Apparatus for estimating the amount of particulate matter deposited

Abstract

To improve the accuracy of estimation when predicting the amount of particulate matter deposited. [Solution] A trained model obtained by machine learning takes as input variables a first parameter that reflects the amount of fuel injected into the engine, a second parameter that reflects the amount of intake air into the engine, a third parameter that reflects combustion conditions different from the amount of fuel injected into the engine and the amount of intake air into the engine, and a fourth parameter that includes at least one of the presence or absence of knocking and the presence or absence of misfires in the engine, and as an output variable the amount of particulate matter deposited, and estimates the amount of particulate matter deposited based on the first, second, third, and fourth parameters.

Description

Technical Field

[0001] The present disclosure relates to an apparatus for estimating the deposition amount of particulate matter.

Background Art

[0002] Conventionally, as an apparatus for estimating the deposition amount of this kind of particulate matter, there has been proposed one used in an engine device including an engine having a filter for removing particulate matter in an exhaust system, which estimates the deposition amount of particulate matter in the filter (see, for example, Patent Document 1). In this apparatus, the deposition amount of particulate matter in the filter is estimated using the intake air temperature, the cylinder wall temperature of the engine, and the flow rate of the fluid flowing into the filter.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the above-described apparatus for estimating the deposition amount of particulate matter, further improvement in the estimation accuracy when estimating the deposition amount of particulate matter is recognized as an important problem.

[0005] The main object of the apparatus for estimating the deposition amount of particulate matter of the present disclosure is to improve the estimation accuracy when estimating the deposition amount of particulate matter.

Means for Solving the Problems

[0006] The particulate matter deposition amount estimation device of this disclosure employs the following means to achieve the above-mentioned main objective. The particulate matter deposition amount estimation device of this disclosure is used in an engine system equipped with an engine that has a filter for removing particulate matter attached to the exhaust system, and estimates the amount of particulate matter deposited in the filter, and the gist of the device is to estimate the amount of particulate matter deposited based on a trained model obtained by machine learning, which takes as input variables a first parameter that reflects the amount of fuel injection in the engine, a second parameter that reflects the amount of intake air in the engine, a third parameter that reflects combustion conditions different from the amount of fuel injection in the engine and the amount of intake air, and a fourth parameter that includes at least one of the presence or absence of knocking and the presence or absence of misfire in the engine, and the amount of particulate matter deposited as an output variable, and the first, second, third, and fourth parameters. [Brief explanation of the drawing]

[0007] [Figure 1] A schematic diagram of the engine device 10 according to the embodiment of this disclosure. [Figure 2] An explanatory diagram illustrating how to create a pre-trained model. [Figure 3] A flowchart showing an example of a decision routine executed by ECU70. [Modes for carrying out the invention]

[0008] Embodiments of this disclosure will be described with reference to the drawings. Figure 1 is a schematic diagram of the engine device 10 of an embodiment of this disclosure. Figure 2 is an explanatory diagram for illustrating how to create a trained model. As shown in the figures, the engine device 10 of the embodiment comprises an engine 12 and an electronic control unit (hereinafter referred to as "ECU") 70 that controls the engine 12. The engine device 10 is installed, for example, in an engine vehicle that runs using power from the engine 12, a hybrid vehicle that has a motor in addition to the engine 12, or in stationary equipment (for example, construction equipment) that operates using power from the engine 12.

[0009] Engine 12 is configured as a multi-cylinder internal combustion engine that uses fuel such as gasoline or a mixture of gasoline and alcohol (e.g., ethanol) to output power through four strokes: intake, compression, expansion (explosive combustion), and exhaust. By having port injection valves 26 and in-cylinder injection valves 27, engine 12 can be operated in any of the following modes: port injection mode, in-cylinder injection mode, and shared injection mode. In port injection mode, air cleaned by the air cleaner 22 is drawn into the intake manifold 23, passes through the throttle valve 24, and fuel is injected from the port injection valve 26 to mix the air and fuel. This mixture is then drawn into the combustion chamber 29 via the intake valve 28 and explodes and burns due to an electric spark from the spark plug 30. The energy from this explosive combustion converts the reciprocating motion of the piston 32, which is pushed down in the cylinder, into the rotational motion of the crankshaft 14. In the in-cylinder injection mode, air is drawn into the combustion chamber 29, similar to the port injection mode, and fuel is injected from the in-cylinder injection valve 27 during the intake and compression strokes. This fuel is then explosively combusted by an electric spark from the spark plug 30 to obtain rotational motion of the crankshaft 14. In the shared injection mode, fuel is injected from the port injection valve 26 when air is drawn into the combustion chamber 29, and also from the in-cylinder injection valve 27 during the intake and compression strokes. This fuel is then explosively combusted by an electric spark from the spark plug 30 to obtain rotational motion of the crankshaft 14. These injection modes are switched based on the operating state of the engine 12. The exhaust gas discharged from the combustion chamber 29 through the exhaust valve 33 to the exhaust pipe 35 (exhaust system) is then discharged to the outside air via a purification device 36 and a PM filter 37. The purification device 36 has a catalyst (three-way catalyst) 36a that purifies harmful components such as carbon monoxide (CO), hydrocarbons (HC), and nitrogen oxides (NOx) in the exhaust gas. The PM filter 37 captures particulate matter (PM: Particulate Matter) such as soot in the exhaust.

[0010] The ECU70 is a microcomputer equipped with a CPU, ROM, RAM, flash memory, input / output ports, and communication ports, as well as various drive circuits and various logic ICs. The ECU70 receives signals from various sensors via its input ports. For example, the ECU70 receives the crank angle θcr from the crank position sensor 14a, which detects the rotational position of the crankshaft 14 of the engine 12, and the coolant temperature Tw from the coolant temperature sensor 15, which detects the coolant temperature of the engine 12. The ECU70 also receives the cam angles θci and θco from the cam position sensor 16, which detects the rotational position of the intake camshaft that opens and closes the intake valve 28 and the rotational position of the exhaust camshaft that opens and closes the exhaust valve 33. The ECU70 also receives input from the intake air volume Qa and intake air temperature Ta from the airflow meter 23a and temperature sensor 23t, which are installed upstream of the throttle valve 24 in the intake manifold 23, the throttle opening TH from the throttle position sensor 24a, which detects the position of the throttle valve 24, and the surge pressure Ps from the surge pressure sensor 25a, which is installed in the surge tank 25. The ECU70 also receives input from the front air-fuel ratio AF1 and rear air-fuel ratio AF2 from the front air-fuel ratio sensor 37a and rear air-fuel ratio sensor 37b, which are installed upstream of the purification device 36 in the exhaust manifold 35 and between the purification device 36 and the PM filter 37, respectively, and the differential pressure ΔP (differential pressure between the upstream and downstream sides) in front of and behind the PM filter 37 from the differential pressure sensor 38a. The ECU70 also receives input from the fuel level Fl from the fuel level gauge 39, which is installed in the fuel tank and detects the remaining amount of fuel, and the knock signal Sn from the knock sensor 40, which detects knocking.

[0011] The ECU70 outputs various control signals via its output ports. For example, the ECU70 outputs control signals to the throttle valve 24 of the engine 12, to the in-cylinder injection valve 27, and to the spark plug 30.

[0012] The ECU70 performs various calculations. For example, the ECU70 calculates the rotational speed Ne of the engine 12 based on the crank angle θcr from the crank position sensor 14a. The ECU70 also calculates the load ratio KL of the engine 12 based on the intake air volume Qa from the airflow meter 23a and the rotational speed Ne of the engine 12. The load ratio KL is defined as the ratio of the volume of air actually inhaled in one cycle to the stroke volume per cycle of the engine 12.

[0013] When the engine unit 10 is mounted in a vehicle, the ECU 70 controls the operation of the engine 12 based on the accelerator opening and vehicle speed, including intake air volume control, fuel injection control, and ignition control. For intake air volume control, the required air volume Qa* is set based on the accelerator opening and vehicle speed, the target opening TH* of the throttle valve 24 is set so that the difference between the intake air volume Qa and the required air volume Qa* cancels out, and the throttle valve 24 is controlled so that the difference between the throttle opening TH and the target opening TH* cancels out. For fuel injection control, the basic injection amount Qfbs is set based on the intake air volume Qa so that the difference between the front air-fuel ratio AF1 and the target air-fuel ratio AF* cancels out, an enrichment coefficient kf is set, and the required injection amount Qf* is set as the product of the basic injection amount Qfbs and the enrichment coefficient kf. The enrichment coefficient kf is set to a value of 1 when no fuel enrichment is performed, and to a value greater than 1 when fuel enrichment is performed. Next, based on the load factor KL of the engine 12, a fuel injection ratio Rp is set, which indicates the ratio of the fuel injection amount from the port injection valve 26 to the fuel injection amount of the engine 12. The port injection valve 26 and the in-cylinder injection valve 27 are controlled so that fuel injection amount Qfp* (=Qf*Rp), which is the required injection amount Qf* multiplied by the fuel injection ratio Rp, is injected from the port injection valve 26, and fuel injection amount Qfd* (=Qf*(1-Rp)), which is the required injection amount Qf* multiplied by a value obtained by subtracting the fuel injection ratio Rp from 1, is injected from the in-cylinder injection valve 27. For ignition control, a target ignition timing Ti* is set based on the rotational speed Ne and the required air amount Qa*, and the spark plug 30 is controlled to perform ignition based on the target ignition timing Ti*.

[0014] When the ignition timing arrives, the ECU70 calculates the rotational speed fluctuation ΔNe (= rotational speed Ne - previous Neb) as a variable indicating the inter-cylinder torque imbalance output from the engine 12 by subtracting the rotational speed (previous) Neb at the previous ignition timing from the rotational speed Ne currently being calculated. For each ignition cycle in which the crank angle θcr of the crankshaft 14 of the engine 12 changes by a predetermined angle and the ignition cylinder changes (for example, every 180 degrees if the engine 12 is a 4-cylinder engine), the ECU70 calculates the 30-degree time T30, which is the time required for the crank angle θcr of the crankshaft 14 to rotate 30 degrees from top dead center. Subsequently, the ECU70 calculates the change in time ΔT30 by subtracting the 30-degree time T30 calculated one ignition cycle ago from the most recently calculated 30-degree time T30. When the change in required time ΔT30 is less than the misfire detection threshold ΔT30ref, it is determined that no misfire has occurred in the engine 12. When the change in required time ΔT30 is greater than or equal to the misfire detection threshold ΔT30ref, it is determined that a misfire has occurred in the engine 12. Hereinafter, this determination will be referred to as "misfire determination".

[0015] The ECU70 calculates the fuel consumption Pc based on the remaining fuel amount Fl from the fuel gauge 39.

[0016] The ECU70 stores in ROM a trained model obtained through machine learning, with the following as input variables (input to the input layer): a fuel injection amount parameter (first parameter) P1 that reflects the amount of fuel injected in the engine 12; an intake air amount parameter (second parameter) P2 that reflects the amount of intake air in the engine 12; a combustion condition parameter (third parameter) P3 that reflects combustion conditions different from the fuel injection amount and intake air amount in the engine 12; a combustion abnormality parameter (fourth parameter) P4 that indicates whether or not there are combustion abnormalities such as knocking and misfires in the engine 12; a combustion result parameter (fifth parameter) P5 that reflects the imbalance between cylinders in the engine 12; and a fuel type parameter (sixth parameter) P6 that reflects the type of fuel used in the engine 12; and the PM deposit amount Vpm, which is the amount of particulate matter deposited, as an output variable (output to the output layer). The fuel injection quantity parameter P1 can be at least one of the following: the required injection quantity Qf*, the injection quantity Qfp* from the port injection valve 26, the injection quantity Qfd* from the in-cylinder injection valve 27, or the enrichment coefficient kf. The intake air quantity parameter P2 can be at least one of the following: the intake air quantity Qa from the airflow meter 23a, the throttle opening TH from the throttle position sensor 24a, or the load factor KL. The combustion condition parameter P3 can be the target ignition timing Ti*. The combustion abnormality parameter P4 can be at least one of the following: the result of the knocking determination based on the knock signal Sn from the knock sensor 40, or the result of the misfire determination. The combustion result parameter P5 can be at least one of the following: the rotational speed fluctuation ΔNe, the front air-fuel ratio AF1 from the front air-fuel ratio sensor 37a, or the fuel consumption Fc. The fuel type parameter P6 can be at least one of the following: the fuel octane number or the type of additive. The PM accumulation amount Vpm is calculated based on the differential pressure ΔP from the differential pressure sensor 38a. In this machine learning approach, the tester operates the engine 12 of the engine unit 10 and acquires a dataset that includes the fuel injection amount parameter P1, the intake air amount parameter P2, the combustion condition parameter P3, the combustion abnormality parameter P4, the combustion result parameter P5, the fuel type parameter P6, and the PM accumulation amount Vpm.The ECU70 repeatedly acquires this dataset at predetermined intervals. Once the ECU70 has acquired the number of datasets necessary to build a highly accurate neural network, it creates a trained model using a neural network that takes the fuel injection amount parameter P1, the intake air amount parameter P2, the combustion condition parameter P3, the combustion abnormality parameter P4, the combustion result parameter P5, and the fuel type parameter P6 as input variables (as shown in Figure 2), and outputs the PM accumulation amount Vpm as an output variable (as shown in Figure 2), and stores it in ROM.

[0017] Next, the operation of the engine device 10 of the embodiment configured in this way, in particular, the operation when estimating the PM accumulation amount Vpm, will be described. Figure 3 is a flowchart of an example of a determination routine executed by the ECU 70. This routine is executed at predetermined intervals while the engine 12 is running.

[0018] When this routine is executed, the CPU of the ECU70 inputs the fuel injection amount parameter P1, the intake air amount parameter P2, the combustion condition parameter P3, the combustion abnormality parameter P4, the combustion result parameter P5, and the fuel type parameter P6 (S100).

[0019] Then, using the pre-trained model stored in ROM, the fuel injection amount parameter P1, the intake air amount parameter P2, the combustion condition parameter P3, the combustion anomaly parameter P4, the combustion result parameter P5, and the fuel type parameter P6, the PM deposition amount Vpm is estimated (S110), and this routine ends. By estimating the PM deposition amount using the pre-trained model created by machine learning and the input including the combustion anomaly parameter P4, the estimation accuracy when estimating the deposition amount of particulate matter is improved.

[0020] In the above-described embodiment, as the engine 22, an exhaust gas recirculation device having an EGR pipe connected to the downstream of the catalyst 36a for supplying exhaust gas to the intake surge tank 25 on the intake side and an EGR valve disposed in the EGR pipe and driven by a stepping motor may be used, and the EGR rate Re may be used as an input. The EGR rate Re is the ratio of the EGR amount Qe, which is the amount of exhaust gas refluxing to the intake pipe 23, to the sum of the intake air amount Qa from the air flow meter 23a and the EGR amount Qe. The EGR amount Qe is calculated based on the intake air amount Qa, the calculated rotational speed Ne, and the opening degree EV of the EGR valve.

[0021] In the above-described embodiment, at least the fuel injection amount parameter P1, the intake air amount parameter P2, the combustion condition parameter P3, and the combustion abnormality parameter P4 may be used as inputs when creating the learned model.

[0022] As described above, the embodiments have been used to explain the embodiments for implementing the present disclosure. However, the present disclosure is not limited to such embodiments, and it is needless to say that the present disclosure can be implemented in various forms without departing from the gist of the present disclosure.

Industrial Applicability

[0023] The present disclosure can be used in the manufacturing industry of particulate matter deposition amount estimation devices and the like.

Explanation of Reference Numerals

[0024] 10 Engine device, 12 Engine, 70 Electronic control unit (ECU).

Claims

[Claim 1] A particulate matter deposition amount estimation device used in an engine system equipped with an engine that has a filter attached to the exhaust system to remove particulate matter, which estimates the amount of particulate matter deposited in the filter, A trained model obtained by machine learning, with input variables being a first parameter reflecting the fuel injection amount in the engine, a second parameter reflecting the intake air amount in the engine, a third parameter reflecting combustion conditions different from the fuel injection amount and intake air amount in the engine, and a fourth parameter including at least one of the presence or absence of knocking and the presence or absence of misfires in the engine, and output variables being the amount of deposition of particulate matter, and the amount of deposition of particulate matter being estimated based on the first, second, third, and fourth parameters. A device for estimating the amount of particulate matter deposited.

Images