Internal combustion engine control device and state quantity estimation method

Abstract

An internal combustion engine control device comprising: an exhaust gas flow rate calculation unit; an exhaust gas temperature detection unit; a catalyst upstream state quantity estimation unit that inputs a catalyst upstream air-fuel ratio, an exhaust gas flow rate, and an exhaust gas temperature and estimates a catalyst upstream NOx discharge quantity, a catalyst upstream HC discharge quantity, and a catalyst upstream CO discharge quantity as the catalyst upstream state quantities; a catalyst downstream state quantity estimation unit that inputs the catalyst upstream air-fuel ratio, exhaust gas flow rate, exhaust gas temperature, and catalyst upstream state quantities and estimates the catalyst temperature of a three-way catalyst, an oxygen storage capacity, a catalyst downstream air-fuel ratio, a catalyst downstream NOx discharge quantity, a catalyst downstream HC discharge quantity, and a catalyst downstream CO discharge quantity as catalyst downstream state quantities; and a correction unit that corrects the catalyst downstream NOx discharge quantity, catalyst downstream HC discharge quantity, and catalyst downstream CO discharge quantity by using the NOx sensor value detected by a downstream NOx sensor.

Description

[Technical Field]

[0001] The present invention relates to an internal combustion engine control device and a state quantity estimation method. [Background technology]

[0002] A control technology has been known in the past in which a three-way catalyst is provided in the exhaust pipe of an internal combustion engine, and exhaust gas sensors installed before and after the three-way catalyst are used to detect the oxygen storage state within the catalyst and correct the air-fuel ratio (or equivalence ratio) based on the detected oxygen storage state. In this control technology, a rich correction of the air-fuel ratio is determined based on the oxygen storage state detected by time integration of the product of the amount of air taken into the internal combustion engine and the difference between the air-fuel ratio of the exhaust gas and the stoichiometric air-fuel ratio. Furthermore, an exhaust gas sensor installed downstream of the three-way catalyst detects the amount of oxygen released downstream of the three-way catalyst, thereby performing feedback correction of the air-fuel ratio control. In the following description, the three-way catalyst will also be abbreviated as "catalyst." [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-71104 [Patent Document 2] Japanese Patent Publication No. 2022-81026 Summary of the Invention [Problem to be solved by the invention]

[0004] In recent years, automobile exhaust gas regulations have become stricter year by year, creating a demand for on-board monitoring (OBM) that monitors in real time within an ECU multiple regulated exhaust gas components (emissions) that fall under regulations on harmful substances emitted from internal combustion engines. To detect the deterioration of emissions due to the gradual deterioration of the catalyst in the exhaust pipe during vehicle use, OBM requires high-precision measurement of emissions downstream of the catalyst in real time and the presentation (indication) of the measured information to the vehicle user. One possible way to present (indicate) the measured information to the vehicle user is to turn on an abnormality lamp.

[0005] Among the above-mentioned emissions, real-time measurement of NOx, one of the regulated exhaust gas components, can be achieved using a nitrogen oxide (NOx) sensor for automobiles. However, real-time measurement of other regulated exhaust gas components, hydrocarbons (HC) and carbon monoxide (CO), has traditionally required very expensive and large exhaust analyzers, such as those typically used for research and development. However, it is practically impossible to install such expensive and large exhaust analyzers on automobiles used on public roads. In other words, it has been difficult with conventional technology to directly measure HC and CO in real time inside an automobile during normal use.

[0006] For example, the above-mentioned Patent Document 1 discloses a technology for estimating the purification rates of HC, CO, and NOx using a catalyst state estimation model that uses exhaust information measured by an exhaust gas sensor upstream of the catalyst as model input. However, Patent Document 1 does not disclose a configuration for measuring or estimating the NOx, HC, and CO emissions downstream of the catalyst in real time and displaying the NOx, HC, and CO emissions.

[0007] Furthermore, Patent Document 2 discloses a technology that uses a state space model to estimate the variation and catalyst state of an exhaust gas sensor installed upstream or downstream of a catalyst, and corrects the exhaust gas sensor output value based on the estimated variation or performs exhaust gas sensor degradation diagnosis based on the estimated variation. However, as can be easily understood by those skilled in the art, Patent Document 2 merely describes a technology for individually correcting or diagnosing the output values of various exhaust gas sensors actually installed upstream or downstream of a catalyst. Therefore, Patent Document 2 does not disclose a configuration for estimating HC and CO emissions downstream of a catalyst and displaying the estimated results to vehicle users without installing HC and CO sensors, which are too expensive and large to install in a vehicle, as described above. Overall, Patent Documents 1 and 2 do not disclose a configuration capable of performing so-called OBM with high accuracy, which measures and displays the downstream HC, CO, and NOx emissions of multiple regulated exhaust gas components in real time.

[0008] The present invention has been made in view of the above circumstances, and has as its object to display the amounts of HC, CO, and NOx emissions downstream of the catalyst in real time. [Means for solving the problem]

[0009] The internal combustion engine control device according to the present invention controls an internal combustion engine that is provided in an exhaust pipe, has an upstream air-fuel ratio sensor that is arranged upstream of a three-way catalyst having oxygen storage capacity and detects the air-fuel ratio of exhaust gas upstream of the catalyst, and has a downstream NOx sensor that is arranged downstream of the three-way catalyst and detects NOx downstream of the three-way catalyst. This internal combustion engine control device includes an exhaust gas flow rate calculation unit that calculates the exhaust gas flow rate of exhaust gas; a catalyst upstream state quantity estimation unit that receives as input the catalyst upstream air-fuel ratio, the exhaust gas flow rate, and the exhaust gas temperature detected by the exhaust gas temperature sensor, and estimates the catalyst upstream NOx emissions, the catalyst upstream HC emissions, and the catalyst upstream CO emissions as catalyst upstream state quantities; a catalyst downstream state quantity estimation unit that receives as input the catalyst upstream air-fuel ratio, the exhaust gas flow rate, the exhaust gas temperature, and the catalyst upstream state quantities, and estimates the catalyst temperature, oxygen storage capacity, the catalyst downstream air-fuel ratio, the catalyst downstream NOx emissions, the catalyst downstream HC emissions, and the catalyst downstream CO emissions as catalyst downstream state quantities; and a correction unit that corrects the catalyst downstream NOx emissions, the catalyst downstream HC emissions, and the catalyst downstream CO emissions using the NOx sensor value detected by the downstream NOx sensor. [Effects of the Invention]

[0010] According to the present invention, the NOx emissions downstream of the catalyst, the HC emissions downstream of the catalyst, and the CO emissions downstream of the catalyst are estimated and corrected, so that it is possible to display the corrected NOx emissions downstream of the catalyst, the HC emissions downstream of the catalyst, and the CO emissions downstream of the catalyst in real time without providing HC and CO sensors. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a schematic diagram illustrating the overall configuration of an internal combustion engine control system according to a first embodiment of the present invention. [Figure 2] 1 is a block diagram showing an example of the hardware configuration of an ECU according to a first embodiment of the present invention. [Figure 3A] 1 is a schematic diagram showing an example of the configuration of a post-processing system according to a first embodiment of the present invention. [Figure 3B] 3 is a diagram illustrating the relationship between the equivalence ratio of exhaust gas and the output of the air-fuel ratio sensor according to the first embodiment of the present invention. FIG. [Figure 3C]FIG. 3 is a diagram illustrating the relationship between the equivalence ratio of exhaust gas and the output of the rear oxygen sensor according to the first embodiment of the present invention. [Figure 4A] 1 is a diagram illustrating the tendency of the equivalence ratio of H2O (water), CO (carbon monoxide), CO2 (carbon dioxide), H2 (hydrogen), and O2 (oxygen) according to the first embodiment of the present invention. FIG. [Figure 4B] FIG. 2 is a diagram illustrating the tendency of HC (hydrocarbons) and NOx (nitrogen oxides) with respect to the equivalence ratio according to the first embodiment of the present invention. [Figure 5] FIG. 2 is a diagram illustrating the main reaction process of a three-way catalyst (ceria-based) used in the aftertreatment system according to the first embodiment of the present invention. [Figure 6] FIG. 3 is a diagram illustrating the tendency of the purification rate of a three-way catalyst with respect to the exhaust gas equivalence ratio at temperatures equal to or higher than the catalyst activation temperature according to the first embodiment of the present invention. [Figure 7] 3 is a diagram illustrating the catalyst upstream equivalence ratio, the catalyst downstream equivalence ratio, and the output behavior of the rear oxygen sensor 22 according to the first embodiment of the present invention. FIG. [Figure 8] FIG. 3 is a diagram illustrating the hysteresis of the output characteristics of the rear oxygen sensor according to the first embodiment of the present invention. [Figure 9] 3 is a diagram illustrating the output of the rear oxygen sensor and the changes over time in the NOx concentration and HC concentration downstream of the catalyst according to the first embodiment of the present invention. FIG. [Figure 10] FIG. 2 is a diagram illustrating the relationship between the degree of catalyst deterioration and the oxygen storage capacity of a three-way catalyst according to the first embodiment of the present invention. [Figure 11] FIG. 2 is a diagram illustrating the relationship between the oxygen storage ratio of the three-way catalyst according to the first embodiment of the present invention and the NOx purification rate of the three-way catalyst. [Figure 12] FIG. 10 is a diagram illustrating the results of a comparison of the catalyst upstream equivalence ratio, catalyst downstream equivalence ratio, oxygen storage ratio, and output behavior of the rear oxygen sensor when a new catalyst and a deteriorated catalyst according to the first embodiment of the present invention are used. [Figure 13] 1 is a block diagram showing an example of the internal configuration of an ECU according to a first embodiment of the present invention. [Figure 14]4 is a flowchart showing an example of processing by an ECU according to the first embodiment of the present invention. [Figure 15] 3 is a flowchart illustrating an example of a Kalman filter algorithm according to the first embodiment of the present invention. [Figure 16A] FIG. 2 is a diagram for explaining a first machine learning model constituting a first state space model and a second machine learning model constituting a second state space model according to the first embodiment of the present invention. [Figure 16B] FIG. 2 is a diagram illustrating an example of a neural network model according to the first embodiment of the present invention. [Figure 17] FIG. 2 is a block diagram of a first and second machine learning model for explaining a machine learning process according to the first embodiment of the present invention. [Figure 18] 4 is a flowchart of a first learning process and a second learning process according to the first embodiment of the present invention. [Figure 19] FIG. 10 is a block diagram showing an example of the internal configuration of an ECU according to a second embodiment of the present invention. [Figure 20] 10 is a flowchart showing an example of processing by an ECU according to a second embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In this specification and drawings, components having substantially the same functions or configurations are designated by the same reference numerals, and redundant description will be omitted.

[0013] [First embodiment] FIG. 1 is a schematic diagram of the overall configuration of an internal combustion engine control system 100. As shown in FIG.

[0014] The internal combustion engine control system 100 includes an internal combustion engine 1 and an ECU (Electronic Control Unit) 28 attached to the internal combustion engine 1. The ECU 28 is electrically connected to various sensors, actuators, etc. that constitute the internal combustion engine 1, and is an example of an internal combustion engine control device that controls the internal combustion engine 1.

[0015] The internal combustion engine 1 includes a flow sensor 2 (air flow sensor), a turbocharger 3, an air bypass valve 4, an intercooler 5, a boost temperature sensor 6, a throttle valve 7, an intake manifold 8, a boost pressure sensor 9, a flow enhancement valve 10, an intake valve 11, a valve opening / closing phase sensor 12, an exhaust valve 13, a valve opening / closing phase sensor 14, a fuel injector 15, a spark plug 16, a knock sensor 17, a crank angle sensor 18, a wastegate valve 19, a catalyst upstream air-fuel ratio sensor 20, an exhaust purification catalyst (three-way catalyst) 21, a rear oxygen sensor (catalyst downstream oxygen sensor) 22, an EGR (Exhausted Gas Recirculation) pipe 23, an EGR cooler 24, an EGR valve 25, an exhaust gas temperature sensor 26, a differential pressure sensor 27, and a catalyst downstream NOx sensor 29. The rear oxygen sensor 22 can be replaced with a catalyst downstream (rear) air-fuel ratio sensor (not shown).

[0016] An intake air temperature sensor (not shown) is attached to the flow rate sensor 2 provided in the intake air flow path of the internal combustion engine 1. The intake air temperature sensor measures the intake air temperature. The turbocharger 3 is composed of a compressor 3a having compressor blades facing the intake passage, and a turbine 3b connected to the compressor 3a so as to rotate integrally with the compressor 3a and having turbine blades facing the exhaust passage. The compressor 3a and turbine 3b are rotatably supported within the turbocharger 3. The turbine 3b converts the energy of the exhaust gas from the internal combustion engine 1 into rotational energy. The compressor 3a connected to the turbine 3b compresses the intake air that flows in from the intake passage using the rotational energy of the turbine 3b.

[0017] In the intake air flow path of the internal combustion engine 1, the intercooler 5 is provided downstream of the compressor 3a of the turbocharger 3, and cools the intake air temperature that has been increased by adiabatic compression by the compressor 3a. The supercharger temperature sensor 6 is mounted downstream of the intercooler 5, and measures the temperature of the intake air cooled by the intercooler 5 (supercharger temperature).

[0018] The throttle valve 7 is provided downstream of the intercooler 5 and narrows the intake flow path to control the amount of intake air flowing into the cylinders of the internal combustion engine 1. The throttle valve 7 is configured as an electronically controlled butterfly valve whose valve opening can be controlled by the ECU 28. Downstream of the throttle valve 7, an intake manifold 8 to which a boost pressure sensor 9 is attached is connected.

[0019] The intake manifold 8 provided downstream of the throttle valve 7 may be integrated with the intercooler 5. In this case, the volume from downstream of the compressor 3a to the cylinder can be reduced, which improves acceleration / deceleration response and controllability.

[0020] The air bypass valve 4 is disposed in the intake bypass passage that connects the upstream and downstream of the compressor 3a in the intake passage to prevent an excessive increase in pressure from downstream of the compressor 3a to the upstream of the throttle valve 7. For example, if the throttle valve 7 is suddenly closed in a supercharging state, the air bypass valve 4 opens under the control of the ECU 28, causing the compressed intake air downstream of the compressor 3a to flow back through the bypass passage to the upstream of the compressor 3a. As a result, the supercharging pressure is immediately reduced, preventing a phenomenon known as surging and appropriately preventing damage to the compressor 3a.

[0021] The flow enhancement valve 10 is located downstream of the intake manifold 8 and enhances the turbulence of the flow inside the cylinder by causing a bias in the intake air drawn into the cylinder. When exhaust gas recirculation combustion, which will be described later, is performed, the ECU 28 can close the flow enhancement valve 10 to promote turbulent combustion and stabilize the combustion that occurs in the combustion chamber.

[0022] The intake valve 11 and the exhaust valve 13 each have a variable valve mechanism for continuously varying the valve opening / closing phase. The variable valve mechanisms of the intake valve 11 and the exhaust valve 13 are respectively equipped with valve opening / closing phase sensors 12, 14 for detecting the valve opening / closing phase. Each cylinder of the internal combustion engine 1 is provided with a direct injection fuel injection valve 15 that injects fuel pressurized by a fuel pump (not shown) into the combustion chamber of the cylinder. Note that the fuel injection valve 15 may be a port injection valve that injects fuel into the intake passage. Alternatively, a plurality of direct injection and port injection fuel injection valves 15 may be used for each cylinder.

[0023] Spark plugs 16, which have electrodes exposed inside the cylinders and generate sparks to ignite a combustible mixture, are installed in the cylinders of the internal combustion engine 1. Knock sensors 17 are installed in the cylinder block and detect the presence or absence of knock by detecting cylinder block vibrations caused by combustion pressure vibrations generated in the combustion chamber. Crank angle sensor 18 is installed on the crankshaft and outputs a signal corresponding to the rotation angle of the crankshaft to the ECU 28.

[0024] In the exhaust pipe of the internal combustion engine 1, an upstream air-fuel ratio sensor (catalyst upstream air-fuel ratio sensor 20) is disposed downstream of the turbine 3b of the turbocharger 3 and upstream of a three-way catalyst (exhaust purification catalyst 21) having oxygen storage capacity that is provided in the exhaust pipe. The upstream air-fuel ratio sensor (catalyst upstream air-fuel ratio sensor 20) detects the exhaust gas composition detected from the exhaust gas, i.e., the catalyst upstream air-fuel ratio of the exhaust gas, and outputs a signal indicating the catalyst upstream air-fuel ratio to the ECU 28. The exhaust purification catalyst 21 is an example of a three-way catalyst and is provided downstream of the catalyst upstream air-fuel ratio sensor 20. The exhaust purification catalyst 21 purifies harmful exhaust gas components such as carbon monoxide, nitrogen compounds, and unburned hydrocarbons in the exhaust gas through catalytic reactions. As described above, a rear oxygen sensor 22 is disposed downstream of the exhaust purification catalyst 21. The rear oxygen sensor 22 detects the amount of oxygen contained in the exhaust gas after purification by the exhaust purification catalyst 21. Hereinafter, the exhaust purification catalyst 21 will also be referred to as a three-way catalyst or a catalyst.

[0025] The turbocharger 3 is equipped with an air bypass valve 4 and a wastegate valve 19. The air bypass valve 4 is arranged in a bypass flow path connecting the upstream and downstream of the compressor 3a to prevent an excessive increase in pressure from the downstream of the compressor 3a to the upstream of the throttle valve 7. If the throttle valve 7 is suddenly closed in a supercharging state, the air bypass valve 4 is opened under the control of the ECU 28, causing the compressed intake air downstream of the compressor 3a to flow back through the bypass flow path to the upstream of the compressor 3a. As a result, the supercharging pressure is immediately reduced, preventing a phenomenon known as surging and appropriately preventing damage to the compressor 3a.

[0026] The wastegate valve 19 is disposed in an exhaust bypass passage that connects the upstream and downstream of the turbine 3b in the exhaust passage. The wastegate valve 19 is an electrically operated valve whose valve opening can be freely controlled according to the boost pressure under the control of the ECU 28. When the ECU 28 adjusts the opening of the wastegate valve 19 based on the boost pressure detected by the boost pressure sensor 9, part of the exhaust gas passes through the bypass passage, thereby reducing the energy imparted by the exhaust gas to the turbine 3b. As a result, the wastegate valve 19 can adjust the boost pressure to the target pressure.

[0027] The EGR pipe 23 connects the exhaust flow path downstream of the exhaust purification catalyst 21 with the intake flow path upstream of the compressor 3a, and diverts exhaust gas from downstream of the exhaust purification catalyst 21 and recirculates it to the upstream of the compressor 3a. An EGR cooler 24 provided in the EGR pipe 23 cools the diverted exhaust gas. An EGR valve 25 is provided in the EGR pipe 23 between the EGR cooler 24 and the upstream of the compressor 3a, and controls the flow rate of exhaust gas recirculated to the upstream of the compressor 3a. The EGR pipe 23 is also provided with an exhaust gas temperature sensor 26 that detects the exhaust gas temperature of the exhaust gas upstream of the EGR valve 25, and a differential pressure sensor 27 that detects the differential pressure between the upstream and downstream of the EGR valve 25.

[0028] The ECU 28 is a control device that includes a CPU (Central Processing Unit) 205, a ROM (Read Only Memory) 204, a RAM (Random Access Memory) 203 (described later) shown in FIG. 2 , as well as an A / D (Analog-to-Digital) converter, a driver circuit, and the like (not shown), and controls the various components of the internal combustion engine 1 and executes various data processing. The ECU 28 controls the operation of actuators such as the throttle valve 7, the fuel injection valve 15, the variable valve mechanisms of the intake valve 11 and the exhaust valve 13, and the EGR valve 25. The ECU 28 also detects the operating state of the internal combustion engine 1 based on signals input from various sensors, and ignites the spark plug 16 at a timing determined according to the operating state.

[0029] [ECU hardware configuration example] FIG. 2 is a block diagram showing an example of the hardware configuration of the ECU 28.

[0030] The ECU 28 includes an input circuit 201 , an input / output port 202 , a RAM 203 , a ROM 204 , a CPU 205 , a throttle valve drive circuit 206 , a fuel injection valve drive circuit 207 , and an ignition output circuit 208 .

[0031] 2 shows an example in which output signals from the throttle sensor of the throttle valve 7, the flow rate sensor 2, the supercharging temperature sensor 6, the supercharging pressure sensor 9, the valve opening / closing phase sensors 12 and 14, the knock sensor 17, the crank angle sensor 18, the catalyst upstream air-fuel ratio sensor 20, and the rear oxygen sensor 22 are input to the input circuit 201 of the ECU 28. The signals input to the input circuit 201 are sent to an input / output port 202.

[0032] The signal transmitted to the input / output port 202 is stored in the RAM 203 and is processed by the CPU 205. A control program describing the details of the processing is written in advance in the ROM 204 and executed by the CPU 205. The ROM 204 stores programs, data, etc. necessary for the operation of the CPU 205, and is used as an example of a computer-readable non-transitory storage medium that stores a program executed by the ECU 28.

[0033] The control signals calculated by the CPU 205 in accordance with the control program are output to various devices such as a throttle valve drive circuit 206, a fuel injection valve drive circuit 207, and an ignition output circuit 208. The throttle valve drive circuit 206 outputs a drive signal to the throttle valve 7 for controlling the opening and closing of the throttle valve 7 . The fuel injection valve drive circuit 207 outputs a drive signal to the fuel injection valve 15 for controlling the opening and closing of the fuel injection valve 15 at the fuel injection timing. The ignition output circuit 208 outputs a drive signal to the ignition output circuit 208 to control the ignition of the spark plug 16 at the ignition timing.

[0034] <Example of aftertreatment system configuration> Next, a configuration example of a well-known aftertreatment system 110 that purifies exhaust gas from an internal combustion engine will be described with reference to FIGS. 3A to 3C.

[0035] FIG. 3A is a schematic diagram showing an example of the configuration of the post-processing system 110. As shown in FIG. As described above, in the aftertreatment system 110, a three-way catalyst is used as an exhaust gas purification catalyst (exhaust purification catalyst 21). A catalyst upstream air-fuel ratio sensor 20 is provided upstream of the three-way catalyst, and a rear oxygen sensor 22 is provided downstream, and the catalyst upstream air-fuel ratio sensor 20 and the rear oxygen sensor 22 are connected to a control device such as an ECU 28. The ECU 28 measures the catalyst upstream air-fuel ratio of the exhaust gas flowing into the three-way catalyst via the catalyst upstream air-fuel ratio sensor 20, and can detect the amount of oxygen contained in the exhaust gas after purification by the catalyst via the rear oxygen sensor 22.

[0036] 3B is a diagram illustrating the relationship between the equivalence ratio of exhaust gas (=stoichiometric air-fuel ratio / air-fuel ratio) and the output of the catalyst upstream air-fuel ratio sensor 20. The horizontal axis of FIG. 3B represents the equivalence ratio, and the vertical axis represents the output of the catalyst upstream air-fuel ratio sensor 20.

[0037] 3B, the output (air-fuel ratio sensor output) of the catalyst upstream air-fuel ratio sensor 20 tends to decrease as the equivalence ratio increases (in other words, as the exhaust gas becomes richer). The ECU 28 converts the catalyst upstream air-fuel ratio sensor signal into a catalyst upstream air-fuel ratio based on the relationship between the equivalence ratio of the exhaust gas and the output of the catalyst upstream air-fuel ratio sensor 20. This enables the ECU 28 to accurately detect the catalyst upstream air-fuel ratio over a wide range, from lean to rich exhaust gas conditions.

[0038] 3C is a diagram illustrating the relationship between the equivalence ratio of exhaust gas and the output of the rear oxygen sensor 22. The horizontal axis of FIG. 3C represents the equivalence ratio, and the vertical axis represents the rear oxygen sensor output.

[0039] The rear oxygen sensor output, also known as the rear oxygen sensor voltage, varies depending on the electromotive force caused by the difference between the oxygen concentration in the exhaust gas and the oxygen concentration in the air. The rear oxygen sensor output exhibits minimal electromotive force under lean conditions and maximizes electromotive force under rich conditions. Therefore, in catalytic control, the rear oxygen sensor output exhibits a sudden change at the stoichiometric air-fuel ratio (equivalent ratio of 1.0). By detecting the timing of this change in the rear oxygen sensor output, the ECU 28 can detect the amount of oxygen in the exhaust gas released downstream of the three-way catalyst.

[0040] <Concentration of chemical species in exhaust gas> 4A and 4B are diagrams illustrating the trends of the concentrations of chemical species in the exhaust gas with respect to the equivalence ratio. FIG. 4A is a diagram illustrating the trend for the equivalence ratio of H2O (water), CO (carbon monoxide), CO2 (carbon dioxide), H2 (hydrogen), and O2 (oxygen). 4A and 4B are graphs illustrating the trends of HC (hydrocarbons) and NOx (nitrogen oxides) relative to the equivalence ratio. The horizontal axis of each graph represents the equivalence ratio, and the vertical axis represents the concentration of the chemical species in the exhaust gas.

[0041] As shown in Figure 4A, the combustion gas composition of hydrocarbon fuels tends to increase in CO (carbon monoxide) and H2 (hydrogen) on the rich side of the stoichiometric air-fuel ratio, and increase in O2 (oxygen) on the lean side. On the other hand, as shown in Figure 4B, NOx (nitrogen oxides) reaches a maximum value slightly leaner than the stoichiometric air-fuel ratio and tends to decrease on the leaner and richer sides. HC (unburned hydrocarbons) are unburned fuel components that are emitted without being burned, and reach a minimum value at the stoichiometric air-fuel ratio. When the air-fuel ratio is excessively lean or richer than the stoichiometric air-fuel ratio, the amount of HC emitted due to incomplete combustion of the fuel tends to increase.

[0042] 4A and 4B, even under theoretical air-fuel ratio conditions where fuel and air (oxygen) are supplied in the right amounts, a certain amount of CO (carbon monoxide) and NOx (nitrogen oxides) is emitted in the high-temperature combustion gas without converting to HO (water) or CO2 (carbon dioxide). For this reason, the exhaust gas needs to be properly purified by the aftertreatment system 110.

[0043] <Reaction process of three-way catalyst> 5 is a diagram illustrating the main reaction process of the three-way catalyst (ceria-based) used in the aftertreatment system 110. In the reaction process shown in FIG. 5, coefficients are omitted.

[0044] The three-way catalytic reaction process mainly consists of an oxidation reaction, a NOx reduction reaction, and an oxygen storage and release reaction. In the oxidation reaction, CO, H2, and HC, which are produced under rich or high-temperature conditions, react with oxygen to produce harmless CO2 and H2O. Unburned hydrocarbons (HC) include components such as methane, propane, ethylene, and butane, and each reaction proceeds at a different rate.

[0045] The NOx reduction reaction is mainly represented by the reaction between CO and NO, producing CO2 and N2. In the oxygen storage and release reaction, the storage and release of oxygen (O2) and the oxidation and reduction reactions of HC, CO, and NO proceed via the catalytic material Ce (cerium). That is, the reaction between cerium dioxide (CeO2) and CO and HC produces CO2 and H2O, while the reaction between cerium trioxide (Ce2O3) and NO produces N2. The oxygen storage ratio Ψ of the three-way catalyst is determined by the balance between the CeO2 and Ce2O3 that are produced at the same time. In other words, if all the Ce2O3 in the catalyst becomes CeO2, it cannot react with NO, and NO cannot be purified.

[0046] Thus, in order to maintain an appropriate purification rate of the three-way catalyst, it is necessary to maintain the balance between CeO2 and Ce2O3, i.e., the oxygen storage ratio Ψ, at a predetermined value. Since all of the above-mentioned reaction processes are strongly dependent on the catalyst temperature, the ECU 28 needs to properly manage the catalyst temperature so that the temperature reaches or exceeds the activation temperature as soon as possible after the internal combustion engine 1 starts.

[0047] Although the system shown in this embodiment is configured to use a ceria-based three-way catalyst, the present invention is not limited to this. Even with catalysts using other materials that exhibit similar effects, the same effects can be achieved without changing the configuration of the present invention by adjusting the constants of the control model. In addition to the reaction mechanism shown in FIG. 5, the catalytic reaction may involve a water-gas shift reaction or the like. The ECU 28 can also accommodate these reaction mechanisms by adjusting the control model constants.

[0048] Figure 6 is a diagram illustrating the tendency of the purification rate of a three-way catalyst versus the exhaust gas equivalence ratio at temperatures above the catalyst activation temperature. The horizontal axis of Figure 6 represents the equivalence ratio, and the vertical axis represents the catalyst purification rate. Figure 6 also shows the catalyst purification rates for NOx, HC, and CO when the equivalence ratio changes from lean to rich. The closer the catalyst purification rate is to 100%, the more the components in the exhaust gas are purified, and the less the components in the exhaust gas are emitted.

[0049] This graph shows that the purification rate characteristics of the three-way catalyst change around the stoichiometric air-fuel ratio (the "control target" in the graph). Under lean conditions, the purification rates for CO and HC are maintained at approximately 90% or higher, while the NOx purification rate decreases as the equivalence ratio decreases. On the rich side, the purification rates for HC and CO tend to decrease as the equivalence ratio increases. Near the stoichiometric air-fuel ratio, purification rates for NOx, HC, and CO can all be achieved at 90% or higher. Therefore, the point near the stoichiometric air-fuel ratio is called the "three-way point." The ECU 28 controls the purification rate of the three-way catalyst to maintain a high level by maintaining the equivalence ratio at the stoichiometric air-fuel ratio, which is the three-way point.

[0050] Figure 7 is a diagram illustrating the behavior of the catalyst upstream equivalence ratio, catalyst downstream equivalence ratio, and output of the rear oxygen sensor 22. The horizontal axis of Figure 7 represents time, and the vertical axis represents the catalyst upstream equivalence ratio, catalyst downstream equivalence ratio, and rear oxygen sensor output. This diagram shows the behavior of the catalyst downstream equivalence ratio (catalyst downstream air-fuel ratio) and the output of the rear oxygen sensor 22 installed downstream of the catalyst when the catalyst upstream equivalence ratio (catalyst upstream air-fuel ratio) is suddenly changed in steps (stepwise) over time to the lean side or rich side, with the equivalence ratio at 1.0 as the center.

[0051] First, as shown in time period (a) along the horizontal time axis in Figure 7, when the catalyst upstream equivalence ratio is set to exactly 1.0, a very small amount of oxygen is discharged downstream of the catalyst, and the rear oxygen sensor output is maintained at an intermediate value between the maximum and minimum electromotive forces.

[0052] Next, as shown in time period (b) in Figure 7, when the catalyst upstream equivalence ratio is decreased from an equivalence ratio of 1.0 to change it in a stepwise manner toward the lean side, the catalyst downstream equivalence ratio gradually decreases toward the lean side, and the rear oxygen sensor output suddenly changes toward the predetermined minimum electromotive force value under lean conditions after a relaxation time required for the gradual decrease of the catalyst downstream equivalence ratio has passed.

[0053] Next, as shown in time period (c) in Figure 7, when the equivalence ratio upstream of the catalyst is increased above 1.0 and shifted in a stepwise manner toward the rich side, the equivalence ratio downstream of the catalyst begins to gradually increase toward the rich side in response. After a relaxation time required for the gradual increase in the equivalence ratio downstream of the catalyst has elapsed, the output of the rear oxygen sensor changes toward the predetermined maximum electromotive force value under rich conditions.

[0054] Next, as shown in time period (d) in Figure 7, when the equivalence ratio upstream of the catalyst is again decreased from 1.0 to a lean-air state in a stepwise manner, the equivalence ratio downstream of the catalyst gradually decreases again to a lean-air state. The rear oxygen sensor output responding to the change in the equivalence ratio upstream of the catalyst changes to the minimum electromotive force value under lean conditions after a relaxation time required for the equivalence ratio downstream of the catalyst to gradually decrease. The time required for the equivalence ratio downstream of the catalyst to change from rich to lean during time period (d) is shorter than the time required for the change from lean to rich during time period (c), which is characterized by a so-called hysteresis characteristic.

[0055] As described above, the delay time of the catalyst downstream equivalence ratio and the delay time of the output of the rear oxygen sensor 22 tend to differ when the mixture is changed from lean to rich and when it is changed from rich to lean. This tendency is due to the fact that the reaction rate of the oxygen storage and release reaction between CeO2 and Ce2O3 in the three-way catalyst (ceria-based) described in Figure 5 differs when the mixture is changed from rich to lean and when it is changed from lean to rich. In addition, since the reaction rate of the three-way catalyst also depends on the catalyst temperature and exhaust gas flow rate, the hysteresis characteristic described above changes depending on the catalyst temperature and exhaust gas flow rate.

[0056] Next, the hysteresis of the oxygen sensor characteristics will be described with reference to FIG. 8 is a graph illustrating the hysteresis of the output characteristics of an oxygen sensor. In this graph, the horizontal axis represents the equivalence ratio, and the vertical axis represents the output of the oxygen sensor.

[0057] The static characteristics of the oxygen sensor have been described with reference to FIG. 3C. Meanwhile, the oxygen sensor also uses the catalytic material described above. Therefore, the dynamic characteristics of the oxygen sensor have hysteresis, which means that the transition from rich to lean differs from the transition from lean to rich. That is, the relaxation time required for the oxygen sensor signal to change from the predetermined maximum electromotive force value under rich conditions to the predetermined minimum electromotive force value under lean conditions is shorter than the relaxation time required for the oxygen sensor signal to change from the predetermined minimum electromotive force value under lean conditions to the predetermined maximum electromotive force value under rich conditions. Furthermore, the behavior of the oxygen sensor signal described above is affected by changes in the properties of the materials that make up the oxygen sensor over time and by the temperature of the oxygen sensor.

[0058] Here, referring also to Figure 9, we will explain the changes over time in the output of the rear oxygen sensor 22 and the NOx concentration downstream of the catalyst when the internal combustion engine 1 is controlled from a state (stoichiometry control) where the equivalence ratio upstream of the catalyst is 1.0 (stoichiometry), goes through a period where fuel cut is performed (lean condition), and then returns to operation under the stoichiometric air-fuel ratio condition. 9 is a diagram illustrating the change over time in the output of the rear oxygen sensor 22 and the NOx and HC concentrations downstream of the catalyst. The horizontal axis of Fig. 9 represents time, and the vertical axis represents the catalyst upstream equivalence ratio, rear oxygen sensor output, NOx concentration downstream of the catalyst, and HC concentration downstream of the catalyst.

[0059] First, when the catalyst upstream equivalence ratio is set to exactly 1.0 as shown in time period (a) along the horizontal time axis in FIG. 9, the rear oxygen sensor output is maintained at an intermediate state between the maximum and minimum electromotive forces.

[0060] Next, as shown in time period (b) in FIG. 9, the operation transitions from the catalyst upstream equivalence ratio of 1.0 to a lean condition due to fuel cut. Next, as shown in time period (c) in Figure 9, the system returns to operation with the catalyst upstream equivalence ratio at 1.0. At this time, the change in the rear oxygen sensor output from the intermediate state with a catalyst upstream equivalence ratio of 1.0 to the lean side differs from the change (relaxation time) from the lean side to the intermediate state with a catalyst upstream equivalence ratio of 1.0.

[0061] In other words, the relaxation time required for the oxygen sensor signal to change from the predetermined minimum electromotive force value under lean conditions due to fuel cut to the intermediate state under stoichiometry control conditions (see time period (a) in Figure 7) is longer than the relaxation time required for the signal to change from the intermediate state under stoichiometry control conditions to the predetermined minimum electromotive force value under lean conditions. During this relaxation time (delay period) until the rear oxygen sensor output returns to the intermediate state, a spike-like increase in the NOx concentration downstream of the catalyst may be observed.

[0062] To address this increase in the NOx concentration downstream of the catalyst, for example, when restarting operation with a catalyst upstream equivalence ratio of 1.0 after a fuel cut, the catalyst upstream equivalence ratio is increased above 1.0 to temporarily increase the NOx catalytic conversion rate (see FIG. 6), thereby reducing NOx emissions downstream of the catalyst. However, there is a trade-off between increasing the NOx catalytic conversion rate through rich correction and the catalytic conversion rates of HC and CO (see FIG. 6). Therefore, in air-fuel ratio control for the internal combustion engine 1, it is necessary to perform rich correction control with an appropriate correction amount and period, taking into account the internal state of the catalyst. To consider the internal state of the catalyst in air-fuel ratio control for the internal combustion engine 1, for example, a rear oxygen sensor signal detecting the oxygen concentration in the exhaust gas downstream of the catalyst can be used as a judgment criterion. However, as mentioned above, it is important to note that there is a delay period in the response of the rear oxygen sensor 22 (see FIG. 8).

[0063] Here, the output of the rear oxygen sensor 22 and the changes over time in the NOx concentration and HC concentration downstream of the catalyst will be explained in the order of stoichiometry control, appropriate rich correction, and excessive rich correction.

[0064] (Stoichiometry control) The solid lines show the output of the rear oxygen sensor 22 and the changes over time in the NOx and HC concentrations downstream of the catalyst when stoichiometry control is performed. As shown in Figure 7, the output of the rear oxygen sensor 22 increases with a delay when the engine changes from a lean to a rich condition. The NOx concentration downstream of the catalyst exhibits a behavior in which a momentary increase in NOx occurs during the delay period until the output of the rear oxygen sensor 22 returns to normal, resulting in the emission of HC. On the other hand, the HC concentration downstream of the catalyst remains almost unchanged, preventing the emission of HC.

[0065] (appropriate rich correction) The dashed lines show the output of the rear oxygen sensor 22 and the changes over time in the NOx and HC concentrations downstream of the catalyst when appropriate rich correction is performed. When firing operation is restarted after a fuel cut, once rich correction is performed, the equivalence ratio upstream of the catalyst becomes rich. The output of the rear oxygen sensor 22 reaches the equivalence ratio more quickly than under stoichiometry control. Furthermore, because the NOx and HC concentrations downstream of the catalyst remain almost unchanged, NOx and HC emissions are prevented.

[0066] (Excessive rich correction) The rough dashed lines show the output of the rear oxygen sensor 22 and the changes over time in the NOx and HC concentrations downstream of the catalyst when excessive rich correction is made. The rear oxygen sensor 22 detects the oxygen content of the gas downstream of the catalyst. Therefore, when excessive rich correction is made, the internal state of the catalyst has already changed to the maximum or minimum oxygen storage state by the time the output of the rear oxygen sensor 22 reacts. Furthermore, when excessive rich correction is made, the NOx concentration decreases as shown in Figure 4B, and the catalyst purification rate of NOx increases as shown in Figure 6, preventing NOx emissions.

[0067] On the other hand, the HC concentration increases as shown in Figure 4B, and the catalyst purification rate of HC decreases as shown in Figure 6, resulting in the emission of HC. In other words, if a control method is adopted in which the rich correction is stopped after the output of the rear oxygen sensor 22 reacts, the timing of stopping the rich correction is too late for the catalyst, and HC cannot be appropriately prevented. For this reason, in air-fuel ratio control for an internal combustion engine, it is necessary to perform rich correction control for an appropriate period of time, taking into account the state inside the catalyst, which cannot be directly observed from the outside.

[0068] 10 is a diagram illustrating the relationship between the degree of catalyst deterioration and the oxygen storage capacity (OSC) of a three-way catalyst, where the horizontal axis represents the degree of catalyst deterioration and the vertical axis represents the oxygen storage capacity.

[0069] Catalyst degradation refers to a state in which the catalytic activity of a three-way catalyst is reduced due to thermal influences and poisoning by sulfur contained in the fuel. As shown in Figure 10, the oxygen storage capacity of a three-way catalyst is roughly proportional to the degree of catalyst degradation in the left-right direction in Figure 10. In other words, as catalyst degradation progresses, the oxygen storage capacity of the three-way catalyst tends to decrease.

[0070] The following describes the effect that changes in oxygen storage capacity have on the purification function of a three-way catalyst. Figure 11 is a diagram illustrating the relationship between the oxygen storage ratio of a three-way catalyst, i.e., the value obtained by dividing the current oxygen storage amount of the three-way catalyst by the above-mentioned oxygen storage capacity, and the NOx purification rate. The horizontal axis of Figure 11 represents the oxygen storage ratio, and the vertical axis represents the NOx purification rate. In Figure 11, the solid line represents the change in the NOx purification rate of a new three-way catalyst, and the dashed line represents the change in the NOx purification rate of a deteriorated three-way catalyst.

[0071] When the oxygen storage ratio is low, both new and deteriorated three-way catalysts have high oxygen storage ratios, and the new three-way catalyst has a higher NOx reduction rate than the deteriorated three-way catalyst.

[0072] As shown in Figure 11, when the oxygen storage ratio exceeds a predetermined value, the NOx purification rate tends to deteriorate rapidly for both new and deteriorated three-way catalysts. This is because, as explained with reference to Figure 5, Ce2O3 in the catalyst is important for NOx purification, and if all Ce2O3 reacts and turns into CeO2, Ce2O3 cannot react with NO and cannot purify NOx. Furthermore, as catalyst deterioration progresses and the oxygen storage capacity of the three-way catalyst decreases, the oxygen storage ratio for the same amount of oxygen stored increases.

[0073] For this reason, the NOx purification rate for a deteriorated three-way catalyst, shown by the dashed line in Figure 11, is lower than the NOx purification rate for a new three-way catalyst, shown by the solid line in Figure 11. Therefore, to maintain a high NOx purification rate, it is necessary not only to maintain the exhaust gas air-fuel ratio at the catalyst inlet at the three-way point, but also to appropriately correct and control the exhaust gas air-fuel ratio at the catalyst inlet so that the oxygen storage rate falls within a range that achieves a predetermined NOx purification rate, taking into account the internal state of the catalyst, which cannot be directly observed from the outside, i.e., the current oxygen storage capacity and oxygen storage rate of the three-way catalyst. Note that the "predetermined range of oxygen storage rate" is represented by a deteriorated control range 36 when the catalyst is deteriorated and a new control range 35 when the catalyst is new, as shown in Figure 11. The new control range 35 is wider than the deteriorated control range 36.

[0074] Figure 12 is a diagram illustrating the results of a comparison of the catalyst upstream equivalence ratio, catalyst downstream equivalence ratio, oxygen storage fraction, and output behavior of the rear oxygen sensor 22 when a new catalyst and a degraded catalyst are used. In Figure 12, the horizontal axis represents time, and the vertical axis represents the catalyst upstream equivalence ratio, catalyst downstream equivalence ratio, oxygen storage fraction, and rear oxygen sensor output. Figure 12 shows the results of a comparison of the output behavior of the rear oxygen sensor 22 installed downstream of the catalyst when the air-fuel ratio is varied over time in steps toward the lean side and the rich side, with an equivalence ratio of 1.0 as the center, for a new catalyst and a degraded catalyst. Note that the dashed line graph in the figure represents a degraded catalyst, and the solid line graph represents a new catalyst.

[0075] As shown in Figure 12, a deteriorated catalyst exhibits a reduced delay in the output behavior of the rear oxygen sensor 22 in response to lean and rich changes in the catalyst downstream equivalence ratio compared to a new catalyst. This can be explained by the change in the catalyst's oxygen storage ratio over time. That is, as the catalyst's oxygen storage capacity decreases due to deterioration, the oxygen storage ratio saturates to its maximum or minimum value more quickly, thereby reducing the period during which the catalyst reacts to store and release oxygen in exhaust gas downstream of the catalyst. Therefore, the rich correction period after returning from a fuel cut, as described with reference to Figure 9, must be set taking into account the deteriorated state of the catalyst.

[0076] 13 is a block diagram showing an example of the internal configuration of the ECU 28 according to the present invention. The ECU 28 includes the following blocks 51 to 56.

[0077] The exhaust gas flow rate calculation unit (exhaust gas flow rate calculation unit 51) receives the engine speed based on the crank angle sensor 18 and the charging efficiency based on the flow rate sensor 2 as inputs, and calculates the flow rate of the exhaust gas.

[0078] The catalyst upstream state quantity estimation unit 52 estimates state quantities (catalyst upstream NOx emissions, catalyst upstream HC emissions, and catalyst upstream CO emissions) of various substances emitted upstream of the three-way catalyst. In the following description, the catalyst upstream NOx emissions, catalyst upstream HC emissions, and catalyst upstream CO emissions are also abbreviated as "catalyst upstream NOx, HC, and CO emissions." This catalyst upstream state quantity estimation unit (catalyst upstream state quantity estimation unit 52) receives as input the catalyst upstream air-fuel ratio detected by the catalyst upstream air-fuel ratio sensor 20, the exhaust gas flow rate, and the exhaust gas temperature detected by the exhaust gas temperature sensor 26, and estimates the catalyst upstream NOx, HC, and CO emissions as catalyst upstream state quantities. This catalyst upstream state quantity estimation unit 52 is configured using a first state space model. This first state space model is configured, for example, using a first machine learning model (neural network model) described later. Then, the catalyst upstream state quantity estimation unit (catalyst upstream state quantity estimation unit 52) inputs the catalyst upstream air-fuel ratio, exhaust gas flow rate, and exhaust gas temperature into a first machine learning model that has been trained in advance, and estimates the catalyst upstream NOx, HC, and CO emissions output by the first machine learning model as catalyst upstream state quantities.

[0079] The catalyst downstream state quantity estimation unit 53 estimates state quantities of various substances emitted downstream of the three-way catalyst (catalyst downstream NOx emissions, catalyst downstream HC emissions, catalyst downstream CO emissions, oxygen storage ratio, catalyst downstream air-fuel ratio, and catalyst temperature). In the following description, catalyst downstream NOx emissions, catalyst downstream HC emissions, and catalyst downstream CO emissions are also abbreviated as "catalyst downstream NOx, HC, and CO emissions." The catalyst downstream state quantity estimation unit 53 receives the catalyst upstream air-fuel ratio, exhaust gas flow rate, exhaust gas temperature, and catalyst upstream state quantities (catalyst upstream NOx, HC, and CO emissions), and estimates the catalyst temperature, oxygen storage capacity, catalyst downstream air-fuel ratio (rear equivalence ratio), and catalyst downstream NOx, HC, and CO emissions as catalyst downstream state quantities. The catalyst downstream state quantity estimation unit 53 is configured with a second state space model. The second state space model is configured with, for example, a second machine learning model (a neural network model separate from the first state space model configured by the catalyst upstream state quantity estimation unit 52). Then, the catalyst downstream state quantity estimation unit (catalyst downstream state quantity estimation unit 53) inputs the catalyst upstream air-fuel ratio, exhaust gas flow rate, and exhaust gas temperature, as well as the catalyst upstream state quantity, into a second machine learning model that has been trained in advance, and estimates the catalyst temperature, oxygen storage capacity, catalyst downstream air-fuel ratio, and catalyst downstream NOx, HC, and CO emissions of the three-way catalyst output by the second machine learning model as the catalyst downstream state quantities.

[0080] The correction unit (correction unit 54) corrects the catalyst downstream NOx, HC, and CO emission amounts using the NOx sensor value detected by the downstream NOx sensor (catalyst downstream NOx sensor 29). For example, the correction unit 54 calculates correction amounts for the catalyst downstream NOx, HC, and CO emission amounts based on the catalyst downstream state quantities estimated by the second state space model of the catalyst downstream state quantity estimation unit 53. At this time, the correction unit 54 calculates a correction parameter that is the difference between the catalyst downstream NOx emission amount and the NOx sensor value based on the corrected catalyst downstream NOx emission amount and the NOx sensor value detected by the downstream NOx sensor (catalyst downstream NOx sensor 29). Then, the correction unit 54 corrects the estimated catalyst downstream NOx, HC, and CO emission amounts based on the correction parameter.

[0081] The correction unit (correction unit 54) corrects the catalyst downstream NOx, HC, and CO emission amounts using an extended Kalman filter. This correction unit 54 is configured by an extended Kalman filter that receives as input the catalyst downstream NOx, HC, and CO emission amounts and the signal from the catalyst downstream NOx sensor 29. The ECU 28 then outputs the catalyst downstream NOx, HC, and CO emission amounts, which are the output of the correction unit 54, to a monitoring unit (not shown), which is a diagnostic device or the like, thereby achieving OBM. The output from the correction unit 54 to the monitoring unit is represented by an arrow extending upward from the correction unit 54. The second state space model of the catalyst downstream state quantity estimation unit 53 estimates the oxygen storage ratio, the catalyst downstream air-fuel ratio, and the catalyst temperature, which are related to the catalyst deterioration state, and therefore can take catalyst deterioration into account and estimate the catalyst downstream NOx, HC, and CO emission amounts that change depending on the deterioration state.

[0082] The diagnosing unit (NOx sensor diagnosing unit 55) diagnoses the downstream NOx sensor (catalyst downstream NOx sensor 29) based on the catalyst downstream NOx emission amount corrected by the correction unit (correction unit 54) and the NOx sensor value detected by the downstream NOx sensor (catalyst downstream NOx sensor 29). At this time, the diagnosing unit (NOx sensor diagnosing unit 55) diagnoses the downstream NOx sensor (catalyst downstream NOx sensor 29) based on the catalyst downstream NOx emission amount corrected based on the correction parameter calculated by the correction unit 54 and the NOx sensor value detected by the downstream NOx sensor (catalyst downstream NOx sensor 29). For example, the NOx sensor diagnosing unit 55 diagnoses the catalyst downstream NOx sensor 29 by comparing the catalyst downstream NOx emission amount with the NOx sensor value of the catalyst downstream NOx sensor 29. Then, the NOx sensor diagnosing unit 55 outputs the diagnosis result of the catalyst downstream NOx sensor 29. Therefore, the degree of deterioration of the catalyst downstream NOx sensor 29 can be determined early, and measures such as replacement of the catalyst downstream NOx sensor 29 can be taken.

[0083] The fuel injection amount correction unit (fuel injection amount correction unit 56) calculates a correction value for the injection amount of fuel injected by the fuel injection unit based on the catalyst downstream NOx, HC, and CO emission amounts corrected by the correction unit (correction unit 54). To this end, the fuel injection amount correction unit 56 receives the catalyst downstream NOx, HC, and CO emission amounts as input and calculates a fuel injection amount correction value for reducing the catalyst downstream NOx, HC, and CO emission amounts using a predetermined fuel injection amount correction map 56a. For example, if the catalyst downstream NOx emission amount increases, the fuel injection amount correction unit 56 outputs a fuel injection amount correction value that shifts the fuel injection amount to the rich side to the fuel injector 15. Then, the injection amount is adjusted so that the NOx, HC, and CO emission amounts are zero or reduced as much as possible as a result of combustion. As a result, the fuel injector 15 injects fuel whose injection amount is adjusted so that appropriate NOx, HC, and CO emission amounts are obtained after combustion. Note that the arguments of the fuel injection amount correction map 56a can also be set to two inputs: the catalyst downstream NOx and HC emission amounts.

[0084] Next, an example of the processing of the ECU 28 will be described with reference to FIG. FIG. 14 is a flowchart showing an example of processing by the ECU 28 according to the present invention.

[0085] First, the exhaust gas flow rate calculation unit 51 of the ECU 28 calculates the exhaust gas flow rate using the engine speed based on the crank angle sensor 18 and the charging efficiency based on the flow rate sensor 2 as inputs (S1).

[0086] Next, the ECU 28 acquires the exhaust gas flow rate, the exhaust gas temperature based on the exhaust gas temperature sensor 26, and the catalyst upstream air-fuel ratio based on the catalyst upstream air-fuel ratio sensor 20 in a storage medium such as a RAM of the ECU 28, and inputs these values into the first state space model of the catalyst upstream state quantity estimation unit 52 (S2).

[0087] Next, the catalyst upstream state quantity estimation unit 52 estimates catalyst upstream state quantities such as catalyst upstream NOx, HC, and CO emissions based on the exhaust gas flow rate, exhaust gas temperature, and catalyst upstream air-fuel ratio input to the first state space model (S3). Thus, the processing of step S3 includes a first estimation step of inputting the exhaust gas flow rate, exhaust gas temperature, and catalyst upstream air-fuel ratio to a first machine learning model that has been trained in advance, and estimating the catalyst upstream NOx, HC, and CO emissions output by the first machine learning model as catalyst upstream state quantities.

[0088] Next, the catalyst downstream state quantity estimation unit 53 inputs the exhaust gas flow rate, exhaust gas temperature, catalyst upstream air-fuel ratio, and catalyst upstream NOx, HC, and CO emission amounts into the second state space model. Then, the catalyst downstream state quantity estimation unit 53 estimates catalyst downstream state quantities such as the catalyst temperature, oxygen storage capacity, catalyst downstream air-fuel ratio (rear equivalence ratio), catalyst downstream NOx, HC, and CO emission amounts based on the various information input into the second state space model (S4). Thus, the processing of step S4 includes a second estimation step of inputting the catalyst upstream air-fuel ratio, exhaust gas flow rate, exhaust gas temperature, and the catalyst upstream state quantities estimated by the first machine learning model into a second machine learning model that has been trained in advance, and estimating the catalyst temperature, oxygen storage capacity, catalyst downstream air-fuel ratio, and catalyst downstream NOx, HC, and CO emission amounts of the three-way catalyst output by the second machine learning model as catalyst downstream state quantities.

[0089] Next, the ECU acquires the sensor value of the catalyst downstream NOx sensor 29 (S5). The sensor value of the catalyst downstream NOx sensor 29 is input to the extended Kalman filter of the correction unit and the NOx sensor diagnosis unit 55.

[0090] Next, the correction unit 54 inputs the downstream-catalyst NOx, HC, and CO emission amounts input from the catalyst downstream state quantity estimation unit 53 to an extended Kalman filter, and corrects the downstream-catalyst NOx, HC, and CO emission amounts using the extended Kalman filter (S6). The process of step S6 embodies a process in which the correction unit 54 calculates a correction parameter, which is the difference between the downstream-catalyst NOx emission amount based on the corrected downstream-catalyst NOx emission amount and the NOx sensor value detected by the downstream NOx sensor (downstream-catalyst NOx sensor 29). The downstream-catalyst NOx, HC, and CO emission amounts calculated in step S6 are used to correct the fuel injection amount in the subsequent fuel injection amount correction unit 56. The process performed by the correction unit 54 using the extended Kalman filter is shown in FIG. 15, which will be described later.

[0091] Next, the correction unit 54 transmits the catalyst downstream NOx, HC, and CO emission amounts corrected in step S6 to a predetermined display unit, such as a diagnostic device, external to the ECU 28 via well-known communication means, such as CAN (Controller Area Network) communication. As a result, the catalyst downstream NOx, HC, and CO emission amounts can be displayed in the diagnostic device (not shown) that receives the catalyst downstream NOx, HC, and CO emission amounts (S7), thereby realizing OBM. Possible forms of OBM include, for example, displaying a time series graph or displaying numerical values.

[0092] Furthermore, the NOx sensor diagnostic unit 55 compares the estimated value of the catalyst downstream NOx emission amount corrected based on the correction parameters obtained by the correction unit 54 in step S6 with the sensor value acquired from the catalyst downstream NOx sensor 29 in step S5. Then, the NOx sensor diagnostic unit 55 diagnoses the catalyst downstream NOx sensor 29 based on the comparison result (S8). This diagnostic result can also be output to an external diagnostic device (not shown). This can improve the reliability of the values of the catalyst downstream NOx, HC, and CO emission amounts. After step S8, the ECU 28 ends this process.

[0093] On the other hand, after step S6, in parallel with step S8, fuel injection amount correction unit 56 calculates a fuel injection amount correction value for reducing catalyst downstream NOx, HC, and CO emissions based on the estimated values (correction values) of catalyst downstream NOx, HC, and CO emissions input from correction unit 54 (S9). The fuel injection amount of fuel injected from fuel injection valve 15 is corrected by the fuel injection amount correction value calculated by fuel injection amount correction unit 56, and fuel is injected from fuel injection valve 15 in the corrected fuel injection amount. After step S9, ECU 28 ends this process.

[0094] Fig. 15 is a flowchart showing an example of a Kalman filter algorithm used in the correction unit 54 that corrects the catalyst downstream NOx, HC, and CO emissions shown in Fig. 13. Here, the algorithm and a method for applying the Kalman filter, which is one of the components of the present invention, to this control will be described. The Kalman filter is based on a state equation that includes system noise Q and observation noise R defined by the following equations (1) and (2).

[0095]

number

[0096] The Kalman filter is divided into a prediction step and a filtering step. In the prediction step, the correction unit 54 updates the internal state variable vector x and the covariance matrix P using the following equations (3) and (4) based on the input variables and the system noise Q (S11, S12).

[0097]

number

[0098] Next, in the filtering step, the correction unit 54 calculates the Kalman gain K, which is defined by the updated covariance matrix P and the observation noise R, using the following equation (5) (S13).

[0099]

number

[0100] Next, the correction unit 54 updates the internal state variable vector x and the covariance matrix P again using the Kalman gain K(k) and the observation data (y(k): NOx sensor detected value) according to the following equations (6) and (7) (S14, S15).

[0101]

number

[0102] In equation (6), the term multiplied by the Kalman gain K(k) represents the difference with respect to the detected value y(k) of the catalyst downstream NOx sensor 29. This difference represents the correction parameter. In this way, the internal state variable vector x and the covariance matrix P are corrected using the actual observation data (NOx sensor detected value).

[0103] Through the above calculations, the correction unit 54 can estimate, based on measurable output information, the behavior of the internal state variable x(k|k), which is difficult to measure directly by the catalyst downstream NOx sensor 29. For this reason, the correction unit 54 according to this embodiment estimates the catalyst downstream NOx, HC, and CO emissions using a Kalman filter, based on the detection value of the catalyst downstream NOx sensor 29.

[0104] Next, the first machine learning model and the second machine learning model will be described with reference to FIGS. 16A and 16B. FIG. 16A is a diagram illustrating a first machine learning model that constitutes a first state space model and a second machine learning model that constitutes a second state space model.

[0105] In this embodiment, the first machine learning model and the second machine learning model are configured as neural network models. A neural network model is a mathematical model that mimics the structure of the neural circuits in the human brain. As shown in the upper and lower left of FIG. 16A, a weight w and a bias b are set for each neuron that configures the neural network model. In this neural network model, a1 to an For each input signal, weights w1 to w n The function f(z) is obtained by multiplying the function f(z) by , and then adding a bias b. Here, the equation z and the input signal a satisfy the following relationship. z=w1a1+w2a2+…+w n a n +b a=f(z)

[0106] Furthermore, as shown in the lower right of Figure 16A, a function called an activation function is defined for each neuron. Activation functions such as logistic functions, ramp functions, and sigmoid functions are set as appropriate. A layer is formed from multiple neurons, and a hidden layer is set between the input layer and the output layer. By increasing the number of neurons and the number of hidden layers, the neural network model can approximate more complex input-output relationships. There is a trade-off between approximation accuracy and model size, and a compromise that satisfies both requirements must be selected.

[0107] FIG. 16B is a diagram illustrating an example of a neural network model. The upper part of FIG. 16B shows an example of a neural network model configured by supervised learning. The neural network model is configured with an input layer, multiple intermediate layers, and an output layer. For example, an exhaust gas sensor signal is set in the input layer, and the catalyst deterioration diagnosis result is set in the output layer. Here, the catalyst deterioration diagnosis result set in the output layer is, for example, the measured catalyst upstream NOx, HC, and CO emissions in the first state space model shown in FIG. 12. Also, in the second state space model shown in FIG. 12, the measured catalyst downstream NOx, HC, and CO emissions. After various values are set in the input layer and the output layer, the neural network model can approximate the input / output relationship by changing the weights of each neuron through machine learning (supervised). Note that the backpropagation method can be applied to this machine learning (supervised) algorithm.

[0108] The lower part of Fig. 16B shows an example of a neural network model used when estimating various values. The neural network model used in the estimation calculation is the neural network model configured by supervised learning shown in the upper part of Fig. 16A. An actual sensor signal is input to the input layer of the neural network model, passes through the middle layer, and the estimated catalyst deterioration diagnosis result is output from the output layer. In the case of the first state space model shown in Fig. 12, estimated catalyst upstream NOx, HC, and CO emissions are output. In addition, in the case of the second state space model shown in Fig. 12, estimated catalyst downstream NOx, HC, and CO emissions are output.

[0109] Next, the machine learning steps of the first machine learning model and the second machine learning model will be described with reference to FIG. 17 is a block diagram of the first and second machine learning models for explaining the machine learning process. Here, the machine learning process (first learning process and second learning process) shown in FIG. 17 is performed using test equipment well known to those skilled in the art, such as an engine testing laboratory. Therefore, the machine learning process can be performed without actually driving an automobile equipped with the internal combustion engine 1 that is the subject of this machine learning on the road.

[0110] First, a block 61 calculates an exhaust gas flow rate using as input the engine speed based on the crank angle sensor 18 of the internal combustion engine control system 100 and the charging efficiency based on the flow rate sensor 2. Note that the block 61 may be the same as the exhaust gas flow rate calculation unit 51 shown in FIG.

[0111] Next, block 62 executes supervised machine learning of the first machine learning model, as shown in the upper part of Fig. 16B. To this end, block 62 receives as input the exhaust gas temperature based on the exhaust gas temperature sensor 26 in the internal combustion engine control system 100, the catalyst upstream air-fuel ratio based on the catalyst upstream air-fuel ratio sensor 20, the exhaust gas flow rate calculated in block 61, and upstream-catalyst state quantities (actual exhaust gas information upstream of the catalyst) obtained by upstream-catalyst NOx, HC, and CO sensors additionally provided outside the internal combustion engine control system 100 for the machine learning process (first learning process) of the first machine learning model. Then, the upstream-catalyst NOx, HC, and CO estimated by block 62 are output to block 63 as estimated upstream-catalyst state quantities.

[0112] In this embodiment, the exhaust gas temperature and the catalyst upstream air-fuel ratio in the machine learning are acquired from the exhaust gas temperature sensor 26 and the catalyst upstream air-fuel ratio sensor 20 arranged in the internal combustion engine control system 100. However, the present invention is not limited to this, and the exhaust gas temperature and the catalyst upstream air-fuel ratio measured by measuring instruments arranged outside the internal combustion engine control system 100 may also be input to the block 62.

[0113] 16B, block 63 executes machine learning (supervised) of the second machine learning model. For this purpose, block 63 receives the above-mentioned exhaust gas temperature, catalyst upstream air-fuel ratio, exhaust gas flow rate, and information from the catalyst upstream NOx, HC, and CO sensors output from block 62, as well as catalyst downstream state quantities (actual exhaust gas information downstream of the catalyst) obtained from catalyst downstream (post-catalyst) NOx, HC, and CO sensors additionally provided outside the internal combustion engine control system 100 for the pre-learning step (second learning step) of the second machine learning model.

[0114] Next, the first learning process and the second learning process will be described with reference to FIG. FIG. 18 is a flowchart of the first learning process and the second learning process.

[0115] First, block 61 measures the exhaust gas flow rate calculated using as input the engine speed based on the crank angle sensor 18 of the internal combustion engine control system 100 and the charging efficiency based on the flow rate sensor 2. Also, an exhaust gas temperature sensor (not shown) measures the exhaust gas temperature, and an upstream-of-catalyst air-fuel ratio sensor (not shown) measures the upstream-of-catalyst air-fuel ratio. Also, an upstream-of-catalyst HC, CO, and NOx sensor (not shown) measures the upstream-of-catalyst state quantities. Also, a downstream-of-catalyst HC, CO, and NOx sensor (not shown) measures the downstream-of-catalyst state quantities (S21). These upstream-of-catalyst air-fuel ratio sensor, upstream-of-catalyst HC, CO, and NOx sensor, and downstream-of-catalyst HC, CO, and NOx sensors (not shown) are all provided in measuring devices located outside the internal combustion engine control system 100.

[0116] Next, block 62 acquires the measured exhaust gas flow rate from block 61. Block 62 also acquires the measured exhaust gas temperature and catalyst upstream air-fuel ratio. Block 62 also acquires catalyst upstream state quantities, which are sensor values obtained from the catalyst upstream NOx, HC, and CO sensors (S22).

[0117] Next, block 62 inputs the measured exhaust gas flow rate, exhaust gas temperature, and catalyst upstream air-fuel ratio, as well as the measured catalyst upstream state quantities, as training data, and executes learning of a first machine learning model (first state space model) (first learning step). Here, block 62 executes learning of a first machine learning model (first state space model) that estimates the catalyst upstream NOx, HC, and CO emissions as catalyst upstream state quantities based on the input training data (S23).

[0118] The first machine learning model (first state space model) approximates the complex input / output relationship between the catalyst upstream air-fuel ratio, exhaust gas flow rate, and exhaust gas temperature, which are the input side of the neural network model, and the catalyst upstream state quantities outside the internal combustion engine control system 100, which are the output side of the neural network model.

[0119] Next, block 63 acquires the measured exhaust gas flow rate, exhaust gas temperature, and catalyst upstream air-fuel ratio, and acquires the catalyst upstream state quantities, which are the catalyst upstream NOx, HC, and CO emissions estimated by the first machine learning model. Furthermore, block 63 acquires the catalyst downstream state quantities measured by the catalyst downstream HC, CO, and NOx sensors from the catalyst downstream HC, CO, and NOx (S24).

[0120] Next, block 63 inputs the measured exhaust gas flow rate, exhaust gas temperature, and catalyst upstream air-fuel ratio, the catalyst upstream state quantities estimated by the first machine learning model, and the measured catalyst downstream state quantities (catalyst downstream NOx, HC, and CO emission amounts) as training data, and executes learning of a second machine learning model (second state space model) (second learning step).Here, block 63 executes learning of a second machine learning model (second state space model) that estimates the catalyst temperature, oxygen storage capacity, catalyst downstream air-fuel ratio, and catalyst downstream NOx, HC, and CO emission amounts of the three-way catalyst as the catalyst downstream state quantities (S25).

[0121] The second machine learning model (second state space model) approximates the complex input / output relationship between the values of the exhaust gas flow rate, exhaust gas temperature, catalyst upstream air-fuel ratio, and catalyst upstream state quantities, which are the input side of the neural network model, and the above-mentioned catalyst downstream state quantities outside the internal combustion engine control system 100, which are the output side of the neural network model.

[0122] The first learning process and the second learning process can be executed using a general-purpose computer provided separately from the ECU 28.

[0123] The ECU 28 according to the first embodiment described above estimates the amount of NOx emissions downstream of the catalyst, the amount of HC emissions downstream of the catalyst, and the amount of CO emissions downstream of the catalyst based on information obtained from existing sensors and the like installed in the internal combustion engine 1. This makes it possible to display the amount of NOx emissions downstream of the catalyst, the amount of HC emissions downstream of the catalyst, and the amount of CO emissions downstream of the catalyst in real time without providing expensive HC and CO sensors upstream of the catalyst.

[0124] Here, the catalyst upstream state quantity estimation unit 52 of the ECU 28 can estimate the catalyst upstream NOx, HC, and CO emissions as catalyst upstream state quantities by inputting the catalyst upstream air-fuel ratio, exhaust gas flow rate, and exhaust gas temperature into a first machine learning model. The first machine learning model is trained in advance using the measured catalyst upstream air-fuel ratio, exhaust gas flow rate, exhaust gas temperature, etc., and the measured catalyst upstream state quantities as training data, and can easily estimate the catalyst upstream NOx, HC, and CO emissions from various complex input values.

[0125] Furthermore, the catalyst downstream state quantity estimation unit 53 of the ECU 28 can estimate the catalyst downstream state quantities, including NOx, HC, and CO emissions, the oxygen storage ratio, the catalyst downstream air-fuel ratio, and the catalyst temperature, as catalyst downstream state quantities by inputting the catalyst upstream air-fuel ratio, exhaust gas flow rate, and exhaust gas temperature, and the catalyst upstream state quantities estimated by the catalyst upstream state quantity estimation unit 52, into a second machine learning model. The second machine learning model is trained in advance using the measured exhaust gas flow rate, exhaust gas temperature, and catalyst upstream air-fuel ratio, the catalyst upstream state quantities estimated by the first machine learning model, and the measured catalyst downstream state quantities (catalyst downstream NOx, HC, and CO emissions) as training data. The second machine learning model can easily estimate the catalyst downstream NOx, HC, and CO emissions, the oxygen storage ratio, the catalyst downstream air-fuel ratio, and the catalyst temperature from various complex input values.

[0126] Furthermore, the correction unit 54 of the ECU 28 uses an extended Kalman filter to correct the catalyst downstream state quantities (catalyst downstream NOx, HC, and CO emission amounts) estimated by the catalyst downstream state quantity estimation unit 53. Then, the corrected catalyst downstream state quantities (catalyst downstream NOx, HC, and CO emission amounts) are displayed in real time in the monitoring unit.

[0127] Furthermore, the NOx sensor diagnostic unit 55 of the ECU 28 can diagnose the catalyst downstream NOx sensor 29 by comparing the catalyst downstream NOx value corrected by the correction unit 54 with the sensor value of the catalyst downstream NOx sensor 29. This makes it possible to detect the occurrence of an abnormality in the catalyst downstream NOx sensor 29 at an early stage, allowing for early replacement of the catalyst downstream NOx sensor 29. Furthermore, if the ECU 28 determines from the diagnosis result that the catalyst downstream NOx sensor 29 has deteriorated, it can use the catalyst downstream NOx value corrected by the correction unit 54 for control instead of using the sensor value of the catalyst downstream NOx sensor 29 for control.

[0128] Furthermore, a fuel injection amount correction unit 56 of the ECU 28 corrects the fuel injection amount based on the catalyst downstream state quantities (catalyst downstream NOx, HC, and CO emission amounts) corrected by the correction unit 54. This allows the ECU 28 to maintain the equivalence ratio at the stoichiometric air-fuel ratio, which is the three-way point, and to perform control to maintain the purification rate of the three-way catalyst at a high level.

[0129] [Second embodiment] In the first embodiment described above, for example, the internal combustion engine 1 is mainly configured to burn gasoline, but the present invention is not limited to this. For example, the present invention can also be applied to an internal combustion engine 1 that uses alternative fuels to gasoline, such as synthetic fuels derived from hydrogen and so-called biofuels based on natural materials derived from plants and animals.

[0130] In the second embodiment of the present invention, the learning data of the input / output relationships used in the first state space model and the second state space model are switched depending on the type of fuel to accommodate changes in fuel composition, such as hydrogen-derived synthetic fuels and biofuels, thereby enabling the ECU to estimate the amounts of NOx, HC, and CO emissions downstream of the catalyst depending on the type of fuel.

[0131] 19 is a block diagram showing an example of the internal configuration of an ECU 28A according to the second embodiment. The ECU 28 includes the following blocks 51 to 56.

[0132] The exhaust gas flow rate calculation unit 51 receives the engine speed based on the crank angle sensor 18 and the charging efficiency based on the flow rate sensor 2 as inputs and calculates the exhaust gas flow rate.

[0133] The catalyst upstream state quantity estimation unit 52 receives as input the exhaust gas temperature based on the exhaust gas temperature sensor 26, the catalyst upstream air-fuel ratio based on the catalyst upstream air-fuel ratio sensor 20, and the exhaust gas flow rate, and constructs a first state space model that estimates the catalyst upstream NOx, HC, and CO emissions. This first state space model is provided with a plurality of learning data corresponding to the various fuels described above. For example, the first state space model is provided in a switchable manner among a first machine learning model A obtained by performing the first learning process when hydrogen-derived synthetic fuel is being burned in the internal combustion engine 1, a first machine learning model B obtained by performing the first learning process when biofuel is being burned in the internal combustion engine 1, and a first machine learning model C obtained by performing the first learning process when standard gasoline is being burned.

[0134] The catalyst downstream state quantity estimation unit 53 is a second state space model that receives the exhaust gas temperature, the catalyst upstream air-fuel ratio, and the catalyst upstream NOx, HC, and CO emissions as inputs and estimates the catalyst downstream NOx, HC, and CO emissions, the oxygen storage ratio, the catalyst downstream air-fuel ratio, and the catalyst temperature. This second state space model is provided with a plurality of learning data corresponding to the various fuels. For example, the second state space model is provided in a switchable manner among a second machine learning model A obtained by performing the second learning process when a hydrogen-derived synthetic fuel is being burned in the internal combustion engine 1, a second machine learning model B obtained by performing the second learning process when a biofuel is being burned in the internal combustion engine 1, and a second machine learning model C obtained by performing the second learning process when standard gasoline is being burned.

[0135] The fuel composition discrimination unit (fuel composition discrimination unit 57) discriminates the composition of the fuel. For example, the fuel composition discrimination unit 57 has a reception unit (not shown) that receives a fuel composition identification number that is previously assigned to fuel such as gasoline, hydrogen-derived synthetic fuel, or biofuel. This fuel composition discrimination unit 57 switches between the first machine learning models A to C or the second machine learning models A to C described above based on the fuel composition identification number received by a reception unit (not shown) configured inside the ECU 28. In this way, the fuel composition discrimination unit 57 can set the first state space model and the second state space model to suit the set fuel.

[0136] For this reason, the catalyst upstream state quantity estimation unit (catalyst upstream state quantity estimation unit 52) has multiple types of first machine learning models that are learned in advance according to the composition of the fuel, and selects one first machine learning model from the multiple types of first machine learning models based on the fuel composition determined by the fuel composition determination unit (fuel composition determination unit 57). For this reason, the catalyst upstream state quantity estimation unit 52 is able to estimate the catalyst upstream NOx, HC, and CO emissions as catalyst upstream state quantities using the first machine learning model selected in accordance with the fuel burned in the internal combustion engine 1.

[0137] Furthermore, the catalyst downstream state quantity estimation unit (catalyst downstream state quantity estimation unit 53) has multiple types of second machine learning models that have been trained in advance according to the composition of the fuel, and selects one second machine learning model from the multiple types of second machine learning models based on the fuel composition determined by the fuel composition determination unit (fuel composition determination unit 57). Therefore, the catalyst downstream state quantity estimation unit 53 can estimate the catalyst temperature, oxygen storage capacity, catalyst downstream air-fuel ratio, and catalyst downstream NOx, HC, and CO emissions of the three-way catalyst as catalyst downstream state quantities using the second machine learning model selected in accordance with the fuel burned in the internal combustion engine 1.

[0138] The correction unit 54 is an example of a correction unit that derives the catalyst downstream NOx, HC, and CO emission amounts based on the catalyst downstream NOx, HC, and CO emission amounts estimated using the second state space model. The correction unit 54 is configured by an extended Kalman filter that receives as input the catalyst downstream NOx, HC, and CO emission amounts and the catalyst downstream NOx sensor 29 signal.

[0139] The NOx sensor diagnostic unit 55 compares the catalyst downstream NOx emission amount with the sensor value of the catalyst downstream NOx sensor 29, diagnoses the catalyst downstream NOx sensor 29, and outputs the diagnostic result.

[0140] The fuel injection amount correction unit 56 receives the catalyst downstream NOx, HC, and CO emission amounts as input and calculates a fuel injection amount correction value for reducing the catalyst downstream NOx, HC, and CO emission amounts using a predetermined fuel injection amount correction map 56a. For example, if the catalyst downstream NOx emission amount increases, the fuel injection amount correction unit 56 outputs a fuel injection amount correction value that shifts the fuel injection amount to the rich side to the fuel injector 15. Here, the arguments of the fuel injection amount correction map 56a can be set to two inputs: the catalyst downstream NOx emission amount and the catalyst downstream HC emission amount.

[0141] The ECU 28A outputs the catalyst downstream NOx, HC, and CO, which are the outputs of the correction unit 54, to a monitoring unit (not shown) such as a diagnostic device, thereby realizing OBM compatible with various gasoline alternative fuels.

[0142] Next, the processing performed by the ECU 28A according to the second embodiment will be described. FIG. 20 is a flowchart showing an example of processing by the ECU 28A.

[0143] First, the exhaust gas flow rate calculation unit 51 of the ECU 28A calculates the exhaust gas flow rate using the engine speed based on the crank angle sensor 18 and the charging efficiency based on the flow rate sensor 2 as inputs (S1).

[0144] Next, the ECU 28A acquires the exhaust gas flow rate, the exhaust gas temperature based on the exhaust gas temperature sensor 26, and the catalyst upstream air-fuel ratio based on the catalyst upstream air-fuel ratio sensor 20 in a storage medium such as a RAM of the ECU 28A, and inputs these values into the first state space model of the catalyst upstream state quantity estimation unit 52 (S31).

[0145] Next, the fuel composition determining unit 57 receives a fuel composition identification number. Then, based on the received fuel composition identification number, the fuel composition determining unit 57 switches and sets the first state space model of the catalyst upstream state quantity estimating unit 52 to one of the first machine learning models A to C. At the same time, based on the received fuel composition identification number, the fuel composition determining unit 57 switches and sets the second state space model of the catalyst downstream state quantity estimating unit 53 to one of the second machine learning models A to C (S32).

[0146] Next, the catalyst upstream state quantity estimation unit 52 inputs the exhaust gas flow rate, exhaust gas temperature, and catalyst upstream air-fuel ratio into the first state space model to which the first machine learning model was switched and set in step S32, and estimates the catalyst upstream NOx, HC, and CO emissions corresponding to the fuel composition (S33).

[0147] Next, the catalyst downstream state quantity estimation unit 53 inputs the exhaust gas flow rate, exhaust gas temperature, catalyst upstream air-fuel ratio, and catalyst upstream NOx, HC, and CO emission amounts to the second state space model to which the second machine learning model has been switched and set in step S32.The catalyst downstream state quantity estimation unit 53 then estimates catalyst downstream state quantities such as the catalyst temperature, oxygen storage capacity (OSC), catalyst downstream air-fuel ratio (rear equivalence ratio), catalyst downstream NOx, HC, and CO emission amounts corresponding to the fuel composition (S34).

[0148] Next, the ECU acquires the sensor value of the catalyst downstream NOx sensor 29 (S35). The sensor value of the catalyst downstream NOx sensor 29 is input to the extended Kalman filter of the correction unit and the NOx sensor diagnosis unit 55.

[0149] Next, the correction unit 54 uses an extended Kalman filter to determine the amounts of NOx, HC, and CO emissions downstream of the catalyst based on the amounts of NOx, HC, and CO emissions downstream of the catalyst (S36).

[0150] The processing of steps S7 to S9 after step S36 is the same as the processing of each step in the ECU 28 according to the first embodiment described with reference to FIG. 14, and therefore detailed description thereof will be omitted.

[0151] In the ECU 28A according to the second embodiment described above, the fuel composition discrimination unit 57 is provided, thereby enabling the catalyst upstream state quantity estimation unit 52 to select an optimal first machine learning model in accordance with the fuel composition discrimination result. Similarly, the catalyst downstream state quantity estimation unit 53 can select an optimal second machine learning model in accordance with the fuel composition discrimination result by the fuel composition discrimination unit 57. As a result, the catalyst upstream NOx, HC, and CO emission amounts estimated by the first machine learning model and the catalyst downstream NOx, HC, and CO emission amounts estimated by the second machine learning model are all estimated as valid values.

[0152] The present invention is not limited to the above-described embodiment, and it goes without saying that various other applications and modifications are possible without departing from the gist of the present invention as set forth in the claims. For example, the above-described embodiment has described the system configuration in detail and specifically to clearly explain the present invention, and is not necessarily limited to a system including all of the described configurations. Furthermore, it is also possible to add, delete, or replace part of the configuration of this embodiment with other configurations. In addition, the control lines and information lines shown are those that are considered necessary for the explanation, and do not necessarily show all the control lines and information lines in the product. In reality, it can be assumed that almost all components are interconnected. [Explanation of symbols]

[0153] 1...internal combustion engine, 20...catalyst upstream air-fuel ratio sensor, 21...exhaust purification catalyst, 22...rear oxygen sensor, 28...ECU, 51...exhaust gas flow rate calculation unit, 52...catalyst upstream state quantity estimation unit, 53...catalyst downstream state quantity estimation unit, 54...correction unit, 55...NOx sensor diagnosis unit, 56...fuel injection amount correction unit, 56a...fuel injection amount correction map, 57...fuel composition determination unit, 100...internal combustion engine control system, 110...aftertreatment system

Claims

1. 1. An internal combustion engine control device for controlling an internal combustion engine, the internal combustion engine including: an upstream air-fuel ratio sensor provided in an exhaust pipe, arranged upstream of a three-way catalyst having an oxygen storage capacity, for detecting an air-fuel ratio upstream of the catalyst of exhaust gas; and a downstream NOx sensor arranged downstream of the three-way catalyst, for detecting NOx downstream of the three-way catalyst, an exhaust gas flow rate calculation unit that calculates an exhaust gas flow rate of the exhaust gas; a catalyst upstream state quantity estimation unit that receives the catalyst upstream air-fuel ratio, the exhaust gas flow rate, and the exhaust gas temperature detected by an exhaust gas temperature sensor as inputs and estimates catalyst upstream NOx emission amounts, catalyst upstream HC emission amounts, and catalyst upstream CO emission amounts as catalyst upstream state quantities; a catalyst downstream state quantity estimation unit that receives the catalyst upstream air-fuel ratio, the exhaust gas flow rate, the exhaust gas temperature, and the catalyst upstream state quantities as inputs and estimates the catalyst temperature of the three-way catalyst, the oxygen storage capacity, the catalyst downstream air-fuel ratio, the catalyst downstream NOx emissions, the catalyst downstream HC emissions, and the catalyst downstream CO emissions as catalyst downstream state quantities; a correction unit that corrects the catalyst downstream NOx emission amount, the catalyst downstream HC emission amount, and the catalyst downstream CO emission amount using the NOx sensor value detected by the downstream NOx sensor. Internal combustion engine control device.

2. The catalyst upstream state quantity estimation unit inputs the catalyst upstream air-fuel ratio, the exhaust gas flow rate, and the exhaust gas temperature into a first machine learning model that has been trained in advance, and estimates the catalyst upstream NOx emission amount, the catalyst upstream HC emission amount, and the catalyst upstream CO emission amount output by the first machine learning model as the catalyst upstream state quantities. The internal combustion engine control device according to claim 1.

3. The catalyst downstream state quantity estimation unit inputs the catalyst upstream air-fuel ratio, the exhaust gas flow rate, the exhaust gas temperature, and the catalyst upstream state quantity into a second machine learning model that has been trained in advance, and estimates the catalyst temperature, the oxygen storage capacity, the catalyst downstream air-fuel ratio, the catalyst downstream NOx emissions, the catalyst downstream HC emissions, and the catalyst downstream CO emissions output by the second machine learning model as the catalyst downstream state quantities.

3. The internal combustion engine control device according to claim 2.

4. The correction unit corrects the catalyst downstream NOx emission amount, the catalyst downstream HC emission amount, and the catalyst downstream CO emission amount using an extended Kalman filter. The internal combustion engine control device according to claim 3.

5. a fuel injection amount correction unit that calculates a correction value for the injection amount of fuel injected by a fuel injection unit based on the catalyst downstream NOx emission amount, the catalyst downstream HC emission amount, and the catalyst downstream CO emission amount corrected by the correction unit; The internal combustion engine control device according to claim 3.

6. a diagnosis unit that diagnoses the NOx sensor based on the catalyst downstream NOx emission amount corrected by the correction unit and the NOx sensor value detected by the downstream NOx sensor; The internal combustion engine control device according to claim 3.

7. a fuel composition determination unit that determines the composition of the fuel; the catalyst upstream state quantity estimation unit has a plurality of types of first machine learning models that have been trained in advance according to the composition of the fuel, and selects one of the plurality of types of first machine learning models based on the composition of the fuel determined by the fuel composition determination unit; The catalyst downstream state quantity estimation unit has a plurality of types of second machine learning models that are trained in advance according to the composition of the fuel, and selects one of the plurality of types of second machine learning models based on the composition of the fuel determined by the fuel composition determination unit. The internal combustion engine control device according to claim 3.

8. 1. A state quantity estimation method for estimating a state quantity of an internal combustion engine, the state quantity of which is provided in an exhaust pipe, the state quantity of which is estimated by an upstream air-fuel ratio sensor disposed upstream of a three-way catalyst having an oxygen storage capacity, the upstream air-fuel ratio sensor detecting an air-fuel ratio upstream of the catalyst of exhaust gas, and the downstream NOx sensor disposed downstream of the three-way catalyst, the downstream NOx sensor detecting NOx downstream of the three-way catalyst, calculating an exhaust gas flow rate of the exhaust gas; a step of estimating, as catalyst upstream state quantities, catalyst upstream NOx emission amounts, catalyst upstream HC emission amounts, and catalyst upstream CO emission amounts, using the catalyst upstream air-fuel ratio, the exhaust gas flow rate, and the exhaust gas temperature detected by an exhaust gas temperature sensor as inputs; a step of using the catalyst upstream air-fuel ratio, the exhaust gas flow rate, the exhaust gas temperature, and the catalyst upstream state quantities as inputs, and estimating the catalyst temperature of the three-way catalyst, the oxygen storage capacity, the catalyst downstream air-fuel ratio, catalyst downstream NOx emissions, catalyst downstream HC emissions, and catalyst downstream CO emissions as catalyst downstream state quantities; and correcting the estimated amount of NOx emissions downstream of the catalyst, the amount of HC emissions downstream of the catalyst, and the amount of CO emissions downstream of the catalyst using a NOx sensor value detected by the downstream NOx sensor. State estimation method.

9. measuring the catalyst upstream air-fuel ratio, the exhaust gas flow rate upstream of the three-way catalyst, and the exhaust gas temperature; a first learning step of inputting the measured exhaust gas flow rate, the exhaust gas temperature, the catalyst upstream air-fuel ratio, and the measured catalyst upstream NOx emission amounts, the catalyst upstream HC emission amounts, and the catalyst upstream CO emission amounts as training data into a first machine learning model that estimates the catalyst upstream NOx emission amounts, the catalyst upstream HC emission amounts, and the catalyst upstream CO emission amounts as the catalyst upstream state quantities. The state quantity estimation method according to claim 8 .

10. measuring the catalyst downstream NOx emissions, the catalyst downstream HC emissions, and the catalyst downstream CO emissions; a second learning step of inputting the measured exhaust gas flow rate, the exhaust gas temperature, and the catalyst upstream air-fuel ratio, the catalyst upstream state quantities estimated by the first machine learning model, and the measured catalyst downstream NOx emissions, the catalyst downstream HC emissions, and the catalyst downstream CO emissions as training data into a second machine learning model that estimates the catalyst temperature of the three-way catalyst, the oxygen storage capacity, the catalyst downstream air-fuel ratio, the catalyst downstream NOx emissions, the catalyst downstream HC emissions, and the catalyst downstream CO emissions as the catalyst downstream state quantities. The state quantity estimation method according to claim 9 .

11. a first estimation step of inputting the catalyst upstream air-fuel ratio, the exhaust gas flow rate, and the exhaust gas temperature into the first machine learning model that has been trained in advance, and estimating the catalyst upstream NOx emission amount, the catalyst upstream HC emission amount, and the catalyst upstream CO emission amount output by the first machine learning model as the catalyst upstream state quantities. The state quantity estimation method according to claim 10.

12. a second estimation step of inputting the catalyst upstream air-fuel ratio, the exhaust gas flow rate, the exhaust gas temperature, and the catalyst upstream state quantities estimated by the first machine learning model into the second machine learning model that has been trained in advance, and estimating the catalyst temperature of the three-way catalyst, the oxygen storage capacity, the catalyst downstream air-fuel ratio, the catalyst downstream NOx emissions, the catalyst downstream HC emissions, and the catalyst downstream CO emissions that are output by the second machine learning model as the catalyst downstream state quantities. The state quantity estimation method according to claim 11.

13. a step of calculating a correction parameter that is a difference between the catalyst downstream NOx emission amount and the NOx sensor value detected by the downstream NOx sensor, based on the corrected catalyst downstream NOx emission amount and the NOx sensor value detected by the downstream NOx sensor; The estimated catalyst downstream NOx emissions, catalyst downstream HC emissions, and catalyst downstream CO emissions are corrected based on the correction parameters. The state quantity estimation method according to claim 12.

14. and diagnosing the NOx sensor based on the amount of NOx emission downstream of the catalyst corrected based on the correction parameter and a NOx sensor value detected by the downstream NOx sensor. The state quantity estimation method according to claim 13.

Images