation mode is requested or when the exhaust fuel injection purification operation mode is executed.

  In S1, the basic fuel injection amount Gfuel_bs (k) is determined by searching a predetermined map (not shown) according to the operating state of the engine, and the process proceeds to S2. The basic fuel injection amount corresponds to, for example, an in-cylinder fuel injection amount during lean operation (see S13 described later). As will be described in detail later, during the stoichiometric operation and the exhaust fuel injection purification operation, the basic fuel injection amount is calculated based on the outputs of the pre-catalyst LAF sensor and the post-catalyst LAF sensor, and the air-fuel ratio correction coefficient KAF (k). Is multiplied (see S10 described later). Moreover, it shows the operating state of the engine, and examples of the input parameters used for determining the basic fuel injection amount include a driver request torque and an engine speed.

  In S2, it is determined whether or not the devices related to the in-cylinder fuel injection control and the exhaust fuel injection control are normal. The devices related to the determination in S2 are, for example, an intake throttle and an EGR valve (not shown), a pre-catalyst LAF sensor, a post-catalyst LAF sensor, a temperature sensor, and the like that are necessary for performing stoichiometric operation. If the determination in S2 is YES (when the device is normal), the process proceeds to S3. If NO (the device is not normal), the process proceeds to S13, and the lean operation is performed regardless of the state of the two flags. Run.

  In S3, it is determined whether or not the LNT is in an active state. More specifically, in S3, an estimated value of the carrier temperature of LNT is calculated, and when the estimated value is equal to or higher than a predetermined active temperature (for example, 200 ° C.), it is determined that the active state is established. In the case, it is determined that it is not in an active state. If the determination in S3 is YES, the process proceeds to S5, and if NO, the process proceeds to S13 and the lean operation is executed.

  In S5, a stoichiometric operation condition determination process for determining whether or not the stoichiometric operation can be executed is executed, and the process proceeds to S6. In this stoichiometric operation condition determination process, it is determined whether or not the engine is in a state suitable for performing the stoichiometric operation according to the engine operating state, the state of the LNT in the exhaust passage, and the like (see FIG. 7 described later). As a result of this processing, when it is determined that the state is suitable for performing stoichiometric operation, the stoichiometric mode flag F_Stoic_mode (k) is set to “1”, otherwise the flag F_Stoic_mode (k) Is set to “0”.

  In S6, it is determined whether any of the two flags F_Stoic_mode (k) and F_ExINJ_mode (k) is “1”. If the determination in S6 is YES, the process moves to S7, and if NO, the process moves to S13 and the lean operation is executed.

  In S7, an additional injection fuel amount calculation process described later is executed, and the process proceeds to S8. Here, the additional injection is a general term for after injection and exhaust fuel injection. As will be described in detail later with reference to FIG. 10, in this additional injection fuel amount calculation processing, the after injection amount Gfuel_aft (k) and the exhaust fuel injection amount Gfuel_ex_add (k) during the stoichiometric operation or the exhaust fuel injection purification operation To decide.

  In S8, a fuel amount Gfuel_pi (k) (hereinafter referred to as “pilot injection amount”) supplied by pilot injection is calculated, and the process proceeds to S9. The pilot injection amount Gfuel_pi (k) increases in proportion to the engine load such as the engine speed and load parameters (for example, BMEP. In addition, required torque, fuel injection amount, engine torque estimated value, exhaust volume, etc. Etc.) are input by a known method such as map search.

  In S9, a target pre-catalyst air-fuel ratio calculation which will be described later is executed, and the process proceeds to S10. In this target pre-catalyst air-fuel ratio calculation, the target value AFcmd (k) and the provisional value Gfuel_cyl (k) of the in-cylinder fuel injection amount with respect to the output value AFcmd_up (k) of the pre-catalyst LAF sensor are determined (see FIG. 13 described later). . Here, “in-cylinder fuel injection amount” is the total amount of fuel used for combustion in the cylinder during one combustion cycle, and all the fuels injected by pilot injection, main injection, and after injection are combined. It corresponds to a thing. That is, the in-cylinder fuel injection amount does not include the fuel amount injected from the exhaust fuel injector.

  In S10, a pre-catalyst air-fuel ratio feedback calculation described later is executed, and the process proceeds to S11. In the pre-catalyst air / fuel ratio feedback calculation, an air / fuel ratio correction coefficient KAF (k) for controlling the output value AFcmd_up (k) of the pre-catalyst LAF sensor to the target value AFcmd (k) calculated in S7 is calculated.

In S11, the final in-cylinder fuel injection amount Gfuel (k) is determined by multiplying the air-fuel ratio correction coefficient KAF (k) by the provisional value Gfuel_cyl (k) of the in-cylinder fuel injection amount calculated in S10. (Refer to the following formula (8)), the process proceeds to S12.

In S12, the main fuel injection is supplied by subtracting the after-injection amount Gfuel (k) determined in S7 and the pilot injection amount Gfuel_pi (k) determined in S8 from the in-cylinder fuel injection amount Gfuel (k). The fuel amount Gfuel_main (k) (hereinafter referred to as “main injection amount”) to be calculated is calculated (see the following equation (9)), and this process is terminated.

  In S13, the basic fuel injection amount Gfuel_bs (k) obtained in S1 is determined as the final in-cylinder fuel injection amount Gfuel (k) during the lean operation, and the process proceeds to S14. In S14, the fuel injection mode is determined according to a predetermined algorithm (not shown) defined for the lean operation mode, and this process is terminated.

  FIG. 7 is a flowchart showing a specific procedure of stoichiometric operation condition determination processing for updating the stoichiometric mode flag F_Stoic_mode. In other words, FIG. 7 is a flowchart for determining whether to perform stoichiometric operation or lean operation. The process shown in FIG. 7 is executed at the same cycle (TDC synchronization) as a subroutine of the main process shown in FIG.

  In S21, it is determined whether or not a predetermined LNT protection condition set for protecting the LNT from heat is satisfied. When the stoichiometric operation is executed, the exhaust temperature rises and the carrier temperature of the catalyst in the exhaust passage also rises. Since the LNT is close to the engine, the temperature rise during the stoichiometric operation is large. The LNT protection condition is a condition set to prevent the LNT from deteriorating due to a temperature rise. More specifically, in S21, an estimated value of the carrier temperature of LNT is calculated, and when the estimated value is less than a predetermined LNT protection temperature set to, for example, about 630 to 700 ° C., the protection condition is satisfied. Otherwise, it is determined that the protection condition is not satisfied. If the determination in S21 is NO, the process moves to S22, the flag F_Stoic_mode is set to “0” to prohibit the stoichiometric operation, and the process returns to S6 in FIG. If the determination in S21 is YES, the process moves to S23.

  In S23, it is determined whether or not the LNT has reached an active state. More specifically, in S23, it is determined whether or not the estimated value of the carrier temperature of LNT is equal to or higher than a predetermined activation temperature set to about 200 ° C., for example. If the determination in S23 is NO, the process moves to S22, the flag F_Stoic_mode is set to “0” to prohibit the stoichiometric operation, and the process returns to S6 in FIG. If the determination in S23 is YES, the process moves to S24.

  In S24, it is determined whether or not the exhaust fuel injection purification operation mode flag F_ExINJ_mode is “1”. If the determination in S23 is YES, the process moves to S22, the flag F_Stoic_mode is set to “0” to prohibit the stoichiometric operation, and the process returns to S6 in FIG. If the determination in S24 is no, the process moves to S25.

In S25, it is determined whether or not the NOx purification possible condition by the exhaust fuel injection system is satisfied. This NOx purifiable condition is a state where the exhaust fuel injection system can purify NOx in the exhaust gas without discharging unnecessary components (for example, N ₂ O) from the LNT and with an appropriate purification efficiency. This is a condition for determining whether or not. More specifically, in S25, when the estimated value of the carrier temperature of LNT is equal to or higher than the purifiable temperature set at, for example, about 350 to 400 ° C., it is determined that the NOx purifiable condition is satisfied. Note that the case where the determination in S25 is NO includes, for example, a case where the engine is warming up immediately after the start of engine start, a case where the vehicle is running in an urban area, and the temperature of the LNT is lowered.

  If the determination in S25 is NO, the process moves to S26, and a map for the three-way purification mode is put on the exhaust purification using the three-way purification function of the LNT rather than the NOx purification by the exhaust fuel injection system. To update the stoichiometric mode flag F_Stoic_mode (k). More specifically, the engine speed and the engine load parameters (for example, BMEP) are acquired, and these are used as input parameters to search the map for the three-way purification mode as shown in FIG. Determine the value of F_Stoic_mode (k). As shown by broken lines in FIG. 8, when the engine operating state is roughly divided into four regions, a low rotation-high load region, a high rotation-low load region, where the amount of NOx discharged from the engine and flowing into the LNT is large, The stoichiometric operation is selected in the three regions of the high rotation and high load region (F_Stoic_mode ← 1), and the lean operation is selected in the low rotation and low load region where the amount of NOx flowing into the LNT is small (F_Stoic_mode ← 0).

  If the determination in S25 is YES, the process moves to S27, and a stoichiometric mode flag F_Stoic_mode () is used using a map for combined use mode of NOx purification by the exhaust fuel injection system and exhaust purification using the LNT three-way purification function. Update k). More specifically, the value of the flag F_Stoic_mode (k) is determined by acquiring the engine speed and the load parameter and searching a map for the combined mode as shown in FIG. 9 as these input parameters. When the map for the three-way purification mode in FIG. 8 is compared with the map for the combined mode in FIG. 9, the region where the stoichiometric operation is selected (F_Stoic_mode ← 1) is the map for the combined mode in FIG. Is narrow. This is because when the determination in S25 is YES, the amount of NOx that can be purified by the exhaust fuel injection system is larger than when NO.

  FIG. 10 is a flowchart showing a specific procedure of the additional injection fuel amount calculation process. The process shown in FIG. 10 is executed at the same cycle (TDC synchronization) during the stoichiometric operation or the exhaust fuel injection purification operation as a subroutine of the main process shown in FIG. In this additional injection fuel amount calculation processing, the after-injection amount Gfuel_aft (k) which is the amount of fuel supplied by after-injection, and the additional exhaust fuel injection amount Gfuel_ex_add (corresponding to the amount of fuel injected from the exhaust fuel injector as an alternative to after-injection) k).

  In S31, an additional injection fuel amount Gadd, which is a sum of the after injection amount and the additional exhaust fuel injection amount, is calculated, and the process proceeds to S32. More specifically, in S31, calculation is performed by a known method such as map search using the engine speed and load parameters as inputs.

  FIG. 11 is an example of a map for determining the additional injected fuel amount Gadd (k). As shown in FIG. 11, the additional injected fuel amount Gadd is set to a larger value as the engine speed increases or the engine load increases.

In S32, it is determined whether or not a failure flag F_ExINJ_NG indicating that the exhaust fuel injection system has failed is “0”. This failure flag F_ExINJ_NG is updated by a process shown in FIG. 29 described later. F_ExINJ_NG = 0 when the exhaust fuel injection system is normal, and F_ExINJ_NG = 1 when a failure occurs. If the determination in S32 is NO and the fuel cannot be injected from the exhaust fuel injector, the process proceeds to S33. In S33, as shown in the following formula (10), all of the additional injected fuel amount Gadd (k) determined in S31 is distributed to after injection.

  If the determination in S32 is YES and the fuel can be injected from the exhaust fuel injector, the process proceeds to S34. In S34, an estimated value Tcc_hat (k) of the carrier temperature of the LNT is calculated, and this estimated value Tcc_hat ( Based on k), an additional injection ratio Radd (k), which is a ratio between the after injection amount and the additional exhaust fuel injection amount, is calculated, and the process proceeds to S35.

FIG. 12 is an example of a map for determining the additional injection ratio Radd (k). When fuel is injected from an exhaust fuel injector in a state where the carrier temperature of the LNT is low, an unnecessary intermediate product N ₂ O is likely to be generated in the LNT. For this reason, as shown in FIG. 12, when the estimated value of the carrier temperature of LNT is lower than about 400 ° C., the ratio Radd (k) is set to 0 so that fuel is not injected more than necessary from the exhaust fuel injector. To do. In addition, when the estimated value of the carrier temperature of LNT is about 400 ° C. or higher, the ratio Radd (k) is gradually increased as the carrier temperature increases.

Returning to FIG. 10, in S35, as shown in the following equation (11), the additional injected fuel amount Gadd (k) is divided into after injection and exhaust fuel injection in accordance with the ratio Radd (k) determined in S34. This process ends.

  FIG. 13 is a flowchart showing a specific procedure for calculating the target pre-catalyst air-fuel ratio. The process shown in FIG. 13 is executed at the same cycle (TDC synchronization) during the stoichiometric operation or the exhaust fuel injection purification operation as a subroutine of the main process shown in FIG. In the target pre-catalyst air-fuel ratio calculation, the target pre-catalyst air-fuel ratio AFcmd (k) corresponding to the target value for the output of the pre-catalyst LAF sensor during the stoichiometric operation or the exhaust fuel injection purification operation and the in-cylinder fuel injection amount Gfuel_cyl ( k).

  In S41, it is determined whether or not the exhaust fuel injection purification operation mode flag F_ExINJ_mode is “1”. If the determination in S41 is YES (when the exhaust fuel injection purification operation is being performed), the process proceeds to S42, and the target pre-catalyst air-fuel ratio is input by a known method such as map search using the engine speed and load parameters as inputs. AFcmd (k) is determined, and the process proceeds to S43.

  FIG. 14 is a diagram showing an example of a map for determining the target pre-catalyst air-fuel ratio AFcmd during the exhaust fuel injection purification operation. The lower the concentration of oxygen in the exhaust gas flowing into the LNT, the lower the probability that LNT will directly oxidize HC and produce water and carbon dioxide, so the production efficiency of intermediate products that contribute to NOx purification increases. Therefore, the NOx concentration in the exhaust gas flowing into the exhaust volume and the LNT becomes high, and the oxygen in the exhaust gas flowing into the LNT in an operating state (high rotation range or high load range) where efficient NOx purification is required in the LNT. It is preferable to lower the concentration. In order to achieve this in the present invention, as shown in FIG. 14, the target pre-catalyst air-fuel ratio AFcmd is set to become richer as the engine speed increases or the load increases.

Returning to FIG. 13, in S43, the in-cylinder fuel injection amount is determined based on the basic fuel injection amount Gfuel_bs (k) and the target pre-catalyst air-fuel ratio AFcmd (k) determined in S1 of FIG. The provisional value Gfuel_cyl (k) is calculated, and the process returns to S10 of FIG. Here, αst in the following equation (12) is a theoretical air-fuel ratio, and is set to a value (for example, 14.5) corresponding to the fuel used.

If the determination in S41 is NO (when the stoichiometric operation is in progress), the process proceeds to S45, and the basic fuel injection amount Gfuel_bs (k) determined in S1 of FIG. The required fuel amount Gfuel_rq (k) is calculated based on the target value AFcmd_ds_tdc (k) obtained by resampling the target post-catalyst air-fuel ratio AFcmd_ds updated in a cycle of 10 to 50 msec according to the process of No. 16, and the process proceeds to S46. This target value AFcmd_ds_tdc (k) corresponds to a target for the exhaust air / fuel ratio in the LNT downstream of the exhaust fuel injector. Therefore, the required fuel amount Gfuel_rq (k) obtained by the following equation (13) corresponds to a required value for the total fuel amount that is the sum of the fuel injected into the cylinder and the fuel injected from the exhaust fuel injector.

In S46, the provisional value Gfuel_cyl (k) of the in-cylinder fuel injection amount is calculated by subtracting the exhaust fuel injection amount Gfuel_ex (k) from the required fuel amount Gfuel_rq (k) (see the following equation (14)), and S47 Move on.

In S47, the target pre-catalyst air-fuel ratio AFcmd (k) is calculated by dividing the estimated value Gair_cyl_hat of the in-cylinder fresh air amount calculated by a process (not shown) by the provisional value Gfuel_cyl (k) of the in-cylinder fuel injection amount. Then (see the following formula (15)), the process returns to S10 of FIG.

  FIG. 15 is a flowchart showing a specific procedure of the pre-catalyst air-fuel ratio feedback calculation for determining the air-fuel ratio correction coefficient KAF. The process shown in FIG. 15 is executed at the same cycle (TDC synchronization) during the stoichiometric operation or the exhaust fuel injection purification operation as a subroutine of the main process shown in FIG.

  In S51, it is determined whether or not the pre-catalyst LAF sensor has reached activity. If the determination in S51 is NO, the correction coefficient KAF (k) = 1 is set without performing the following feedback calculation (S52), and the process returns to S11 in FIG.

If the determination in S51 is YES, the deviation E_af (k) between the output value AFact_up (k) of the pre-catalyst LAF sensor and the target value AFcmd (k) determined by the processing of FIG. The correction coefficient KAF (k) is determined using a known feedback algorithm so that (see -1)) becomes 0 (S53), and the process returns to S11 of FIG. As an example of the calculation in S53, the following formulas (16-1) to (16-3) show calculation formulas when the correction coefficient KAF (k) is determined using the sliding mode algorithm. In Expression (16-2), “Pole_af” is a switching function setting parameter, and is set to a value larger than −1 and smaller than 0 (for example, −0.65). In the equation (16-3), the two feedback gains “Krch_af” and “Kadp_af” are set to negative values. Note that the compensation speed of the deviation of the pre-catalyst feedback in S53 is preferably set higher than the speed of the post-catalyst feedback in FIG.

  FIG. 16 is a flowchart showing a specific procedure of post-catalyst air-fuel ratio feedback calculation for determining the target post-catalyst air-fuel ratio AFcmd_ds. The processing shown in FIG. 16 is executed in the ECU at a control cycle ΔTex (10 to 50 msec) of the exhaust fuel injection system. In the following description, a value that is updated or sampled at the period ΔTex is denoted by “m” in parentheses. As described in the process of FIG. 13, the target post-catalyst air-fuel ratio AFcmd_ds calculated in the process of FIG. 16 is used as the target value of the exhaust air-fuel ratio in the LNT on the downstream side of the exhaust fuel injector.

  In S61, it is determined whether or not the pre-catalyst LAF sensor has reached activity. If the determination in S61 is NO, the target value AFcmd_ds (m) is set to a predetermined reference value AFcmd_bs (a fixed value, for example, 14.5) without performing the following feedback calculation (S62). Exit. If the determination in S61 is yes, the process moves to S63.

  In S63, it is determined whether or not the stoichiometric mode flag F_Stoic_mode (m) is 1. If the determination in S63 is YES, the process proceeds to S64. If the determination is NO, the process proceeds to S62, and AFcmd_ds (m) = AFcmd_bs is set as described above.

  In S64, it is determined whether or not a reduction process completion flag F_CRD_Done (m) described later is “1”. As described above, the stoichiometric operation starts when the stoichiometric mode flag F_Stoic_mode (m) is changed from “0” to “1” during the lean operation. However, since the lean operation has been performed so far, the LNT stores a large amount of oxygen, and even if the stoichiometric operation is started, the LNT's three-way purification function cannot immediately be exhibited. For this reason, immediately after the flag F_Stoic_mode (m) changes from “0” to “1”, the air-fuel ratio is biased slightly to the rich side (so-called weakly rich) from stoichiometric over a predetermined period, so that the LNT occludes it. A reduction process for releasing the oxygen that has been performed in a short time is executed. This reduction process completion flag F_CRD_Done (m) is a flag indicating that the reduction process immediately after the start of the stoichiometric operation has ended, and is updated by a weak rich mode end determination process shown in FIG. 19 described later. Hereinafter, the operation mode that promotes the reduction of the direct catalyst immediately after the start of the stoichiometric operation is referred to as “weakly rich mode”. An operation mode in which the target post-catalyst air-fuel ratio AFcmd_ds (m) is determined based on the output of the post-catalyst LAF sensor during the stoichiometric operation is referred to as “post-catalyst air-fuel ratio feedback mode”.

  When the determination in S64 is NO, the process proceeds to S65, and the target post-catalyst air-fuel ratio AFcmd_ds (m) is determined under the weak rich mode. More specifically, in S65, the estimated value Tcc_hat (m) of the carrier temperature of the LNT and the estimated value Gex_hat (m) of the exhaust volume are obtained, and based on these two estimated values Tcc_hat (m) and Gex_hat (m). Thus, the target post-catalyst air-fuel ratio AFcmd_ds (m) is determined by searching a predetermined map, and this process is terminated.

  FIG. 17 is an example of a map for determining the target post-catalyst air-fuel ratio AFcmd_ds under the weak rich mode. As shown in FIG. 17, in the weak rich mode, the target post-catalyst air-fuel ratio AFcmd_ds corresponds to the estimated value Tcc_hat of the carrier temperature and the estimated value Gex_hat of the exhaust volume within the weakly rich region (about 14.5 to 13.5). Set to a value. More specifically, the target post-catalyst air-fuel ratio AFcmd_ds is set to the rich side in the weakly rich region as the carrier temperature of the LNT increases or the exhaust volume decreases.

  Returning to FIG. 16, if the determination in S64 is YES, the process proceeds to S66, and the target post-catalyst air-fuel ratio AFcmd_ds (m) is determined under the post-catalyst stoichiometric feedback mode. In S66, the estimated value Tcc_hat (m) of the carrier temperature of the LNT and the estimated value Gex_hat (m) of the exhaust volume are acquired, and predetermined based on these two estimated values Tcc_hat (m) and Gex_hat (m). By searching the map, the three-way purification target air-fuel ratio AFcmd_ds_twc (m) is determined, and the process proceeds to S67.

FIG. 18 is an example of a map for determining the three-way purification target air-fuel ratio AFcmd_ds_twc (m). As shown in FIG. 18, the three-way purification target air-fuel ratio AFcmd_ds_twc (m) is set in the vicinity of the theoretical air-fuel ratio 14.5. The ternary purification target air-fuel ratio AFcmd_ds_twc (m) is corrected to the rich side as the carrier temperature of the LNT increases. Further, as the exhaust volume increases (in other words, as the load increases), the amount of NOx discharged from the engine increases and the exhaust passage speed in the LNT increases. As a result, the NOx purification rate in the LNT decreases. In order to compensate for such a decrease in the NOx purification rate, the three-way purification target air-fuel ratio AFcmd_ds_twc (m) is corrected to the rich side as the exhaust volume increases as shown in FIG. 18, and CO, H ₂ on the LNT is corrected. Increase the amount of reducing agent such as NH ₃ produced.

Returning to FIG. 16, in S67, the deviation E_ds (m) between the output value AFact_ds (m) of the post-catalyst LAF sensor and the ternary purification target air-fuel ratio AFcmd_ds_twc (m) (see the following equation (17-1)) The target post-catalyst air-fuel ratio AFcmd_ds (m) is determined using a known feedback algorithm so as to be 0, and this processing is terminated. As an example of the feedback calculation in S67, the following formulas (17-1) to (17-3) show calculation formulas when the target post-catalyst air-fuel ratio AFcmd_ds (m) is determined using the sliding mode algorithm. In Expression (17-2), “Pole_ds” is a switching function setting parameter, and is set to a value larger than −1 and smaller than 0 (for example, −0.85). In the equation (17-3), the two feedback gains “Krch_ds” and “Kadp_ds” are set to negative values.

  FIG. 19 is a flowchart illustrating a specific procedure of weak rich mode end determination processing for updating the reduction processing completion flag F_CRD_Done. The process shown in FIG. 19 is executed in the ECU at the same control cycle ΔTex (10 to 50 msec) as the post-catalyst air-fuel ratio feedback calculation of FIG. In the weak rich mode end determination process of FIG. 19, the reduction process completion flag F_CRD_Done is updated based on the output of the post-catalyst LAF sensor.

  In S71, it is determined whether or not the stoichiometric mode flag F_Stoic_mode (m) is “0” and the exhaust fuel injection purification operation mode flag F_ExINJ_mode (m) is “0”. If the determination in S71 is YES, that is, if neither the stoichiometric operation nor the exhaust fuel injection purification operation is performed, the process proceeds to S72, the reduction process completion flag F_CRD_Done (m) = 0 is set, and this process ends.

  If the determination in S71 is NO, that is, if the stoichiometric operation is being performed or if the exhaust fuel injection purification operation is being performed, the process proceeds to S73, where the output value AFact_ds (m) of the post-catalyst LAF sensor is a predetermined inversion determination. It is determined whether or not the threshold value AF_ln is greater. As described with reference to FIGS. 16 to 18, immediately after the start of the stoichiometric operation, the air-fuel ratio is set to be slightly rich, and the oxygen stored in the LNT is released and the reduction supplied by the weak enrichment. Used to oxidize agents. Therefore, whether or not the reduction process is completed can be determined by whether or not the output value AFact_ds (m) of the post-catalyst LAF sensor exceeds the inversion determination threshold value AF_ln. The inversion determination threshold value AF_ln is set to a value (for example, 14.6) that is slightly larger than the theoretical air-fuel ratio αst.

  If the determination in S73 is NO, this process ends without updating the flag F_CRD_Done. If the determination in S73 is YES, the process moves to S74, the flag F_CRD_mode (m) = 1 is set to clearly indicate that the reduction process is completed, and this process ends.

  FIG. 20 is a time chart showing a specific example of the processing of FIGS. FIG. 20 shows the stoichiometric mode flag F_Stoic_mode, the reduction process completion flag F_CRD_Done, the target post-catalyst air-fuel ratio AFcmd_ds, the integrated value of the reducing agent amount supplied to the LNT, and the post-catalyst LAF sensor output value AFact_ds in order from the top. . FIG. 20 shows a case where the stoichiometric mode flag F_Stoic_mode is changed from “0” to “1” at time t1.

  As described with reference to FIG. 16, immediately after the start of the stoichiometric operation (time t1 in FIG. 20), the target post-catalyst air-fuel ratio AFcmd_ds is set to be slightly rich (see S65 of FIG. 16). Therefore, after time t1, excess fuel is supplied to the LNT as a reducing agent, thereby releasing the oxygen stored in the LNT.

  Thereafter, in response to the fact that the oxygen stored in the LNT is completely released, the output value AFact_ds of the post-catalyst LAF sensor exceeds the inversion determination threshold AF_ln at time t2 (see S73 in FIG. 19). As a result, the reduction process completion flag F_CRD_Done switches from 0 to 1, the weak rich mode ends, and the post-catalyst air-fuel ratio feedback mode starts (see S64 in FIG. 16). In this post-catalyst air-fuel ratio feedback mode, the target post-catalyst air-fuel ratio AFcmd_ds is determined so that the output value AFact_ds of the post-catalyst LAF sensor becomes the three-way purification target air-fuel ratio AFcmd_ds_twc calculated by the map (FIG. 16 S67).

  FIGS. 21 and 22 are flowcharts showing a specific procedure of the exhaust fuel injection control for determining the exhaust fuel injection mode by the exhaust fuel injector. The processing shown in FIGS. 21 and 22 is executed in the ECU at the same control cycle tm (10 to 50 msec) as the post-catalyst air-fuel ratio feedback calculation of FIG. As shown in FIGS. 21 and 22, the exhaust fuel injection control includes a step of setting target temperatures of the two LAF sensors (S92, S98, S100), a step of determining parameters required for intermittent injection (S115), A step of determining an exhaust fuel injection amount (S93, S99, S107, S111), a step of detecting system deterioration (S108), a step of updating an exhaust fuel injection purification operation mode flag F_ExINJ_mode (S94, S109), including.

  In S91, it is determined whether or not the exhaust fuel injection system is normal (whether or not a failure flag F_ExINJ_NG = 0 described later). If the determination in S91 is YES, the process moves to S95. If the determination in S91 is NO and the fuel cannot be injected from the exhaust fuel injector, the process proceeds to S92. In S92, the target temperature Tcmd_laf_up (m) of the detection element of the pre-catalyst LAF sensor and the detection element of the post-catalyst LAF sensor are set so that the HC deviations of the pre-catalyst LAF sensor and post-catalyst LAF sensor are as small as possible. The target temperature Tcmd_laf_ds (m) is set to a predetermined high temperature side target value Tcmd_laf_high (Tcmd_laf_up (m) = Tcmd_laf_ds (m) = Tcmd_laf_high), and the process proceeds to S93. As described above, when the exhaust fuel injection system is not normal, the stoichiometric operation can be performed under the accurate output of the LAF sensor without HC deviation by equalizing the temperature of the LAF sensor before and after the catalyst. it can. Further, in S93, the exhaust fuel injection amount Gfuel_ex (m) = 0 is set in response to the determination that the fuel cannot be injected from the exhaust fuel injector, and in S94, the exhaust fuel injection purification operation mode flag F_ExINJ_mode (m) = 0. And the process is terminated.

  In S95, it is determined whether or not the LNT protection condition set for protecting the LNT from heat is satisfied. Note that the specific contents and specific determination method of the LNT protection conditions are the same as S21 in FIG. If the determination in S95 is YES and the LNT protection condition is satisfied, the process proceeds to S96. In S96, it is determined whether or not a predetermined NOx purification condition for the exhaust fuel injection system is satisfied. In addition, since the specific content and specific determination method of this NOx purification possible condition are the same as S25 of FIG. 7, detailed description is abbreviate | omitted. If the determination in S96 is YES and the NOx purification possible condition of the exhaust fuel injection system is satisfied, the process proceeds to S97.

  In S97, it is determined whether or not the weak rich mode completion flag F_CRD_Done is “1”. The state where the weak rich mode completion flag F_CRD_Done is not “1” is a state where the LNT reduction process described with reference to FIG. 16 is not completed. In a state where the reduction process of LNT is not completed, NOx cannot be purified with sufficient efficiency even if fuel is injected from the exhaust fuel injector. Therefore, the process proceeds to S115 only when the determination in S97 is YES, and the exhaust fuel injection amount is determined under the exhaust fuel injection purification operation mode.

  If any of the determinations in S95 to S97 is NO, the process proceeds to S98. Here, the case where any of the determinations of S95 to S97 is NO corresponds to a state where fuel can be injected from the exhaust fuel injector, but the exhaust cannot be purified using the exhaust fuel injection system. In this case, as described with reference to FIG. 10, fuel injection from the fuel injection injector may be required as an alternative to after injection during the stoichiometric operation. In S98, similarly to S92, the target temperatures of the pre-catalyst LAF sensor and the post-catalyst LAF sensor are set to predetermined high-temperature side target values, and the process proceeds to S99. In S99, the additional exhaust fuel injection amount calculated in the TDC cycle by the process of FIG. 10 is resampled (Gfuel_ex_add (k) → Gfuel_ex_add (m)), and the value obtained by the resampling is set as the exhaust fuel injection amount (Gfuel_ex (m) = Gfuel_ex_add (m)), the process proceeds to S94. In S94, the exhaust fuel injection purification operation mode flag F_ExINJ_mode (m) = 0 is set, and this process ends.

  In S115, an intermittent injection parameter setting process for setting the injection cycle Tfuel_ex (m) and the injection cycle parameter Nex (m) necessary for the execution of intermittent injection is executed, and the process proceeds to S100. A specific procedure for setting the intermittent injection parameter will be described later with reference to FIG.

  In S100, the temperature of the detection element of the post-catalyst LAF sensor is set to the detection element of the pre-catalyst LAF sensor so that the HC slip downstream of the LNT can be detected using the pre-catalyst LAF sensor and the post-catalyst LAF sensor. The temperature is lowered to S101, and the process proceeds to S101. More specifically, the target temperature Tcmd_laf_up (m) of the detection element of the pre-catalyst LAF sensor is set to the high-temperature side target value Tcmd_laf_high (Tcmd_laf_up (m) = Tcmd_laf_high), and the target temperature Tcmd_laf_ds ( m) is set to a predetermined low temperature side target value Tcmd_laf_low smaller than the high temperature side target value Tcmd_laf_high (Tcmd_laf_ds (m) = Tcmd_laf_low).

  In S101, an estimated value Gnox_hat (m) of the amount of NOx flowing into the LNT and an estimated value Tcc_hat (m) of the carrier temperature of the LNT are acquired, and based on these two estimated values Gnox_hat (m) and Tcc_hat (m), A reference exhaust fuel injection amount Gfuel_ex_bs (m) serving as a reference value for the exhaust fuel injection amount is determined, and the process proceeds to S102. Here, the estimated value Gnox_hat (m) of the NOx amount flowing into the LNT can be calculated by searching a predetermined map (not shown) based on the engine speed and the load parameter, for example. In addition, a NOx sensor may be provided on the upstream side of the LNT, and the calculation may be performed based on the output of the NOx sensor, or may be performed based on the output of the neural network that receives the engine speed, load parameters, and the like. Good.

  FIG. 23 is an example of a map for determining the reference exhaust fuel injection amount Gfuel_ex_bs. As shown in FIG. 23, the injection amount Gfuel_ex_bs is set to a larger value as the amount of NOx flowing into the LNT increases. The injection amount Gfuel_ex_bs is set to a larger value as the carrier temperature of the LNT increases.

  Returning to FIG. 22, in S102, it is determined whether or not a predetermined HC slip feedback control execution condition is satisfied. Here, the contents of the HC slip feedback control execution condition will be described. First, as described with reference to FIG. 5, the HC slip feedback control involves a positive HC slip even if it is slightly downstream of the LNT. That is, during the HC slip feedback, it is necessary to inject extra fuel by an amount corresponding to the HC slip as compared with the injection amount (G2 in FIG. 5) determined so that the HC slip becomes zero.

For example, in a region where the carrier temperature of LNT is relatively low (for example, a region below 350 to 400 ° C.), L ₂ tends to generate N ₂ O from the injected fuel. Since N ₂ O is not an intermediate product that contributes to the purification of NOx, it is preferable to reduce its generation as much as possible. Therefore, in such a low temperature range, since the risk of N ₂ O generation is high, it is not preferable to inject extra fuel to the extent that HC slips.

  Further, in a region where the carrier temperature of LNT is relatively high (for example, a region of 550 ° C. or higher), most of the injected fuel is directly oxidized, and the production efficiency of intermediate products contributing to NOx purification is reduced. . Therefore, in such a high temperature range, the degree of contribution to the improvement of the NOx purification rate of the fuel supplied to the LNT is small, so it is not preferable to inject extra fuel as HC slips.

  As described above, the HC slip feedback control in the low temperature region and the high temperature region is wasteful of fuel. Therefore, in these low temperature range and high temperature range, the HC slip is generated rather than making the exhaust fuel injection amount to an amount that generates the HC slip (see G4 in FIG. 5) by executing the HC slip feedback control. It is preferable to suppress the amount to a level that does not occur (a little less than G2 in FIG. 5). Hereinafter, the control for intentionally suppressing the exhaust fuel injection amount to such an amount that HC slip does not occur is referred to as an HC slip suppression mode.

  The HC slip feedback control execution condition in S102 is a condition for determining whether or not the HC slip feedback control execution condition is preferable for executing the HC slip feedback control as described above. More specifically, the HC slip feedback control execution condition is, for example, that the estimated value of the carrier temperature of LNT is within a range of 400 to 550 ° C. If the determination in S102 is YES, the process proceeds to S103 to execute HC slip feedback control. If the determination in S102 is NO, the process proceeds to S110 to execute the HC slip suppression mode.

Next, a specific procedure (S103 to S108) of HC slip feedback control will be described. During the HC slip feedback control, the final exhaust fuel injection amount Gfuel_ex (m) is proportional to the reference exhaust fuel injection amount Gfuel_ex_bs (m) calculated in S101, and the post-catalyst LAF sensor. Is calculated by adding the feedback correction term (second term on the right side) proportional to the correction value DGfuel_ex (m) calculated based on the output value AFact_ds (m) of (See S107). This basic term is obtained by multiplying a reference exhaust fuel injection amount Gfuel_ex_bs (m) calculated using a map (see FIG. 23) by an adaptive coefficient Kff_ex (m) calculated according to an adaptive correction map described later. Defined.

  The HC slip feedback control includes a step of calculating a correction value DGfuel_ex (m) (S103 to S105), a step of calculating an adaptive coefficient Kff_ex (m) (S106), and a final exhaust fuel injection amount by equation (18). It comprises a step of determining Gfuel_ex (m) (S107) and a step of determining a failure of the exhaust fuel injection system (S108).

First, in S103, the filter value AFact_mav_ds (m) of the post-catalyst LAF sensor output and the ideal exhaust air-fuel ratio AF_exh_id_mav_up (m ) And a NOx purification parameter P_LNT (m) proportional to the HC slip amount is calculated (see the following equation (19)).

In S104, a deviation E_LNT (m) between the NOx purification parameter P_LNT (m) and the target value P_LNT_cmd (m) is calculated (see the following equation (20)), and the process proceeds to S105. During the HC slip feedback control, an amount of HC proportional to the target value P_LNT_cmd (m) of the NOx purification parameter slips steadily to the downstream side of the LNT. In the exhaust purification system shown in FIG. 1, HC slipped from the LNT is oxidized by an oxidation catalyst provided in a CSF provided downstream thereof. Therefore, the target value P_LNT_cmd (m) is set to a value slightly larger than 0 or a value smaller than the upper limit value set according to the HC oxidation treatment capability of the CSF as the HC oxidation catalyst. Also, this target value P_LNT_cmd (m) can be varied according to the engine speed, load parameters, exhaust volume, NOx amount discharged from the engine, carrier temperature of LNT, etc. Also good.

In S105, the correction value DGfuel_ex (m) of the exhaust fuel injection amount is calculated using a known feedback algorithm so that the deviation E_LNT (m) becomes “0”. As an example of the calculation in S105, the following formulas (21-1) and (21-2) show calculation formulas when the correction value DGfuel_ex is determined using the sliding mode algorithm. In the following formula (21-1), “Pole_LNT” is a switching function setting parameter, and is set to a value larger than −1 and smaller than 0 (for example, −0.85). In the equation (21-2), the two feedback gains “Krch_LNT” and “Kadp_LNT” are set to negative values.

  In S106, an adaptive coefficient calculation process for HC slip feedback control is executed, and the process proceeds to S107. As will be described later with reference to FIG. 26, in this adaptive coefficient calculation process, an adaptive coefficient Kff_ex (m) is calculated using an adaptive correction map, which will be described later, and the feedback correction term in equation (18) becomes zero. The adaptive correction map is learned. In S107, the exhaust gas injection amount Gfuel_ex () is obtained by adding the basic term calculated by multiplying the reference exhaust fuel injection amount Gfuel_ex_bs (m) by the adaptation coefficient Kff_ex (m) and the correction value DGfuel_ex (m). m) is calculated (see the above equation (18)), and the process proceeds to S108. In S108, the failure determination process of the exhaust fuel injection system which will be described later with reference to FIG. 29 is executed, and the process proceeds to S109. In S109, the flag F_ExINJ_mode (m) = 1 is set to clearly indicate that the exhaust fuel injection purification operation is being performed, and this process is terminated.

  Next, a specific procedure (S110 to S111) of the HC slip suppression control will be described. The purpose of the HC slip suppression control is to suppress the exhaust fuel injection amount to such an extent that HC slip does not occur in order to avoid unnecessary fuel consumption. However, if the exhaust fuel injection amount is suppressed excessively, the NOx purification rate is also greatly reduced. Therefore, the target of the exhaust fuel injection amount in the HC slip suppression control is slightly smaller than G2 in FIG. However, in the region where no HC slip occurs, the NOx purification parameter P_LNT is uniformly zero. For this reason, the exhaust fuel injection amount cannot be controlled to the target as described above based on the purification parameter P_LNT in the same manner as the HC slip feedback control described above.

  On the other hand, during the HC slip feedback control, the calculation algorithm of the adaptive coefficient Kff_ex of the basic term is learned so that the feedback correction term of the second term on the right side of Expression (18) becomes 0 by the process shown in S106. Therefore, if the HC slip feedback control is repeatedly executed, the exhaust fuel injection amount equivalent to that at the time of the HC slip feedback control (for example, corresponding to G4 in FIG. 5) is not obtained without performing the calculation of the purification parameter P_LNT and the feedback correction term. Can be reproduced only by the fundamental terms. During the HC slip suppression control in which the purification parameter P_LNT cannot be used, the exhaust fuel injection amount calculated by the basic term that has been learned in this way is multiplied by a reduction coefficient smaller than 1 so that the HC slip feedback control is performed. Also, an exhaust fuel injection amount that is moderately suppressed (for example, an amount slightly smaller than G2 in FIG. 5) is calculated.

In S110, an adaptive coefficient Kff_ex (m) is calculated by executing an adaptive coefficient calculation process for HC slip suppression control described later with reference to FIG. 28, and the process proceeds to S111. In S111, an exhaust fuel injection amount Gfuel_ex (m) is calculated by multiplying the basic term of the equation (18) by setting the reduction factor to 0.9 as shown in the following equation (22), The process moves to S109. In S109, the flag F_ExINJ_mode (m) = 1 is set to clearly indicate that the exhaust fuel injection purification operation is being performed, and this process is terminated.

  FIG. 24 is a flowchart showing a specific procedure of the intermittent injection parameter setting process. The process shown in FIG. 24 is executed in the control cycle ΔTex as a subroutine of the main process in FIG.

  In S151, an oxidation capability parameter indicating the HC oxidation capability of LNT is acquired, and the process proceeds to S152. As the oxidation capacity parameter, for example, an estimated value Tcc_hat (m) of an LNT carrier temperature and a load parameter (for example, BMEP. Other engine loads such as required torque, fuel injection amount, estimated engine torque, and exhaust volume) Is used), but the present invention is not limited to this. The HC oxidation capacity of LNT varies depending on the LNT environment specified by the combination of temperature and load parameters, as well as the degree of LNT degradation, individual variations, and the like. Therefore, the oxidation ability parameter may be a numerical value taking into account such a degree of deterioration and individual variation.

  In S152, using the oxidation capacity parameter acquired in S151 as an input, an injection cycle parameter Nex (m) that is an integer greater than 1 is calculated by a known method such as map search, and the process proceeds to S153.

  FIG. 25 is an example of a map for determining the injection cycle parameter Nex. In order to prevent the fuel supplied by intermittent injection from being directly oxidized in the LNT without contributing to the purification of NOx, the fuel is injected during one cycle by increasing the injection period as the oxidizing ability of the LNT increases. It is preferable to increase the amount. Therefore, the injection cycle parameter Nex is set to a larger value as the HC oxidation capability of LNT increases. Further, as described above, the injection cycle parameter Nex is an integer larger than 1. Therefore, it is preferable to change the injection cycle parameter Nex in a step shape according to the oxidation ability of the LNT as shown in FIG.

Returning to FIG. 24, in S153, the control cycle ΔTex multiplied by an integer injection cycle parameter Nex (m) is calculated as the injection cycle Tfuel_ex (m) of intermittent injection (see the following equation (23)). The process ends.

  FIG. 26 is a flowchart showing a specific procedure of adaptive coefficient calculation processing for HC slip feedback control. The process shown in FIG. 26 is executed at the control cycle ΔTex during the execution of the HC slip feedback control as a subroutine of the main process of FIG.

Before describing the specific procedure of the adaptive coefficient calculation process with reference to FIG. 26, the calculation algorithm for calculating the adaptive coefficient Kff_ex and the contents of the learning procedure of the calculation algorithm will be described. In the adaptive coefficient calculation process, as shown in the following equation (24), an estimated value Gnox_hat (m) of the NOx amount flowing into the LNT and an estimated value Tcc_hat (m) of the carrier temperature of the LNT are input, and an adaptive correction map (described later) The adaptive coefficient Kff_ex (m) is calculated by adding the map value M_tcc_nox (m) calculated by the equation (25) and the like and the initial value “1” of the adaptive coefficient Kff_ex.

The following formula (25) is a formula showing a specific configuration of the adaptive correction map, that is, a specific calculation formula of the map value M_tcc_nox (m). In the following equation (25), the value Wex_nox_i (m) (i = 1, 2, 3) is calculated by a weight function map as shown in the upper part of FIG. 27 with the estimated NOx amount Gnox_hat (m) as an input. Are the three types of NOx amount weight function values. The values Wex_tlnt_j (m) (j = 1, 2, 3) are input for the three types of LNT temperatures calculated by the weight function map as shown in the lower part of FIG. 27 with the estimated carrier temperature value Tcc_hat (m) as an input. The value of the weight function. Nine coefficients Dkff_ij (m) (i or j = 1, 2, 3) are region adaptive update values that are determined in association with combinations of nine types of weight functions. These nine area adaptive update values Dkff_ij (m) are values that specify the contribution of the combination of weight functions, in other words, the height of the combination of the weight functions. That is, changing the value of the region adaptive update value Dkff_ij (m) intuitively corresponds to changing the shape of the adaptive correction map for calculating the map value M_tcc_nox (m). In the adaptive coefficient calculation process of FIG. 26, the region adaptive update value Dkff_ij (m) is changed so that the absolute value of the feedback correction term (DGfuel_ex (m)) in Expression (18) becomes small.

  The three weight functions Wex_nox_1 to Wex_nox_3 are defined for the NOx amount estimated value Gnox_hat, and the three weight functions Wex_tlnt_1 to Wex_tlnt_3 are defined for the carrier temperature estimated value Tcc_hat. These weight functions are normalized so that the total sum of all weight function values becomes 1 for any temperature or NOx amount as shown in FIG.

  Returning to FIG. 26, in S121, the estimated value Gnox_hat (m) of the NOx amount and the estimated value Tcc_hat (m) of the carrier temperature are obtained, and a map as shown in FIG. 27 is searched based on these estimated values. Three NOx amount weight function values Wex_nox_i (m) and three LNT temperature weight function values Wex_tlnt_j (m) are calculated, and the process proceeds to S122.

In S122, the adaptive error signal Eadp_ex (m) is calculated by multiplying the correction value DGfuel_ex (m) calculated in S105 of FIG. 22 by “−1” (see the following equation (26-1)), and in S123 Move. In S123, nine weighted adaptive error signals WEadp_ex_ij (m) (i or j = 1, 2, 3) are calculated according to the following equation (26-2), and the process proceeds to S124. In S124, each region adaptive update value Dkff_ij (m) is calculated by integrating each weighted adaptive error signal WEadp_ex_ij (m) multiplied by a negative gain Kadp_ff (see the following equation (26-3)). To S125. In other words, the region adaptive update value Dkff_ij (m) is calculated so that the absolute value of the correction value DGfuel_ex (m) is reduced by the processing of S121 to S124 described above. Therefore, the processing of S121 to S124 constitutes a basic term correction means for correcting the adaptive correction map. In S125, the map value M_tcc_nox (m) is calculated by inputting the plurality of weight function values and region adaptive update values calculated as described above to the adaptive correction map of Expression (25), and further, Expression (24) To calculate the adaptive coefficient Kff_ex (m), and the process returns to S107 in FIG.

  FIG. 28 is a flowchart showing a specific procedure of adaptive coefficient calculation processing for HC slip suppression control. The process shown in FIG. 28 is executed at the control cycle ΔTex during the HC slip suppression control as a subroutine of the main process of FIG.

  In S131, the estimated value Gnox_hat (m) of the NOx amount and the estimated value Tcc_hat (m) of the carrier temperature are obtained, and by searching a map as shown in FIG. 27 based on these estimated values, the weight for each NOx amount is obtained. The function value Wex_nox_i (m) and each LNT temperature weight function value Wex_tlnt_j (m) are calculated, and the process proceeds to S132. In S132, the plurality of weight function values calculated as described above and the region adaptive update value Dkff_ij (m calculated by the learning processing of Expressions (26-1) to (26-3) during the HC slip feedback control. ), The adaptive coefficient Kff_ex (m) is calculated according to the above equations (24) and (25), and the process returns to S109 of FIG.

  FIG. 29 is a flowchart showing a specific procedure of the failure determination process of the exhaust fuel injection system. The process shown in FIG. 29 is executed in the control cycle ΔTex during the HC slip feedback control as a subroutine of the main process in FIG. In this failure determination process, the exhaust fuel injection system fails by comparing the correction value DGfuel_ex (m) and the region adaptive update value Dkff_ij (m) calculated during the HC slip feedback control with the threshold values set for each. It is determined whether or not.

  In S141, it is determined whether or not the correction value DGfuel_ex (m) is smaller than a predetermined lower limit failure threshold DG_Low_NG. If the determination in S141 is no, the process proceeds to S142. In S142, it is determined whether any of the nine area adaptive update values Dkff_ij (m) is smaller than a predetermined lower limit failure threshold DKFF_Low_NG. If the determination in S142 is no, the process moves to S145. If the determination in either S141 or S142 is YES, it is determined that the exhaust fuel injection system has failed, the process proceeds to S143, the failure flag F_ExINJ_NG (m) = 1 is set, and a warning lamp (not shown) is turned on. Then (S144), the process proceeds to S109 in FIG.

  Here, the reason why a failure of the exhaust fuel injection system can be determined by the processing of S141 and S142 will be described. The exhaust fuel injection amount during HC slip feedback control is determined so that HC slip occurs. As the correction value DGfuel_ex (m) used in the determination in S141 becomes smaller, the final exhaust fuel injection amount Gfuel_ex (m) is corrected to a smaller value (see Expression (18)). Further, as the region adaptive update value Dkff_ij (m) used in the determination in S142 becomes smaller, the adaptation coefficient Kff_ex (m) becomes smaller (see equations (24) and (25)), and the final exhaust fuel injection amount Gfuel_ex (m ) Is corrected to a small value (see equation (18)). Therefore, it can be said that the state in which the correction value or the region adaptive update value is smaller than the respective threshold values is a state in which the amount of exhaust fuel injection necessary for generating the HC slip is smaller than that in the normal state.

  According to the determination in S141 or S142, it is possible to specify a failure in which the oxidation ability of LNT is lower than that in the normal state. When the oxidation capacity of LNT decreases, intermediate products are less likely to be produced, and much of the fuel supplied from the exhaust fuel injector does not contribute to the production of intermediate products on the LNT and is discharged downstream as HC. This is because that. Further, when the determination in S141 or S142 is YES, the amount of intermediate product produced is smaller than the original amount, and therefore the NOx purification rate is also lowered than the original. Such a decrease in the oxidation ability of LNT can be caused by, for example, deterioration of LNT.

  In S145, it is determined whether or not the correction value DGfuel_ex (m) is greater than a predetermined upper limit failure threshold DG_High_NG. If the determination in S145 is no, the process moves to S146. In S146, it is determined whether any of the nine area adaptive update values Dkff_ij (m) is larger than a predetermined upper limit failure threshold DKFF_Low_NG. If the determination in S146 is YES, it is determined that the exhaust fuel injection system is normal, the failure flag F_ExINJ_NG (m) = 0 is set (S147), and the process proceeds to S109 in FIG. If the determination in any one of S145 and S146 is YES, it is determined that the exhaust fuel injection system has failed, the failure flag F_ExINJ_NG (m) = 1 is set (S143), and a warning lamp is lit ( S144), the process proceeds to S109 in FIG. The state in which the correction value or the region adaptive update value is larger than the respective threshold values is contrary to the above-described cases of S141 and S142, and the exhaust fuel injection amount necessary for generating the HC slip is more than normal. It can be said that the situation has increased.

  According to the determination in S145 or S146, it is possible to identify a failure in which the oxidation ability of LNT is higher than that in the normal state. This is because when the oxidation capacity of the LNT increases, most of the fuel supplied from the exhaust fuel injector is directly oxidized on the LNT without contributing to the generation of intermediate products. In addition, when the determination in S145 or S146 is YES, the amount of intermediate product produced is smaller than the original amount, so the NOx purification rate is also lower than the original. Such an increase in the oxidation capacity of LNT is caused when a metal-based (for example, Pt) additive is added to the fuel or when the particle size of the fuel injected from the exhaust fuel injector is smaller than normal. It can occur when it becomes smaller.

  FIG. 30 is a flowchart showing a specific procedure of intermittent injection control performed under the exhaust fuel injection amount Gfuel_ex and the injection cycle Tfuel_ex set by the processing of FIGS. 21 and 22. FIG. 31 is a time chart showing a specific example of intermittent injection control realized by the flowchart of FIG. 30. In the process shown in FIG. 30, when the exhaust fuel injection purification operation mode flag F_ExINJ_mode is “1” and the intermittent injection of fuel from the exhaust fuel injector is requested, the cycle ΔTinj (for example, sufficiently shorter than the control cycle ΔTex) (Several microseconds to several tens of microseconds). In the following description, a value updated or sampled with a period ΔTinj is denoted by “j” in parentheses.

  In S161, the injection cycle updated at the cycle ΔTex by the process of FIG. 24 is oversampled (Tfuel_ex (m) → Tfuel_ex (j)), and whether the injection control timer TM_INJ (j) is larger than the injection cycle Tfuel_ex (j). Is determined. The injection control timer TM_INJ (j) is incremented by the time elapsed every control cycle ΔTinj (see S168 described later), and reset every injection cycle Tfuel_ex (j) (refer to S162 described later). If the determination in S161 is YES, the injection control timer TM_INJ (j) is reset to “0” (S162), and then the process proceeds to S163. If the determination in S161 is NO, the process proceeds to S163 without resetting the timer.

In S163, the exhaust fuel injection amount updated at the period ΔTex by the process of FIG. 22 is oversampled (Gfuel_ex (m) → Gfuel_ex (j)), and the injection amount Gfuel_ex (j) per unit time is calculated as one injection cycle. In the meantime, it is converted into the exhaust fuel injection time Tfuel_inj (j) corresponding to the time for which fuel injection is performed by the exhaust fuel injector (that is, the valve opening time), and the process proceeds to S164. The conversion from the injection amount to time is realized by the following equation (27), for example. In the following equation, Ginj corresponds to the amount of fuel injected from the exhaust fuel injector during the period ΔTinj.

  In S164 and S165, it is determined whether or not the injection control timer TM_INJ (j) has exceeded one of a predetermined injectable period Tfuel_max (j) and injection time Tfuel_inj (j). This injectable period Tfuel_max (j) is an upper limit for the injection control timer TM_INJ (j) that is determined so that intermittent injection is reliably performed as shown in FIG. 31, and is a value smaller than the injection cycle Tfuel_ex (j) ( For example, Tfuel_max (j) = Tfuel_ex (j) / 2) is set.

  If both S164 and S165 are NO, exhaust fuel injection is executed (exhaust fuel injector is opened) (see S166), and the process proceeds to S168. If either of the determinations at S164 and S165 is YES, the exhaust fuel injection is stopped (the exhaust fuel injector is driven to close the valve) (see S167), and the process proceeds to S168. In S168, the injection control timer TM_INJ (j) is updated (TM_INJ (j) ← TM_INJ (j) + ΔTinj), and this process ends.

Note that if the determination in S164 is YES, the actual injection time Tfuel_inj (j) is limited to the injectable period Tfuel_max (j), and therefore the fuel that is actually injected from the exhaust fuel injector per unit time. The amount decreases by a predetermined amount ΔG from the amount determined in the process of FIG. 22 (Gfuel_ex → Gfuel_ex−ΔG). Therefore, in consideration of this point, when the determination in S164 is YES, for example, the ideal exhaust air-fuel ratio AF_ex_id_up calculated by the equation (4-2) or (5-1) in the processing of FIG. Each may be replaced by the following formula (28-1) or (28-2).

Next, a second embodiment of the present invention will be described with reference to the drawings.
FIG. 32 is a diagram showing a configuration of the engine 1 and its exhaust purification system 2B according to the present embodiment. The exhaust purification system 2B of FIG. 32 includes the exhaust purification system 2 of the first embodiment shown in FIG. 1, the position where the pre-catalyst LAF sensor 51B is provided, and the configuration of the ECU 3B that performs in-cylinder fuel injection control and exhaust fuel injection control Is different.

  The pre-catalyst LAF sensor 51B is provided between the LNT 41 and the exhaust fuel injector 452 in the exhaust passage 11. In this case, unlike the system 2 of FIG. 1, the output value AFact_up ′ of the LAF sensor 51B is directly affected by the intermittent injection from the exhaust fuel injector 452, but if the processing shown below is changed, FIG. The in-cylinder fuel injection control and the exhaust fuel injection control shown in FIG.

First, by providing the pre-catalyst LAF sensor 51B on the downstream side of the exhaust fuel injector 452, the output value AFact_up '(m) of the pre-catalyst LAF sensor 51B is directly used as the ideal exhaust air-fuel ratio AF_exh_id_up (upstream of the LNT). Available as m). That is, in the exhaust purification system 2B, the expressions (4-1) to (4-2) are replaced with the following expression (29).

However, as described with reference to FIGS. 2 and 36, when intermittent injection is performed by the exhaust fuel injector, the output value AFact_up ′ of the pre-catalyst LAF sensor 51B is sawed similarly to the output value AFact_ds of the post-catalyst LAF sensor 52. It will be wavy. Therefore, in the exhaust purification system 2B of FIG. 32, during the intermittent injection, it is necessary to apply a moving average filter similar to the equation (3) to the output value AFact_up ′ of the pre-catalyst LAF sensor 51B. . More specifically, for example, the arithmetic expression (29) of the ideal exhaust air / fuel ratio AF_exh_id_up is replaced with the following expression (30).

Since the ideal exhaust air / fuel ratio AF_exh_id_up calculated by the equation (30) already includes the phase delay due to the moving average, the equation (7) for the purification parameter P_LNT is expressed by the following equation (31): Is replaced by

Secondly, the processing of S45 to S47 of the target pre-catalyst air-fuel ratio calculation of FIG. 13 is changed as follows. In the exhaust purification system 2 of FIG. 1, the pre-catalyst LAF sensor 51 is provided on the upstream side of the exhaust fuel injector 452, and therefore, the exhaust fuel injection amount Gfuel_ex () is calculated from the target value AFcmd_ds_tdc (k) calculated according to the processing of FIG. After removing the influence of k), the target value AFcmd (k) for the output value AFact_up (k) of the pre-catalyst LAF sensor 51 was calculated (see equations (12) to (15)). In contrast, in the exhaust purification system 2B of FIG. 32, the pre-catalyst LAF sensor 51B is provided on the downstream side of the exhaust fuel injector 452, and therefore the target value AFcmd '(k) of the output value AFact_up' (k) of the catalyst LAF sensor 51B. ) Need not be excluded from the influence of the exhaust fuel injection amount Gfuel_ex (k). That is, in the exhaust purification system 2B, the expression (14) is replaced with the following expression (32). According to the exhaust gas purification system 2B of FIG. 32, the above-described replacement has substantially the same effect as the exhaust gas purification system 2 of FIG.

  Although one embodiment of the present invention has been described above, the present invention is not limited to this. The present invention utilizes a phenomenon in which the difference between the output value of the air-fuel ratio sensor and the true value varies depending on the HC concentration even though the concentration of the entire fuel component in the exhaust gas is constant. In the above embodiment, as described above with reference to FIG. 4, as the embodiment of the present invention is realized, the output value shifts to the lean side with respect to the true value as the HC concentration in the exhaust gas becomes higher. Although the case where the air-fuel ratio sensor having characteristics is used in the post-catalyst air-fuel ratio sensor 52 (see FIG. 1) on the downstream side of the catalyst has been described in detail, the present invention is not limited to this. For example, if there is an air-fuel ratio sensor having a characteristic that the difference between the output value and the true value changes to the rich side as the HC concentration in the exhaust gas increases, such an air-fuel ratio sensor is used as the post-catalyst air-fuel ratio sensor 52. Even if it uses for, it can achieve the same objective and effect as the invention which concerns on the said embodiment. When the air-fuel ratio sensor whose output value shifts in the direction opposite to that in FIG. 4 is used in this way, the sign of the NOx purification parameter P_LNT defined by the above formula (6) or (7) is reversed, and consequently It should be noted that the sign of the target value P_LNT_cmd with respect to the NOx purification parameter P_LNT is also reversed. Considering that the sign of the NOx purification parameter P_LNT is reversed as described above, the specific control procedure described with reference to FIGS. 6 to 31 is the air-fuel ratio in which the shift direction of the output value is opposite to that in FIG. The sensor can also be applied to an exhaust purification system using the post-catalyst air-fuel ratio sensor 52.

1. Engine (internal combustion engine)
DESCRIPTION OF SYMBOLS 11 ... Exhaust passage 2, 2B ... Exhaust gas purification system 3, 3B ... ECU (Sampling means, moving average value calculation means, injection amount calculation means, intermittent injection execution means, injection cycle calculation means, estimation means, injection cycle parameter setting means)
41 ... LNT (catalyst)
452 ... Exhaust fuel injector (injector)
52 ... LAF sensor after catalyst (air-fuel ratio sensor)

Claims (3)

An internal combustion engine comprising: an injector that intermittently injects fuel upstream of a catalyst provided in an exhaust passage of the internal combustion engine; and an air-fuel ratio sensor that generates a signal corresponding to the air-fuel ratio of exhaust downstream of the injector An exhaust purification system of
Sampling means for acquiring an output value of the air-fuel ratio sensor at a predetermined sampling period;
A moving average value calculating means for calculating an average value over a predetermined moving average section of the output value acquired by the sampling means;
An injection amount calculating means for calculating a fuel injection amount of the injector based on the average value;
Intermittent injection execution means for driving the injector according to the injection period set to an integral multiple of the sampling period and the calculated fuel injection amount;
Injection cycle calculating means for calculating the sampling cycle multiplied by an injection cycle parameter that is a predetermined integer as the injection cycle of the intermittent injection;
An exhaust purification system for an internal combustion engine , comprising : an injection cycle parameter setting unit that changes the injection cycle parameter in a step-like manner according to a state of a vehicle on which the internal combustion engine is mounted . Before Symbol moving average value calculating means, an internal combustion engine exhaust gas purification system according to claim 1, characterized in that to calculate the average value of the injection cycle parameters as the number of taps corresponding to the moving average section. Estimating means for estimating the HC oxidation ability of the catalyst;
3. The exhaust gas purification system for an internal combustion engine according to claim 2, wherein the injection cycle parameter setting unit changes the injection cycle parameter in a step shape according to an estimation result by the estimation unit.

Images