Control device for an internal combustion engine system

Abstract

To provide a control device of an internal combustion engine system capable of early updating a learning value to correct a fuel injection amount, and more efficiently purifying an exhaust gas by early eliminating a deviation between an actual air fuel ratio and a theoretical air fuel ratio.SOLUTION: A control device corrects a fuel injection amount obtained by stoichiometric control using an air-fuel ratio feedback control by using a learning value. The control device has: a forced learning start determination portion for starting a forced learning mode to forcibly calculate the learning value in a case when a forced learning execution order for forcibly calculating the learning value is input from a user or when feeding of a fuel is determined; a forced learning value calculation / storage portion for calculating the learning values on the basis of a state of the air-fuel ratio feedback control in an operation state as a corresponding learning region and storing the same in a nonvolatile storage device; and a forced learning releasing portion for releasing a forced learning mode in a case when the calculation and storage of the learning value are terminated.SELECTED DRAWING: Figure 14

Description

Technical Field

[0001] The present invention relates to a control device for an internal combustion engine system that adjusts the fuel injection amount to an internal combustion engine using air-fuel ratio feedback control.

Background Art

[0002] For example, in a forklift or the like equipped with an LPG engine (natural gas engine), as a measure against exhaust gas, the fuel amount is adjusted to be near the stoichiometric air-fuel ratio, and the exhaust gas is purified using a three-way catalyst provided in the exhaust path. In order to adjust the fuel amount to be near the stoichiometric air-fuel ratio, an air-fuel ratio related information detection device (O2 sensor or A / F sensor) is provided in the exhaust path. Then, an injector is arranged in the intake path so as to achieve the target air-fuel ratio, and based on the detection signal (air-fuel ratio related information) from the air-fuel ratio related information detection device (O2 sensor or A / F sensor), the fuel amount from the injector is adjusted by air-fuel ratio feedback control.

[0003] In the case of an LPG engine, since the fuel amount is insufficient only with the fuel amount supplied from the injector, a fuel mixer that mixes fuel into the intake air is arranged in the intake path separately from the injector. The fuel mixer has a needle that can adjust the opening degree of a fuel pipe connected from an LPG regulator to the intake path, and a stepping motor that drives the needle.

[0004] A control device for an internal combustion engine system using natural gas, gasoline, or the like as fuel controls each actuator that adjusts the operating state based on detection signals from various detection devices (various sensors) that detect various operating states of the internal combustion engine, so that the actual air-fuel ratio is near the stoichiometric air-fuel ratio. However, the composition of fuels such as natural gas and gasoline varies due to various factors such as season and region, and there are individual differences in each detection device and each actuator, and they deteriorate over time with long-term use, so the actual air-fuel ratio is likely to deviate from the stoichiometric air-fuel ratio.

[0005] Even if the actual air-fuel ratio deviates from the stoichiometric air-fuel ratio, it is possible to make it follow near the stoichiometric air-fuel ratio by air-fuel ratio feedback control. However, when the deviation is large, the time required to follow near the stoichiometric air-fuel ratio by air-fuel ratio feedback control becomes long, and during that time, there is a possibility that the exhaust gas may not be sufficiently purified. Also, generally, air-fuel ratio feedback control starts after the internal combustion engine has completed warm-up. Therefore, during warm-up, it is not possible to make it follow near the stoichiometric air-fuel ratio by air-fuel ratio feedback control, and there is a possibility that the exhaust gas may not be sufficiently purified during warm-up. Thus, conventionally, learning control for correcting the fuel injection amount has been performed to eliminate the deviation between the actual air-fuel ratio and the stoichiometric air-fuel ratio.

[0006] For example, Patent Document 1 discloses a control device for a natural gas internal combustion engine that performs air-fuel ratio feedback control to calculate an air-fuel ratio correction coefficient according to the output of an oxygen concentration sensor (O2 sensor) provided in the exhaust system of the internal combustion engine and uses this to adjust the fuel injection amount. In Cited Document 1, furthermore, the average value of the air-fuel ratio correction coefficient is used as a parameter indicating the fuel composition, and the ignition timing is also corrected accordingly.

Prior Art Documents

Patent Documents

[0007]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0008] Although there is no particular description of the learning control for correcting the fuel injection amount in the control device for an internal combustion engine for natural gas described in Patent Document 1, generally, during steady operation without acceleration or deceleration after warm-up of the internal combustion engine, the air-fuel ratio feedback control is continuously performed, and a learning value is calculated based on the operation and state of the air-fuel ratio correction coefficient. Therefore, when the internal combustion engine has been warmed up and steady operation has continued for a relatively long time, the learning value is calculated, so it takes time until the learning value is calculated. In a forklift or the like that frequently performs acceleration and deceleration, the learning frequency becomes very low, which is not very preferable. In particular, when the composition of the replenished fuel has a relatively large variation when refueling, the deviation between the actual air-fuel ratio and the theoretical air-fuel ratio becomes large until the learning value is updated, which is not preferable.

[0009] The present invention has been devised in view of such points, and an object thereof is to provide a control device for an internal combustion engine system that can update a learning value for correcting a fuel injection amount at an early stage, eliminate the deviation between the actual air-fuel ratio and the theoretical air-fuel ratio at an early stage, and purify exhaust gas more efficiently.

Means for Solving the Problems

[0010] To solve the above problems, a first invention of the present invention is a control device for an internal combustion engine system, wherein the internal combustion engine system includes an internal combustion engine, an air-fuel ratio related information detection device that detects air-fuel ratio related information related to the air-fuel ratio of the exhaust of the internal combustion engine, and the control device that adjusts the fuel injection amount to the internal combustion engine by stoichiometric control using air-fuel ratio feedback control based on the operating state including the air-fuel ratio related information in the internal combustion engine. a throttle device for adjusting a throttle valve of the internal combustion engine from the control device, It has. And the control device the intake air amount of the internal combustion engine has a plurality of learning regions corresponding thereto, stores the learning values of each of the learning regions in a non-volatile storage device, and adjusts the fuel injection amount to the internal combustion engine according to the operating state. the intake air amount of the internal combustion engine It is corrected using the learning value corresponding to the learning area according to the situation. Further, when an instruction to execute forced learning for forcibly calculating the learning value for correcting the fuel injection amount is input from the user, or when the fuel supply detection unit that detects fuel replenishment determines that fuel has been replenished, the control device starts a forced learning mode for forcibly calculating the learning value of each of the learning areas, and when the forced learning mode is started, the learning value of each of the learning areas is calculated based on the state of the air-fuel ratio feedback control based on the air-fuel ratio related information in the corresponding learning area, and the calculated learning value of each of the learning areas is stored in the non-volatile storage device. A forced learning value calculation / storage unit, and a forced learning cancellation unit that cancels the forced learning mode when the calculation and storage of the learning value of each of the learning areas are completed in the forced learning value calculation / storage unit. the intake air amount of the internal combustion engine It is calculated based on the state of the air-fuel ratio feedback control based on the air-fuel ratio related information in the intake air amount of the internal combustion engine , and the calculated learning value of each of the learning areas is stored in the non-volatile storage device. and. In the reinforcement learning mode, the control device automatically changes the intake air amount of the internal combustion engine by adjusting the throttle device so as to be each of the learning regions in the reinforcement learning value calculation and storage unit, and calculates and stores the learning values of each of the learning regions. It is a control device for an internal combustion engine system.

[0011] Next, the second invention of the present invention is A control device for an internal combustion engine system according to the first invention, wherein the internal combustion engine system has a load device capable of adjusting a load of the internal combustion engine from the control device. The control device has a plurality of learning regions corresponding to the rotational speed and intake air amount of the internal combustion engine, stores the learning values of each of the learning regions in a non-volatile storage device, and corrects the fuel injection amount to the internal combustion engine according to the operating state using the learning value corresponding to the learning region corresponding to the rotational speed and intake air amount of the internal combustion engine. When the reinforcement learning mode is started, the learning values of each of the learning regions are calculated based on the state of the air-fuel ratio feedback control based on the air-fuel ratio related information at the rotational speed and intake air amount of the internal combustion engine corresponding to the corresponding learning region, and the calculated learning values of each of the learning regions are stored in a non-volatile storage device. The control device has a reinforcement learning value calculation and storage unit for storing. In the reinforcement learning mode, the control device calculates and stores the learning values of each of the learning regions while automatically changing the rotational speed and intake air amount of the internal combustion engine by adjusting the throttle device and the load device so as to be each of the learning regions. It is a control device for an internal combustion engine system.

[0012] Next, the third invention of the present invention is the control device for the internal combustion engine system according to the first invention above Or the second invention wherein the internal combustion engine system has an accelerator pedal operated by the user, and the control device, before starting the forced learning mode in the forced learning value calculation / storage unit, selects a plurality of the learning areas according to the accelerator pedal depression amount, which is the depression amount of the user's accelerator pedal, from among the plurality of the learning areas for selects a plurality of the learning areas to do It is a control device for an internal combustion engine system.

[0013] Next, the fourth invention of the present invention is the above any one of the first invention to the third inventionA control device for an internal combustion engine system according to the present invention, wherein the control device has a forced warm-up promotion unit that automatically promotes warm-up when it is determined that the warm-up of the internal combustion engine has not been completed before starting the reinforcement learning mode. When the forced warm-up promotion unit promotes warm-up, when the warm-up is completed, the promotion of warm-up is terminated and the process proceeds to the process of the reinforcement learning value calculation and storage unit. It is a control device for an internal combustion engine system.

[0014] Next, a fifth invention of the present invention is A control device for an internal combustion engine system, the internal combustion engine system including: an internal combustion engine; an air-fuel ratio related information detection device that detects air-fuel ratio related information related to the air-fuel ratio of the exhaust of the internal combustion engine; the control device that adjusts the fuel injection amount to the internal combustion engine by stoichiometric control using air-fuel ratio feedback control based on the operating state including the air-fuel ratio related information in the internal combustion engine; an accelerator pedal operated by a user; and a throttle opening adjustment device that adjusts the opening of the throttle valve of the internal combustion engine according to the operation amount of the accelerator pedal by the user. And the control device has a plurality of learning regions corresponding to the intake air amount of the internal combustion engine, stores the learning values of each of the learning regions in a non-volatile storage device, and corrects the fuel injection amount to the internal combustion engine according to the operating state using the learning values corresponding to the learning regions corresponding to the intake air amount of the internal combustion engine. Further, when an instruction to forcibly calculate the learning value for correcting the fuel injection amount is input from the user, or when the fuel supply detection unit that detects fuel supply determines that fuel has been supplied, the control device starts a forced learning mode for forcibly calculating the learning values of each of the learning regions, has a forced learning value calculation and storage unit that, when the forced learning mode is started, calculates the learning values of each of the learning regions based on the state of the air-fuel ratio feedback control based on the air-fuel ratio related information at the intake air amount of the internal combustion engine corresponding to the respective learning regions, and stores the calculated learning values of each of the learning regions in a non-volatile storage device, and a forced learning release unit that releases the forced learning mode when the calculation and storage of the learning values of each of the learning regions are completed by the forced learning value calculation and storage unit. And, The internal combustion engine system has an information output device capable of outputting an instruction to the user, In the forced learning mode, The control device uses the information output device to output an instruction to the user in the forced learning value calculation and storage unit so as to be in each of the learning regions, the operation amount of the accelerator pedal corresponding to the intake air amount of each learning region Output an instruction, the intake air amount of the internal combustion engine corresponding to the operation amount of the accelerator pedal It is a control device for an internal combustion engine system that calculates and stores the learning values of the respective learning regions while allowing the user to change them.

[0015] Next, a sixth invention of the present invention is the above Fifth A control device for an internal combustion engine system according to the invention, 、 Before starting the reinforcement learning mode, the control device selects a plurality of the learning regions according to the accelerator pedal depression amount, which is the depression amount of the user's accelerator pedal, from among the plurality of learning regions in the forced learning value calculation and storage unit. for Select a plurality of the learning regions doing It is a control device for an internal combustion engine system.

[0016] Next, a seventh invention of the present invention is a control device for an internal combustion engine system according to the fifth or sixth invention above, wherein the control device has a forced warm-up promotion unit that outputs a warm-up promotion instruction to the user using the information output device when it is determined that the warm-up of the internal combustion engine has not been completed before starting the reinforcement learning mode. When the forced warm-up promotion unit promotes warm-up, when the warm-up is completed, the promotion of warm-up is terminated and the process proceeds to the process of the forced learning value calculation and storage unit. It is a control device for an internal combustion engine system.

Advantages of the Invention

[0017] According to the first invention, when a forced learning execution instruction is input from a user, or when it is determined that there has been fuel replenishment, learning values for respective learning regions corresponding to the driving state are calculated and stored in a non-volatile memory device. Thereby, it is possible to update the learning value for correcting the fuel injection amount at an early stage, eliminate the deviation between the actual air-fuel ratio and the theoretical air-fuel ratio at an early stage, and purify the exhaust gas more efficiently.

[0018] First According to the second invention, when calculating and storing the learning values for respective learning regions corresponding to the driving state, the control device calculates and stores the learning values for respective learning regions while automatically changing the driving state of the internal combustion engine. Therefore, the user does not need to change (adjust) the driving state of the internal combustion engine, and it is convenient because the learning values for respective learning regions can be calculated and stored without trouble.

[0019] According to the third invention, before starting the forced learning mode, the user can specify a desired learning region from among a plurality of learning regions by the depression amount of the accelerator pedal. Therefore, the user can calculate and store only the learning values for the learning regions considered necessary, omit wasted time, and make the time required for the forced learning mode shorter. In addition, it is convenient because the control device calculates and stores the learning values for respective learning regions while automatically changing the driving state of the internal combustion engine.

[0020] It is preferable to execute the forced learning mode in a state where the warm-up of the internal combustion engine is completed. According to the fourth invention, it is convenient because the warm-up of the internal combustion engine can be automatically completed before starting the forced learning mode.

[0021] According to the fifth invention, when calculating and storing the learning values of respective learning regions according to the operating state, in the case of a system that requires the user to change the operating state of the internal combustion engine, an instruction to the user for changing the operating state can be appropriately given by the information output device. Further, compared with a system in which the control device automatically changes the operating state of the internal combustion engine, the system can be configured simply.

[0022] According to the sixth invention, before starting the reinforcement learning mode, the user can specify a desired learning region from among a plurality of learning regions by the amount of depression of the accelerator pedal. Therefore, since the user can calculate and store only the learning values of the learning regions considered necessary, wasteful time can be omitted and the time required for the reinforcement learning mode can be made shorter. Note that an instruction to the user for changing the operating state can be appropriately given by the information output device.

[0023] It is preferable to execute the reinforcement learning mode in a state where the warm-up of the internal combustion engine is completed. In the case of a system that requires the user to change the operating state of the internal combustion engine, according to the seventh invention, before starting the reinforcement learning mode, an instruction to the user for promoting the warm-up of the internal combustion engine can be output using the information output device, which is convenient.

Brief Description of the Drawings

[0024]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Figure 21

Figure 22

Embodiments for Carrying Out the Invention

[0025] ●[Example of the Schematic Configuration of the LPG Engine System 1 (FIG. 1)] Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. First, with reference to FIG. 1, an example of the schematic configuration of the LPG engine system 1 will be described. In the description of this embodiment, the LPG engine system 1 mounted on a forklift will be described as an example of an internal combustion engine system.

[0026] Hereinafter, the overall schematic configuration of the LPG engine system 1 (corresponding to an internal combustion engine system) will be described in order from the intake side to the exhaust side. Note that the intake pipes 11A, 11B, and the intake manifold 11C correspond to the intake path, and the exhaust manifold 12A and the exhaust pipe 12B correspond to the exhaust path.

[0027] An air cleaner 21 is provided in the intake pipe 11A. Also, the inflow side of the intake pipe 11B is connected to the outflow side of the intake pipe 11A.

[0028] An intake air flow detection device 15 is provided in the intake pipe 11B. The intake air flow detection device 15 is, for example, an intake air flow sensor, and outputs a detection signal corresponding to the flow rate of the intake air flowing in the intake pipe 11B to the control device 60. The control device 60 can detect the flow rate of the intake air based on the detection signal from the intake air flow detection device 15.

[0029] In addition, a fuel pipe 33 of a fuel mixer 30 is connected to the intake pipe 11B. The fuel mixer 30 is connected to a main fuel pipe 51 from an LPG regulator 50, and fuel (fuel gas) is supplied from the LPG regulator 50. The fuel supplied from the main fuel pipe 51 flows from the fuel pipe 33 into the intake pipe 11B and is mixed with the intake air. Further, a fuel tank 53 is connected to the LPG regulator 50 to supply fuel. A fuel quantity detection device 27 is provided in the fuel tank 53. The fuel quantity detection device 27 is, for example, a fuel quantity sensor, and outputs a detection signal corresponding to the quantity of fuel in the fuel tank 53 to a control device 60. The control device 60 can detect the quantity of fuel in the fuel tank 53 based on the detection signal from the fuel quantity detection device 27.

[0030] The fuel mixer 30 is also provided with a step motor 31 and a needle 32. The control device 60 controls the position of the needle 32 by controlling the step motor 31, and adjusts the opening degree of the fuel pipe 33, thereby adjusting the quantity of fuel mixed into the intake air. The fuel mixer 30 is provided in the intake path, mixes fuel with the intake air flowing in the intake path, and the quantity of fuel to be mixed can be adjusted by the step motor 31.

[0031] A throttle device 43 is provided on the downstream side of the intake air with respect to the connection location of the fuel pipe 33 in the intake pipe 11B. The throttle device 43 is provided with a throttle valve 22 that adjusts the opening degree of the intake pipe 11B by an electric motor or the like according to the depression amount of the accelerator pedal by the user. The control device 60 detects the depression amount of the accelerator pedal by the user based on the detection signal from an accelerator pedal depression amount detection device 16 (such as an accelerator pedal depression amount sensor), obtains the target opening degree of the throttle valve 22 according to the detected depression amount, and outputs a control signal to the throttle device 43 so that the opening degree of the throttle valve 22 becomes the target opening degree.

[0032] An injector 40 is provided on the downstream side of the throttle valve 22 in the intake pipe 11B with respect to the intake air. An auxiliary fuel pipe 52 from an LPG regulator 50 is connected to the injector 40, and fuel (fuel gas) is supplied from the LPG regulator 50. The fuel supplied from the auxiliary fuel pipe 52 is injected from the injector 40 into the interior of the intake pipe 11B. The control device 60 can adjust the amount of fuel (fuel injection amount) injected from the injector by controlling the injector 40.

[0033] On the outflow side of the intake pipe 11B, the inflow side of an intake manifold 11C is connected. Also, the outflow side of the intake manifold 11C is connected to the intake port of an LPG engine 10 (corresponding to an internal combustion engine). And a pressure detection device 23 (for example, a pressure sensor) is provided in the intake manifold 11C. The control device 60 can detect the pressure of the intake air in the intake manifold 11C based on a detection signal from the pressure detection device 23.

[0034] An LPG engine 10 is provided with a rotation detection device 25 and a coolant temperature detection device 26. The rotation detection device 25 is, for example, a rotation sensor, and outputs a detection signal corresponding to the rotation angle of the crankshaft of the LPG engine 10 (that is, the engine speed) to the control device 60. The control device 60 can detect the engine speed of the LPG engine 10 and the rotation angle of the crankshaft, etc. based on the detection signal from the rotation detection device 25. The coolant temperature detection device 26 is, for example, a temperature sensor, and outputs a detection signal corresponding to the temperature of the coolant circulated in the LPG engine 10 to the control device 60. The control device 60 can detect the temperature of the coolant based on the detection signal from the coolant temperature detection device 26.

[0035] The LPG engine system 1 also has a load device 45 that can adjust the load of the LPG engine 10. The load device 45 is a device that serves as a load on the LPG engine 10. For example, in the case of an alternator for power generation or a hydraulic pump in the case of a forklift, it corresponds to the load device. The control device 60 can adjust the increase or decrease of the load on the LPG engine 10 by outputting a control signal to the load device 45 and adjusting the operation of the load device 45.

[0036] The LPG engine 10 is also provided with a spark plug 42. The control device 60 outputs a ignition signal to the spark plug 42 via the igniter 41.

[0037] The inflow side of the exhaust manifold 12A is connected to the exhaust port of the LPG engine 10, and the inflow side of the exhaust pipe 12B is connected to the outflow side of the exhaust manifold 12A.

[0038] An air-fuel ratio related information detection device 24 (for example, an O2 sensor or an A / F sensor) is provided in the exhaust pipe 12B. The O2 sensor outputs a lean detection signal when the exhaust gas is in a lean state where oxygen is excessive with respect to the stoichiometric air-fuel ratio, and outputs a rich detection signal when the exhaust gas is in a rich state where fuel is excessive with respect to the stoichiometric air-fuel ratio. The A / F sensor outputs a detection signal according to the air-fuel ratio of the exhaust gas (including the lean state and the rich state). That is, the air-fuel ratio related information detection device 24 outputs air-fuel ratio related information related to the lean state and the rich state to the control device 60. The control device 60 can detect the value of the air-fuel ratio (in the case of the A / F sensor) or whether the air-fuel ratio is in a rich state or a lean state (in the case of the O2 sensor) based on the air-fuel ratio related information from the air-fuel ratio related information detection device 24.

[0039] A three-way catalyst 70 (exhaust purification device) is connected to the outflow side of the exhaust pipe 12B. In order to efficiently purify the exhaust gas using the three-way catalyst 70, it is necessary to adjust the fuel injection amount so as to be near the stoichiometric air-fuel ratio.

[0040] The LPG engine system 1 is also provided with a forced learning instruction device 28 for inputting from the user a forced learning execution instruction for forcibly calculating a learning value for correcting the fuel injection amount. The forced learning instruction device 28 is, for example, a forced learning button, and when the control device 60 detects that the forced learning instruction device 28 has been operated, it determines that a forced learning execution instruction has been input from the user.

[0041] The LPG engine system 1 is also provided with information output devices 29A and 29B capable of outputting various information from the control device 60 to the user. Examples of the information output from the information output devices 29A and 29B include instruction information from the control device 60 to the user. The information output device 29A is, for example, a display, and can display characters or the like based on a control signal from the control device 60 to output an instruction to the user. The information output device 29B is, for example, a speaker, and can output a voice based on a control signal from the control device 60 to output an instruction to the user.

[0042] The control device 60 includes a CPU 61, a RAM 62, a storage device 63 (ROM), a timer 64, a non-volatile storage device 65, and the like. Detection signals from the various detection devices described above are input to the control device 60 (CPU 61), and the control device 60 (CPU 61) outputs control signals to the various actuators described above. Note that the input and output of the control device 60 are not limited to the above detection devices and actuators. Also, the temperature, pressure, etc. of each part may be calculated by estimation without mounting a sensor. The control device 60 detects the operating state of the LPG engine 10 based on detection signals from various detection devices including the above detection devices, and controls various actuators including the above actuators. The storage device 63 is, for example, a storage device such as a Flash-ROM, and stores programs, data, etc. for controlling the LPG engine 10 and performing self-diagnosis and the like. The control device 60 (CPU 61) also includes a fuel supply detection unit 61A, a forced learning start determination unit 61B, a forced warm-up promotion unit 61C, a forced learning value calculation and storage unit 61D, a forced learning cancellation unit 61E, etc., and the details of these will be described later.

[0043] ●[Base Step Number and Base Injection Quantity (Figs. 2 to 4)] The amount of fuel supplied to the LPG engine 10 is the sum of the amount of fuel mixed into the intake air from the fuel mixer 30 and the amount of fuel injected from the injector 40.

[0044] For example, in the storage device 63, a base step number characteristic (see Fig. 2) in which a base step number (Sxx) corresponding to the rotational speed (N1, N2, ···) of the LPG engine and the intake air pressure (P1, P2, ···) in the intake manifold 11C is set is stored. The control device 60 detects the operating state of the LPG engine, and controls the step motor 31 of the fuel mixer 30 based on the base step number obtained according to the operating state, and adjusts the amount of fuel mixed into the intake air from the fuel mixer 30. Note that the base step number characteristic is not limited to the characteristic shown in Fig. 2 corresponding to the rotational speed of the LPG engine and the intake air pressure in the intake manifold, and may be set according to various operating states.

[0045] Also, for example, in the storage device 63, a base injection quantity characteristic (see Fig. 3) in which a base injection quantity (Bxx) corresponding to the rotational speed (N1, N2, ···) of the LPG engine and the intake air pressure (P1, P2, ···) in the intake manifold is set is stored. The control device 60 detects the operating state of the LPG engine, and controls the injector 40 based on the base injection quantity obtained according to the operating state, and adjusts the amount of fuel (fuel injection quantity) injected into the intake pipe 11B from the injector 40. Note that the base injection quantity characteristic is not limited to the characteristic shown in Fig. 3 corresponding to the rotational speed of the LPG engine and the intake air pressure in the intake manifold, and may be set according to various operating states.

[0046] Figure 4 shows an image of the fuel quantity supplied to the LPG engine 10. As described above, for the LPG engine, the total fuel quantity Q, which is the sum of the fuel quantity M mixed into the intake air from the fuel mixer 30 and the fuel quantity B (base injection quantity) injected from the injector, is supplied. Further, the control device 60 can instantaneously and highly accurately adjust the fuel quantity B (base injection quantity) injected from the injector 40 between the lower limit quantity Bmin and the upper limit quantity Bmax in comparison with the fuel quantity M from the fuel mixer 30 according to the air-fuel ratio or the like.

[0047] As described above, the control device 60 adjusts the total fuel quantity so as to be near the stoichiometric air-fuel ratio according to the operating state of the LPG engine 10. However, due to variations in the fuel composition, individual differences and aging deterioration of various actuators and sensors, etc., the actual air-fuel ratio is likely to deviate from the stoichiometric air-fuel ratio. To eliminate this deviation, the control device 60 performs stoichiometric control using air-fuel ratio feedback control based on the air-fuel ratio related information detected by the air-fuel ratio related information detection device 24.

[0048] Even if the actual air-fuel ratio deviates from the stoichiometric air-fuel ratio, it is possible to make it follow near the stoichiometric air-fuel ratio by air-fuel ratio feedback control. However, when the deviation is large, the time required to follow by air-fuel ratio feedback control becomes long, and during that time, there is a possibility that the exhaust gas is not sufficiently purified. Also, generally, air-fuel ratio feedback control is started after the internal combustion engine has completed warm-up. Therefore, during warm-up, it is not possible to make it follow near the stoichiometric air-fuel ratio by air-fuel ratio feedback control, and there is a possibility that the exhaust gas cannot be sufficiently purified during warm-up. Thus, conventionally, learning control for obtaining a learning value for correcting the fuel injection quantity has been performed so as to eliminate the deviation between the actual air-fuel ratio and the stoichiometric air-fuel ratio.

[0049] In general learning control, during steady operation where acceleration and deceleration are not performed after warm-up of the internal combustion engine, air-fuel ratio feedback control is continuously performed, and a learning value is calculated based on the operation and state of the air-fuel ratio correction coefficient. Therefore, when the internal combustion engine has been warmed up and steady operation has continued for a relatively long time, the learning value is calculated, so it takes time until the learning value is calculated. In a forklift or the like that frequently performs acceleration and deceleration, the learning frequency becomes extremely low, which is not very preferable. Therefore, the control device 60 performs the "forced learning" process described below to update the learning value for correcting the fuel injection amount at an early stage, eliminate the deviation between the actual air-fuel ratio and the theoretical air-fuel ratio at an early stage, and purify the exhaust gas more efficiently.

[0050] ●[Processing procedure of the control device 60 (Figs. 5 to 13) and example of operation waveform (Fig. 14) in the first embodiment] ●[Overall processing of forced learning (Fig. 5) (First embodiment)] Next, an example of the processing procedure of "forced learning" by the control device 60 in the first embodiment will be described using the flowcharts and the like shown in Figs. 5 to 13. In the first embodiment, in the LPG engine system 1 described with reference to Fig. 1, the control device 60 controls the throttle device 43 to automatically change the operation state while obtaining each learning value. Note that Fig. 14 shows an example of the operation waveform as a result of performing the processing shown in Figs. 5 to 13.

[0051] The control device 60 (CPU 61) starts the processing shown in Fig. 5 at a predetermined time interval (for example, at intervals of several [ms]) after starting the LPG engine 10 (and when the vehicle is stopped (vehicle speed = 0), and in the P or N range, and the parking brake is ON), and proceeds to step S010.

[0052] In step S010, the control device 60 executes [fuel supply determination processing] and proceeds to step S015. The details of [fuel supply determination processing] will be described later.

[0053] In step S015, the control device 60 determines whether the reinforcement learning execution flag is ON. If the reinforcement learning execution flag is ON (Yes), the control device 60 proceeds to step S035. If it is not ON (No), the control device 60 proceeds to step S020. The reinforcement learning execution flag is a flag that is set to ON in step S030 in FIG. 5 and set to OFF in step S080, and is a flag that is set to ON during the execution of "reinforcement learning" (see "Reinforcement Learning Execution Flag" in FIG. 14).

[0054] When the process proceeds to step S020, the control device 60 determines whether the reinforcement learning button (reinforcement learning instruction device 28) has been operated to be ON by the user. If the reinforcement learning button has been operated to be ON (Yes), the control device 60 proceeds to step S030. Otherwise (No), the control device 60 proceeds to step S025.

[0055] When the process proceeds to step S025, the control device 60 determines whether the fuel replenishment flag is ON. If the fuel replenishment flag is ON (Yes), the control device 60 proceeds to step S030. Otherwise (No), the process shown in FIG. 5 ends. The fuel replenishment flag is a flag that is set to ON when it is determined in step S010 of FIG. 5 that the fuel has been replenished, and is set to OFF in step S030, and is a flag stored in the non-volatile memory device.

[0056] When the process proceeds to step S030, the control device 60 sets the reinforcement learning execution flag to ON, sets the fuel replenishment flag to OFF, sets the warm-up flag to ON, and proceeds with the process to step S035. In the example of FIG. 14, at time T2, since the reinforcement learning button has been operated to ON, the reinforcement learning execution flag and the warm-up flag are set to ON. Note that the example of FIG. 14 shows an example where it is determined that the fuel has not been replenished. The warm-up flag is a flag that is set to ON during the period in which the forced acceleration process is being executed before executing the reinforcement learning mode described later, and is set to ON in step S030 in FIG. 5 and set to OFF in step S090. In step S030, the control device 60 outputs information to the effect that "reinforcement learning will start" from the information output devices 29A and 29B.

[0057] When the process proceeds to step S035, the control device 60 determines whether the warm-up flag is OFF. If the warm-up flag is OFF (Yes), the control device 60 determines that the warm-up has been completed and proceeds with the process to step S040. If the warm-up flag is ON (No), the control device 60 determines that the vehicle is still warming up and proceeds with the process to step S090. When the acceleration of the warm-up is completed (when the warm-up flag is changed from ON to OFF), the control device 60 ends the acceleration of the warm-up and shifts to the reinforcement learning value calculation and storage process (step S070).

[0058] When the process proceeds to step S090, the control device 60 executes [warm-up acceleration process] and ends the process shown in FIG. 5. The details of the [warm-up acceleration process] will be described later.

[0059] When the process proceeds to step S040, it is determined whether the reinforcement learning mode flag is ON. If the control device 60 determines that the reinforcement learning mode flag is ON (Yes), the process proceeds to step S070; otherwise (No), the process proceeds to step S045. The reinforcement learning mode flag is a flag that is set to ON during the period when the "forced learning mode" that forcibly executes the air-fuel ratio feedback control and the calculation of the learning value is being performed. It is a flag that is set to ON in step S065 in FIG. 5 and set to OFF in step S070.

[0060] When the process proceeds to step S045, the control device 60 determines whether the target area instruction reception flag is ON. If the control device 60 determines that the target area instruction reception flag is ON (Yes), the process proceeds to step S053; otherwise (No), the process proceeds to step S050. The target area instruction reception flag is a flag that is set to ON at the start of the acquisition of the accelerator pedal depression amount when, immediately before starting the "forced learning mode", the accelerator pedal depression amount of the user is acquired and the learning area for calculating the learning value in the "forced learning mode" is selected from among a plurality of learning areas according to the accelerator pedal depression amount, and is set to OFF at the end of the acquisition of the accelerator pedal depression amount.

[0061] When the process proceeds to step S050, the control device 60 initializes and starts the area instruction timer, sets the target area instruction reception flag to ON, and proceeds to step S053. As shown in the example of FIG. 14, the control device 60 sets the target area instruction reception flag to ON and starts the area instruction timer at the timing when the warm-up flag is changed from ON to OFF (the timing when the warm-up is completed) at time T3.

[0062] When the process proceeds to step S053, the control device 60 outputs information to the effect that "Please indicate the upper limit of the learning area for which learning is desired with the depression amount of the accelerator pedal. Learning all learning areas will be performed when the accelerator pedal is fully depressed" from the information output devices 29A and 29B.

[0063] [Example of learning area GR(J, K) (Fig. 10)] Note that the example of the learning area GR(J, K) is as shown in Fig. 10, and a plurality of learning areas GR(J, K) are set according to the operating state. The learning area GR(J, K) shown in Fig. 10 shows an example in which the learning area GR(J, K) is set according to the rotational speed and the intake air amount as the operating state of the LPG engine. For example, the learning area GR(2, 3) indicates that it is a learning area in the case of an operating state where the rotational speed is equal to or higher than N1 and lower than N2, and the intake air amount is equal to or higher than G2 and lower than G3. Further, a learning value is set for each of the learning areas GR(J, K), and each learning value is stored in the non-volatile memory device.

[0064] When the process proceeds to step S055, the control device 60 determines whether or not the area instruction timer has reached the instruction waiting time or more. When the area instruction timer is equal to or longer than the instruction waiting time (Yes), the control device 60 proceeds to step S060, and when not (No), the process shown in Fig. 5 is terminated. Note that the instruction waiting time is, for example, about 1 to 3 [sec].

[0065] When the process proceeds to step S060, the control device 60 acquires the accelerator pedal depression amount, sets the target upper limit (X, Y) according to the acquired accelerator pedal depression amount, and proceeds to step S065. For example, as shown in FIG. 10, when all the learning regions GR (J, K) are the learning regions GR (1, 1) to GR (4, 4), if the accelerator pedal depression amount is 0% or more and less than 25% (100% for full depression), the target upper limit (X, Y) is set to (1, 1). If the accelerator pedal depression amount is 25% or more and less than 50%, the target upper limit (X, Y) is set to (2, 2). If the accelerator pedal depression amount is 50% or more and less than 75%, the target upper limit (X, Y) is set to (3, 3). If the accelerator pedal depression amount is 75% or more, the target upper limit (X, Y) is set to (4, 4). When the control device 60 sets the target upper limit (X, Y) to (3, 3), for example, it outputs information to the effect that "the accelerator pedal depression amount has been acquired. Among the learning regions GR (1, 1) to GR (4, 4), the learning regions GR (1, 1) to GR (3, 3) will be learned" from the information output devices 29A and 29B. (When the target upper limit is set to (3, 3), the learning regions GR (1, 1), (1, 2), (1, 3), (2, 1), (2, 2), (2, 3), (3, 1), (3, 2), and (3, 3) are learned).

[0066] In step S065, the control device 60 sets the forced learning mode flag to ON, sets the target area instruction reception flag to OFF, initializes the target area (J, K) to (1, 1), and proceeds to step S070. Note that the target area (J, K) indicates that the learning region GR (J, K) to be learned first in the "forced learning mode" executed in step S070 is the learning region GR (1, 1).

[0067] In the example of FIG. 14, at time T4A, since the "area instruction timer" has reached or exceeded the "instruction waiting time", the control device 60 acquires the accelerator pedal depression amount, sets the "target upper limit" to (3, 3), sets the "target area" to (1, 1), changes the target area instruction reception flag from ON to OFF, and changes the reinforcement learning mode flag from OFF to ON. Also, during the period from time T3 to time T4A, the user depresses the accelerator pedal to input an instruction for the upper limit of the learning area.

[0068] When the process proceeds to step S070, the control device 60 executes [reinforcement learning value calculation and storage process] (process in the "reinforcement learning mode") and proceeds to step S075. The details of [reinforcement learning value calculation and storage process] will be described later.

[0069] At step S075, the control device 60 determines whether the reinforcement learning mode flag is OFF. If the reinforcement learning mode flag is OFF (Yes), the control device 60 proceeds to step S080; otherwise (No), the process shown in FIG. 5 ends.

[0070] When the process proceeds to step S080, the control device 60 sets the reinforcement learning execution flag to OFF and ends the process shown in FIG. 5. At step S080, the control device 60 outputs information to the effect that "reinforcement learning will end" from the information output devices 29A and 29B. In the example of FIG. 14, at time T5, since the reinforcement learning mode flag is set to OFF, the control device 60 sets the reinforcement learning execution flag to OFF and ends "reinforcement learning".

[0071] ● [Fuel replenishment determination process (FIG. 6)] Next, with reference to FIG. 6, the details of the [fuel replenishment determination process] in step S010 of FIG. 5 will be described. When the control device 60 proceeds to step S010 shown in FIG. 5, it proceeds to step S110 shown in FIG. 6.

[0072] In step S110, the control device 60 acquires the "previous fuel quantity" stored in the non-volatile memory device, acquires the "current fuel quantity" using the fuel quantity detection device 27, and proceeds with the processing to step S115.

[0073] In step S115, the control device 60 determines whether "current fuel quantity" - "previous fuel quantity" is greater than or equal to the replenishment determination fuel quantity. If the control device 60 determines that "current fuel quantity" - "previous fuel quantity" is greater than or equal to the replenishment determination fuel quantity (Yes), it determines that fuel replenishment has occurred and proceeds with the processing to step S120. Otherwise (No), it proceeds with the processing to step S125.

[0074] When proceeding with the processing to step S120, the control device 60 sets the fuel replenishment flag set in the non-volatile memory device to ON and proceeds with the processing to step S125.

[0075] When proceeding with the processing to step S125, the control device 60 substitutes the "current fuel quantity" into the "previous fuel quantity" set in the non-volatile memory device, ends the processing shown in FIG. 6, and returns the processing to step S015 in FIG. 5.

[0076] ● [Warm-up acceleration process (FIG. 7)] Next, with reference to FIG. 7, the details of the [warm-up acceleration process] in step S090 of FIG. 5 will be described. When the control device 60 proceeds with the processing to step S090 shown in FIG. 5, it proceeds with the processing to step S210 shown in FIG. 7.

[0077] In step S210, the control device 60 determines whether the warm-up of the LPG engine 10 is completed. For example, the control device 60 determines that the warm-up is completed when the temperature of the coolant detected using the coolant temperature detection device 26 is equal to or higher than the warm-up determination temperature (e.g., 70 [°C] or higher). If the control device 60 determines that the warm-up is completed (Yes), it proceeds with the processing to step S220. If it determines that the warm-up is not completed (No), it proceeds with the processing to step S230.

[0078] When the process proceeds to step S220, the control device 60 controls the electric motor of the throttle device 43 to forcibly control the opening degree of the throttle valve 22 so that the rotational speed of the LPG engine 10 approaches the idle rotational speed (for example, about 700 [rpm]), sets the warm-up flag to OFF, ends the process shown in FIG. 7, returns the process to below step S090 shown in FIG. 5, and ends the process of FIG. 5. In step S220, the control device 60 outputs information to the effect that "warm-up is completed" from the information output devices 29A and 29B.

[0079] When the process proceeds to step S230, the control device 60 controls the electric motor of the throttle device 43 to forcibly control the opening degree of the throttle valve 22 so that the rotational speed of the LPG engine 10 approaches the warm-up promotion rotational speed (for example, about 3000 [rpm]), ends the process shown in FIG. 7, returns the process to below step S090 shown in FIG. 5, and ends the process of FIG. 5. In step S230, the control device 60 outputs information to the effect that "the rotational speed will be increased to promote warm-up" from the information output devices 29A and 29B.

[0080] ● [Control of Throttle Device (Throttle Valve) (FIG. 8)] Next, the control of the throttle device 43 (throttle valve 22) will be described with reference to FIG. 8. The control of the throttle device (throttle valve) is obtained by adding step SA010 to the processing of the existing control of the throttle device (throttle valve). The control device 60 controls the electric motor of the throttle device 43 according to the depression amount of the user's accelerator pedal to control the opening degree of the throttle valve. However, as shown in FIG. 14, during the period when the "reinforcement learning execution flag" is set to ON, regardless of the depression amount of the accelerator pedal from the user, during the period when the "warming-up flag" is ON, the opening degree of the throttle valve 22 of the throttle device 43 is forcibly controlled to be the "warming-up acceleration rotation speed", and during the period when the "reinforcement learning mode flag" is ON, the opening degree of the throttle valve 22 of the throttle device 43 is forcibly controlled to be the operating state corresponding to each learning region GR (J, K). The control device 60 starts the processing shown in FIG. 8 at, for example, a predetermined time interval (a time interval of about several [ms]) and proceeds to step SA010.

[0081] In step SA010, the control device 60 determines whether the reinforcement learning execution flag is ON. If the reinforcement learning execution flag is ON (Yes), the control device 60 ends the processing shown in FIG. 8. Otherwise (No), the control device 60 proceeds to step SA020.

[0082] When the processing proceeds to step SA020, the control device 60 performs the existing control of the throttle device 43 (throttle valve 22), controls the opening degree of the throttle valve 22 of the throttle device 43 according to the depression amount of the user's accelerator pedal, and ends the processing shown in FIG. 8.

[0083] ● [Reinforcement learning value calculation and storage process (FIG. 9)] Next, with reference to FIG. 9, the details of the [Forced Learning Value Calculation and Storage Process] in step S070 of FIG. 5 will be described. When the control device 60 proceeds to the process of step S070 shown in FIG. 5, it proceeds to the process of step S310 shown in FIG. 9. During the execution of the [Forced Learning Value Calculation and Storage Process], in FIG. 14, the "Forced Learning Mode Flag" is set to ON. Note that (J, K) in the learning area GR(J, K) in FIG. 9 is the value of the target area at that time, and initially it is (1, 1) set in step S065 of FIG. 5.

[0084] In step S310, the control device 60 forcibly controls the opening degree of the throttle valve 22 and the load of the load device 45 so as to be the learning area GR(J, K) corresponding to the target area (J, K), and proceeds to the process of step S320. When the control device 60 performs the process of step S070 after the process of step S065 shown in FIG. 5, since the target area (J, K) is set to (1, 1), the rotation speed is less than N1 and the intake air amount is less than G1 so as to be the learning area GR(1, 1) shown in FIG. 10. The opening degree of the throttle valve 22 and the load of the load device 45 are forcibly controlled. Note that the control device 60 outputs information to the effect that "the rotation speed and the load are automatically adjusted so as to be the learning area GR(J, K) (for example, the learning area GR(1, 1))" from the information output devices 29A and 29B.

[0085] In step S320, the control device 60 executes the [Air-Fuel Ratio Feedback Control Process] and proceeds to the process of step S330. The details of the [Air-Fuel Ratio Feedback Control Process] will be described later.

[0086] In step S330, the control device 60 executes the [Update Process of the Learning Value of the Learning Area GR(J, K)] and proceeds to the process of step S340. The details of the [Update Process of the Learning Value of the Learning Area GR(J, K)] will be described later. When the target area is (1, 1), the learning value of the learning area GR(1, 1) is updated.

[0087] In step S340, the control device 60 determines whether the learning of the learning area GR(J, K) has been completed. When the target area is (1, 1), it determines whether the update of the learning value of the learning area GR(1, 1) has been completed. If the update of the learning value of the learning area GR(J, K) has been completed (Yes), the control device 60 proceeds to step S345, and if not (No), it proceeds to step S350.

[0088] When proceeding to step S345, the control device 60 changes the target area (J, K) to the target area (J, K) for the next learning, and proceeds to step S350. For example, when the target upper limit is (3, 3) and the target area before the change is (1, 1), it updates to (1, 2) as the next target area, and when the target area before the change is (2, 3), it changes to (3, 1) as the next target area.

[0089] When proceeding to step S350, the control device 60 determines whether the next target area (J, K) is an area exceeding the target upper limit (X, Y). For example, when the target upper limit is (3, 3) and the next target area is (3, 4) or (4, 1), the control device 60 determines that it exceeds the target upper limit (X, Y). If the control device 60 determines that the next target area (J, K) exceeds the target upper limit (X, Y) (Yes), it proceeds to step S355, and if not (No), it ends the process shown in FIG. 9 and returns the process to step S075 shown in FIG. 5.

[0090] When proceeding to step S355, the control device 60 sets the forced learning mode flag to OFF, ends the process shown in FIG. 9, and returns the process to step S075 shown in FIG. 5.

[0091] In FIG. 14, at time T5, since the control device 60 has completed the update of the learning value of the learning area GR(3, 3) corresponding to the target upper limit (3, 3), it sets the "forced learning mode flag" to OFF.

[0092] ● [Air-fuel ratio feedback control process (FIG. 11)] Next, with reference to FIG. 11, the [air-fuel ratio feedback control process] of step S320 shown in FIG. 9 will be described. The [air-fuel ratio feedback control process] is obtained by adding step SB010 to the existing [air-fuel ratio feedback control process]. When the control device 60 proceeds to the process of step S320 shown in FIG. 9, it proceeds to the process of step SB010 shown in FIG. 11.

[0093] In step SB010, the control device 60 determines whether the forced learning mode flag is ON. If the forced learning mode flag is ON (Yes), the control device 60 proceeds to step SB030; otherwise (No), it proceeds to step SB020.

[0094] When proceeding to step SB020, the control device 60 determines whether the (existing) feedback execution condition is satisfied. If the (existing) feedback execution condition is satisfied (Yes), the control device 60 proceeds to step SB030; otherwise (No), it proceeds to step SB040.

[0095] When proceeding to step SB030, the control device 60 executes the existing air-fuel ratio feedback control. For example, the control device 60 calculates the air-fuel ratio correction coefficient Kf based on the air-fuel ratio related information (the value of the air-fuel ratio or rich / lean) detected using the air-fuel ratio related information detection device 24. The air-fuel ratio correction coefficient Kf is, for example, as shown in FIG. 13, a correction coefficient multiplied when obtaining the fuel injection amount, and is set to a value in the range of, for example, 0.80 to 1.20 centered around 1.00. Also, since the method of calculating the air-fuel ratio correction coefficient Kf is the same as the existing method, the description is omitted. Then, the control device 60 ends the process shown in FIG. 11 and returns to the process of step S330 shown in FIG. 9.

[0096] When proceeding to step SB040, the control device 60 sets the air-fuel ratio correction coefficient Kf to 1.00, ends the process shown in FIG. 11, and returns to the process of step S330 shown in FIG. 9.

[0097] ●[Learning value update process for learning area GR(J, K) (Fig. 12)] Next, with reference to Fig. 12, the [learning value update process for learning area GR(J, K)] in step S330 shown in Fig. 9 will be described. The [learning value update process for learning area GR(J, K)] is obtained by adding step SD010 to the existing [learning value update process for learning area GR(J, K)]. When the control device 60 proceeds to the process in step S330 shown in Fig. 9, it proceeds to step SD010 shown in Fig. 12.

[0098] In step SD010, the control device 60 determines whether the forced learning mode flag is ON. If the forced learning mode flag is ON (Yes), the control device 60 proceeds to step SD030; otherwise (No), it proceeds to step SD020.

[0099] When proceeding to step SD020, the control device 60 determines whether the execution condition of the (existing) learning value update process is satisfied. If the execution condition of the (existing) learning value update process is satisfied (Yes), the control device 60 proceeds to step SD030; otherwise (No), it ends the process shown in Fig. 12 and returns to step S340 shown in Fig. 9.

[0100] When proceeding to step SD030, the control device 60 executes the existing learning value update process. Based on the operation, state, etc. of the air-fuel ratio correction coefficient Kf, the control device 60 updates the learning value corresponding to the learning area GR(J, K) and stores the updated learning value in the non-volatile memory device. For example, when the target area is (1, 1), the control device 60 updates the learning value corresponding to the learning area GR(1, 1) based on the operation, state, etc. of the air-fuel ratio correction coefficient Kf and stores it in the non-volatile memory device. Since the method of updating the learning value based on the operation and state of the air-fuel ratio correction coefficient Kf is the same as the existing method, the description is omitted. Then, the control device 60 ends the process shown in Fig. 12 and returns to step S340 shown in Fig. 9.

[0101] In FIG. 14, during the period when the control device 60 turns on the "reinforcement learning mode flag", the target areas (J, K) are sequentially changed from (1, 1) to (1, 2) ··· (3, 2), (3, 3), and the learning values of the corresponding learning areas (J, K) are sequentially updated. In FIG. 14, for example, during the period from time T4A to time T4B, the control device 60 automatically adjusts the "throttle valve opening" and the "load of the load device" to the learning area GR(1, 1), and during the period from time T4E to time T4F, the control device 60 automatically adjusts the "throttle valve opening" and the "load of the load device" to the learning area GR(3, 2).

[0102] ● [Calculation process of fuel injection amount (FIG. 13)] Next, with reference to FIG. 13, the [calculation process of fuel injection amount] will be described. The [calculation process of fuel injection amount] has not been changed from the existing [calculation process of fuel injection amount]. The control device 60 starts the process shown in FIG. 13 at a predetermined timing (for example, the timing when the rotation angle of the crankshaft reaches a predetermined angle or the timing at a predetermined time interval), and proceeds to step SE010.

[0103] In step SE010, the control device 60 calculates the fuel injection amount and ends the process shown in FIG. 13. In this example, the fuel injection amount is obtained by multiplying the "base injection amount shown in Figure 3 + learning value (the learning value corresponding to the learning area GR(J, K) according to the operating state at that time) + other correction amounts" by the air-fuel ratio correction coefficient Kf, and further multiplying by other correction coefficients. The other correction amount is, for example, a correction amount for increasing fuel during warm-up, and the other correction coefficient is, for example, a correction coefficient that increases during acceleration or decreases during deceleration.

[0104] In the above-mentioned "reinforcement learning", when the user operates the reinforcement learning button or it is determined that fuel has been replenished, the learning value for correcting the fuel amount is calculated, so that the learning value can be obtained early and reflected in the fuel injection amount.

[0105] As described above, before starting the reinforcement learning mode (the reinforcement learning value calculation and storage process in step S070 of FIG. 5), the control device 60 selects one or a plurality of learning regions from among the plurality of learning regions according to the depression amount of the user's accelerator pedal. Then, while automatically changing the operating state of the LPG engine so as to be each selected learning region, the control device 60 calculates and stores the learning value of each learning region.

[0106] ● [Processing procedure of control device 60 (FIGS. 15 to 21) and example of operation waveform (FIG. 22) in the second embodiment] ● [Overall processing of reinforcement learning (FIG. 15) (second embodiment)] Next, an example of the processing procedure of "reinforcement learning" by the control device 60 in the second embodiment will be described using the flowcharts and the like shown in FIGS. 15 to 22. In the above-described first embodiment, the control device 60 obtained each learning value while automatically changing the operating state by controlling the opening degree of the throttle valve 22 and the like using the electric motor of the throttle device 43. However, in the second embodiment, the throttle device 43 does not have an electric motor, and the opening degree of the throttle valve 22 is directly controlled from the accelerator pedal via a cable or the like. And in the second embodiment, the user is instructed of the depression amount of the accelerator pedal from the information output devices 29A and 29B, and while causing the user to change the operating state, the learning value of each learning region is obtained. In the second embodiment, mainly the above points are different from the first embodiment. Hereinafter, the main differences from the first embodiment will be described.

[0107] [Example of learning region GR(J) (FIG. 17)] In the second embodiment, as shown in FIG. 17, examples of the learning region GR(J) are such that a plurality of learning regions GR(J) are set according to the operating state. The learning region GR(J) shown in the example of FIG. 17 shows an example in which the learning region GR(J) is set according to the intake air amount as the operating state of the LPG engine. For example, the learning region GR(2) indicates that it is a learning region in the case of an operating state where the air volume is equal to or greater than G1 and less than G2. In addition, a learning value is set for each of the learning regions GR(J), and each learning value is stored in the non-volatile memory device.

[0108] Regarding the flowchart of FIG. 5 which is the [overall process of reinforcement learning] in the first embodiment, the flowchart of FIG. 15 which is the [overall process of reinforcement learning] in the second embodiment has different parts set in thick frames compared to the flowchart of FIG. 5. Step S060 is changed to step S060B, step S065 is changed to step S065B, the processing content of step S070 is changed, and the processing content of step S090 is changed. Since the other steps are the same as the flowchart shown in FIG. 5, the description thereof is omitted.

[0109] When the process proceeds to step SB060B, the control device 60 acquires the accelerator pedal depression amount, sets the target upper limit (X) according to the acquired accelerator pedal depression amount, and proceeds to step S065B. For example, as shown in FIG. 17, when all the learning regions GR(J) are the learning regions GR(1) to GR(4), if the accelerator pedal depression amount is 0% or more and less than 25% (100% for full depression), the target upper limit (X) is set to (1), if the accelerator pedal depression amount is 25% or more and less than 50%, the target upper limit (X) is set to (2), if the accelerator pedal depression amount is 50% or more and less than 75%, the target upper limit (X) is set to (3), and if the accelerator pedal depression amount is 75% or more, the target upper limit (X) is set to (4). When the control device 60 sets the target upper limit (X) to (3), for example, it outputs information to the effect of "The accelerator pedal depression amount has been acquired. Among the learning regions GR(1) to GR(4), the learning regions GR(1) to GR(3) will be learned." from the information output devices 29A and 29B (in this example, the learning regions GR(1), (2), and (3) will be learned).

[0110] In step S065B, the control device 60 sets the forced learning mode flag to ON, sets the target area instruction reception flag to OFF, initializes the target area (J) to (1), and proceeds to step S070. Note that the target area (J) indicates that the learning region GR(J) to be learned first in the "forced learning mode" executed in step S070 is the learning region GR(1).

[0111] ● [Fuel replenishment determination process] The second embodiment of the [Fuel replenishment determination process] is the same as the [Fuel replenishment determination process] of the first embodiment shown in FIG. 6, so the description is omitted. As shown in the "Fuel replenishment flag" in FIG. 22, the second embodiment shows an example where "forced learning" is started at the start after fuel replenishment has been performed.

[0112] ● [Warm-up promotion process (FIG. 16)] Next, with reference to FIG. 16, the details of the [warming-up promotion process] in step S090 of FIG. 15 will be described. When the control device 60 proceeds to the process in step S090 shown in FIG. 15, it proceeds to step S210 shown in FIG. 16. In the [warming-up promotion process] of the second embodiment shown in FIG. 16, compared with the [warming-up promotion process] of the first embodiment shown in FIG. 7, the parts indicated by thick frames are changed, step S220 is changed to step S220B, and step S230 is changed to step S230B.

[0113] When proceeding to step S220B, the control device 60 outputs an instruction to the user using the information output devices 29A and 29B so that the rotational speed of the LPG engine 10 approaches the idle rotational speed (for example, about 700 [rpm]), ends the process shown in FIG. 16, returns the process to below step S090 shown in FIG. 15, and ends the process in FIG. 15. In step S220B, the control device 60 outputs information to the effect that "warming up is complete. Please remove your foot from the accelerator pedal and set it to the idle rotational speed" from the information output devices 29A and 29B.

[0114] When proceeding to step S230B, the control device 60 outputs an instruction to the user using the information output devices 29A and 29B so that the rotational speed of the LPG engine 10 approaches the warm-up promotion rotational speed (for example, about 3000 [rpm]), ends the process shown in FIG. 16, returns the process to below step S090 shown in FIG. 15, and ends the process in FIG. 15. In step S230B, the control device 60 outputs information to the effect that "to promote warming up, please step on the accelerator pedal to set it to the warm-up promotion rotational speed (for example, about 3000 [rpm])" from the information output devices 29A and 29B.

[0115] ● [Control of Throttle Device (Throttle Valve)] In the second embodiment, as described above, the throttle device 43 does not have an electric motor, and the opening degree of the throttle valve 22 is directly controlled from the accelerator pedal via a cable or the like. Therefore, the [control of the throttle device (throttle valve)] of the first embodiment shown in FIG. 8 is omitted.

[0116] ●[Forced learning value calculation and memory processing (Fig. 18)] Next, with reference to Fig. 18, the details of the [forced learning value calculation and memory processing] in step S070 of Fig. 15 will be described. When the control device 60 proceeds to the process in step S070 shown in Fig. 15, it proceeds to step S310B shown in Fig. 18. Compared with the [forced learning value calculation and memory processing] of the first embodiment shown in Fig. 9, the [forced learning value calculation and memory processing] of the second embodiment shown in Fig. 18 has been changed at the locations indicated by the thick frames and is significantly different. During the execution of the [forced learning value calculation and memory processing], in Fig. 22, the "forced learning mode flag" is set to ON. Note that (J) in the learning area GR(J) in Fig. 18 is the value of the target area at that time, and initially it is (1) set in step S065B of Fig. 15.

[0117] In step S310B, the control device 60 outputs an instruction to the user from the information output devices 29A and 29B so that the learning area GR(J) corresponding to the target area (J) is obtained, and proceeds to step S315B.

[0118] In the storage device 63 of the control device 60, the rotation speed N(1) (for example, about 1000 [rpm]) corresponding to the learning area GR(1), the rotation speed N(2) (for example, about 2000 [rpm]) corresponding to the learning area GR(2), the rotation speed N(3) (for example, about 3000 [rpm]) corresponding to the learning area GR(3), and the rotation speed N(4) (for example, about 4000 [rpm]) corresponding to the learning area GR(4) are stored. When the process in step S070 is performed after the process in step S065B shown in Fig. 15, the target area (J) is set to (1). In this case, the control device 60 uses the information output devices 29A and 29B to output information to the effect that "please step on the accelerator pedal so that the learning area GR(1) is obtained and set the rotation speed to N(1) (about 1000 [rpm])".

[0119] In step S315B, the control device 60 acquires the rotational speed of the LPG engine and determines whether the acquired rotational speed is within the range of the rotational speed N(J) ± ΔN (about ±200 [rpm]). If the acquired rotational speed is within the range of the rotational speed N(J) ± ΔN (Yes), the control device 60 proceeds to step S320; otherwise (No), it proceeds to step S340B. The user adjusts the depression amount of the accelerator pedal so that the actual rotational speed is within the range of the rotational speed N(J) ± ΔN.

[0120] When the process proceeds to step S320, the control device 60 executes [air-fuel ratio feedback control process] and proceeds to step S330 Although the details of the [air-fuel ratio feedback control process] will be described later, the content of the [air-fuel ratio feedback control process] in the second embodiment is slightly different from the content of the [air-fuel ratio feedback control process] in the first embodiment shown in FIG. 11.

[0121] step S330 In step, the control device 60 executes [update process of learning value in learning area GR(J)] and proceeds to step S340B. Although the details of the [update process of learning value in learning area GR(J)] will be described later, the content of the [update process of learning value in learning area GR(J)] in the second embodiment is slightly different from the content of the [update process of learning value in learning areas GR(J, K)] in the first embodiment shown in FIG. 12. When the target area is (1), the learning value of the learning area GR(1) is updated.

[0122] In step S340B, the control device 60 determines whether the learning of the learning area GR(J) is completed. When the target area is (1), it determines whether the update of the learning value of the learning area GR(1) is completed. If the update of the learning value of the learning area GR(J) is completed (Yes), the control device 60 proceeds to step S345B; otherwise (No), it proceeds to step S350B.

[0123] When the process proceeds to step S345B, the control device 60 changes the target area (J) to the target area (J) for the next learning, and proceeds with the process to step S350B. For example, when the target upper limit is (3), if the target area before the change is (1), it is updated to (2) as the target area for the next time, and if the target area before the change is (3), it is changed to (4) as the target area for the next time.

[0124] When the process proceeds to step S350B, the control device 60 determines whether the target area (J) for the next time is an area exceeding the target upper limit (X). For example, when the target upper limit is (3), if the target area for the next time is (4), the control device 60 determines that it exceeds the target upper limit (X). When the control device 60 determines that the target area (J) for the next time exceeds the target upper limit (X) (Yes), the process proceeds to step S355, and if not (No), the process shown in FIG. 18 ends, and the process returns to step S075 shown in FIG. 15.

[0125] When the process proceeds to step S355, the control device 60 sets the forced learning mode flag to OFF, ends the process shown in FIG. 18, and returns the process to step S075 shown in FIG. 15.

[0126] In FIG. 22, since the control device 60 has completed the update of the learning value of the learning area GR(3) corresponding to the target upper limit (3) at time T6, the "forced learning mode flag" is set to OFF.

[0127] ●[Air-fuel ratio feedback control process (FIG. 19)] Next, with reference to FIG. 19, the [air-fuel ratio feedback control process] in step S320 shown in FIG. 18 will be described. The [air-fuel ratio feedback control process] is obtained by adding step SB010 to the existing [air-fuel ratio feedback control process]. When the control device 60 proceeds with the process to step S320 shown in FIG. 18, it proceeds with the process to step SB010 shown in FIG. 19.

[0128] In step SB010, the control device 60 determines whether the reinforcement learning mode flag is ON. If the reinforcement learning mode flag is ON (Yes), the control device 60 proceeds to step SB030. Otherwise (No), the control device 60 proceeds to step SB020.

[0129] When proceeding to step SB020, the control device 60 determines whether the (existing) feedback execution condition is satisfied. If the (existing) feedback execution condition is satisfied (Yes), the control device 60 proceeds to step SB030B. Otherwise (No), the control device 60 proceeds to step SB040B.

[0130] When proceeding to step SB030B, the control device 60 executes the existing air-fuel ratio feedback control. For example, the control device 60 calculates the air-fuel ratio correction amount Mf based on the air-fuel ratio related information (the value of the air-fuel ratio or rich / lean) detected using the air-fuel ratio related information detection device 24. Note that the air-fuel ratio correction amount Mf is, for example, as shown in FIG. 21, the correction amount added when obtaining the fuel injection amount, centered around 0.00, for example, in the range of -A1 [ms] to +A2 [ms] (or -A3 [mm 3 to +A4 [mm 3 etc.). Also, since the method for calculating the air-fuel ratio correction amount Mf is the same as the existing method, the description is omitted. Then the control device 60 ends the process shown in FIG. 19 and returns to the step shown in FIG. 18 S330 for processing.

[0131] When proceeding to step SB040B, the control device 60 sets 0.00 for the air-fuel ratio correction amount Mf, ends the process shown in FIG. 19, and returns to the step shown in FIG. 18 S330 for processing.

[0132] ● [Update process of the learning value in the learning area GR(J) (FIG. 20)] Next, using FIG. 20, the step shown in FIG. 18 S330The [learning value update process of learning area GR(J)] will be described. The [learning value update process of learning area GR(J)] is obtained by adding step SD010 to the existing [learning value update process of learning area GR(J)]. When the control device 60 proceeds to the step S330 shown in FIG. 18, it proceeds to step SD010 shown in FIG. 20.

[0133] In step SD010, the control device 60 determines whether the forced learning mode flag is ON. When the forced learning mode flag is ON (Yes), the control device 60 proceeds to step SD030, and when it is not (No), it proceeds to step SD020.

[0134] When proceeding to step SD020, the control device 60 determines whether the execution condition of the (existing) learning value update process is satisfied. When the execution condition of the (existing) learning value update process is satisfied (Yes), the control device 60 proceeds to step SD030B, and when it is not (No), it ends the process shown in FIG. 20 and returns to step S340B shown in FIG. 18.

[0135] When proceeding to step SD030B, the control device 60 executes the existing learning value update process. Based on the operation, state, etc. of the air-fuel ratio correction amount Mf, the control device 60 updates the learning value corresponding to the learning area GR(J) and stores the updated learning value in the non-volatile storage device. For example, when the target area is (1), the control device 60 updates the learning value corresponding to the learning area GR(1) based on the operation, state, etc. of the air-fuel ratio correction amount Mf and stores it in the non-volatile storage device. Since the method of updating the learning value based on the operation and state of the air-fuel ratio correction amount Mf is the same as the existing method, the description is omitted. Then, the control device 60 ends the process shown in FIG. 20 and returns to step S340B shown in FIG. 18.

[0136] In FIG. 22, during the period when the control device 60 turns on the "reinforcement learning mode flag", the target area (J) is sequentially changed from (1) to (2) and then to (3), and the learning values of the corresponding learning areas (J) are sequentially updated. In FIG. 22, for example, during the period from time T4A to time T4B, the user adjusts the depression amount of the accelerator pedal so that the rotational speed N(1) is obtained.

[0137] ●[Calculation process of fuel injection amount (FIG. 21)] Next, with reference to FIG. 21, the [calculation process of fuel injection amount] will be described. Although the [calculation process of fuel injection amount] has not been changed from the existing [calculation process of fuel injection amount], the content of the [calculation process of fuel injection amount] in the second embodiment shown in FIG. 21 is slightly different from the content of the [calculation process of fuel injection amount] in the first embodiment shown in FIG. 13. The control device 60 starts the process shown in FIG. 21 at a predetermined timing (for example, the timing when the rotational angle of the crankshaft reaches a predetermined angle or the timing at a predetermined time interval), and proceeds to step SE010B.

[0138] In step SE010B, the control device 60 calculates the fuel injection amount and ends the process shown in FIG. 21. In this example, the fuel injection amount is obtained by multiplying "the base injection amount shown in FIG. 2 + the learning value (the learning value corresponding to the learning area GR(J) according to the operating state at that time) + the air-fuel ratio correction amount Mf + other correction amounts" by other correction coefficients. The other correction amount is, for example, a correction amount for increasing fuel during warm-up, and the other correction coefficient is, for example, a correction coefficient that increases during acceleration or decreases during deceleration.

[0139] As described above, before starting the reinforcement learning mode (the reinforcement learning value calculation and storage process in step S070 of FIG. 15), the control device 60 selects one or more learning areas corresponding to the depression amount of the user's accelerator pedal from among a plurality of learning areas. Then, the control device 60 outputs an instruction to the user using the information output device so that each selected learning area is obtained, calculates and stores the learning values of each learning area while changing the operating state of the LPG engine for the user.

[0140] In the "reinforcement learning" of the above-described first and second embodiments, when the user operates the reinforcement learning button or when it is determined that fuel has been replenished, a learning value for correcting the fuel quantity is calculated. Therefore, it is possible to obtain the learning value at an early stage and reflect it in the fuel injection quantity.

[0141] Note that the control device 60 (CPU 61) that executes the [fuel replenishment determination process] shown in FIG. 6 corresponds to a fuel replenishment detection unit 61A (see FIG. 1) that detects that fuel has been replenished.

[0142] Further, the control device 60 (CPU 61) that executes the processes of steps S015 to S030 shown in FIGS. 5 and 15 starts a reinforcement learning mode in which learning values in respective learning areas are forcibly calculated when an instruction to forcibly calculate a learning value for correcting the fuel injection quantity is input from the user, or when the fuel replenishment detection unit 61A (see FIG. 1) that detects that fuel has been replenished determines that fuel has been replenished. It corresponds to a reinforcement learning start determination unit 61B (see FIG. 1).

[0143] Further, the control device 60 (CPU 61) that executes the [warming promotion process] shown in FIGS. 6 and 16 corresponds to a forced warming promotion unit 61C (see FIG. 1) that automatically promotes warming or outputs an instruction to promote warming to the user when it is determined that warming of the internal combustion engine has not been completed before starting the reinforcement learning mode.

[0144] Further, the control device 60 (CPU 61) that executes the [reinforcement learning value calculation and storage process], [air-fuel ratio feedback control process], and [update process of learning value in learning area] shown in FIGS. 9, 11, 12, 18, 19, and 20 calculates the learning values in respective learning areas based on the state of air-fuel ratio feedback control based on air-fuel ratio related information in the operating state corresponding to the respective learning areas when starting the reinforcement learning mode, and stores the calculated learning values in respective learning areas in a non-volatile storage device. It corresponds to a reinforcement learning value calculation and storage unit 61D (see FIG. 1).

[0145] In addition, the control device 60 (CPU 61) that executes the processes of steps S350 and S355 in FIG. 9, steps S350B and S355 in FIG. 18, and steps S075 and S080 in FIGS. 5 and 15 corresponds to a forced learning cancellation unit 61E (see FIG. 1) that cancels the forced learning mode when the calculation and storage of the learning values in each learning area are completed in the forced learning value calculation and storage unit 61D (see FIG. 1).

[0146] As described above, by the "forced learning" described in the first and second embodiments, even when refueling with fuel having a large variation in composition, etc., stable combustion (control near the stoichiometric air-fuel ratio) can be realized without waiting for the start of air-fuel ratio feedback control (after warm-up). Also, since the fuel injection amount becomes appropriate, it can contribute to improving fuel efficiency. Further, since there is a margin in the width of the air-fuel ratio feedback control, the range of controllable calorific value can be widened. Also, after the "forced learning" ends, even if the air-fuel ratio related information detection device fails and the air-fuel ratio feedback control cannot be performed, stalling becomes less likely and appropriate evacuation driving can be performed.

[0147] The control device of the internal combustion engine system of the present invention is not limited to the configurations, structures, processing procedures, etc. described in this embodiment, and various changes, additions, and deletions are possible without changing the gist of the present invention. For example, the processing procedure is not limited to the flowchart described in this embodiment, and the operation waveform is also not limited to the waveforms shown in FIGS. 14 and 22.

[0148] In the description of this embodiment, an O2 sensor that detects a rich state / lean state is used as an example of an air-fuel ratio related information detection device that outputs air-fuel ratio related information, but an A / F sensor that detects the value of the air-fuel ratio may be used instead of the O2 sensor.

[0149] In addition, the control device of the LPG engine system of the present invention is not limited to an LPG engine, and can also be applied to an internal combustion engine system using a natural gas engine, a gasoline engine, etc., which controls the air-fuel ratio by stoichiometric control (controlling the air-fuel ratio near the theoretical air-fuel ratio) using air-fuel ratio feedback control based on air-fuel ratio related information. In the description of this embodiment, a forklift is taken as an example for explanation, but it is not limited to a forklift, and it can be applied to various devices using an internal combustion engine system that adjusts the air-fuel ratio by stoichiometric control.

[0150] In the description of this embodiment, when at least one of the cases where the user operates the reinforcement learning instruction device 28 (reinforcement learning button) or when it is detected that the fuel has been replenished is detected, the reinforcement learning flag is set to ON and "reinforcement learning" is executed. However, the replenishment of fuel is omitted, and "reinforcement learning" may be executed only by the operation of the reinforcement learning instruction device 28 (reinforcement learning button) by the user.

[0151] In addition, learning values corresponding to each learning region may be prepared separately for after warm-up and during warm-up, and the [warm-up promotion process] may be omitted. During warm-up, the learning value during warm-up is calculated, stored, and reflected in the fuel injection amount, and after warm-up, the learning value after warm-up is calculated, stored, and reflected in the fuel injection amount.

[0152] In addition, the processes of steps S045 to S060 in FIG. 5 or steps S045 to S060B in FIG. 15 may be omitted, and the target upper limits (X, Y) (or (X)) are always set to the upper limits of the learning regions, and the learning values of all the learning regions (J, K) (or (J)) are always calculated and stored. Also, the number of learning regions is not limited to the number described in this embodiment, and any appropriate plurality of numbers based on the operating state may be used.

[0153] In the first embodiment, as shown in FIG. 10, the learning regions (J, K) were set based on the intake air amount and the rotational speed. However, they may be set based on various operating states such as being set based on the intake air pressure and the rotational speed, or based on the load and the rotational speed. Similarly, in the second embodiment, as shown in FIG. 17, the learning region (J) was set based on the intake air amount. However, it may be set based on various operating states such as being set based on the intake air pressure or the load. Also, in either the first or second embodiment, the intake air flow rate detection device 15 may be omitted, and the intake air amount may be converted from the intake air pressure and the rotational speed.

[0154] Also, a manual of the operation procedure of "forced learning" may be prepared to let the user understand the operation at the time of executing "forced learning", and the information output devices 29A and 29B may be omitted, but it is preferable to have the information output devices 29A and 29B.

[0155] Also, after updating and storing the learning values of the respective learning regions in the "forced learning" described in this embodiment, the learning values may be further updated and stored by the conventional learning control (when a steady state that is not a transient state continues, updating and storing the learning values of the learning region corresponding to the operating state at that time). The "forced learning" described in this embodiment may be added to the conventional learning control, or the "forced learning" described in this embodiment may be implemented instead of the conventional learning control.

[0156] Also, "greater than or equal to" (≧), "less than or equal to" (≦), "greater than" (>), "less than" (<), etc. may or may not include the equal sign. Also, the numerical values used in the description of this embodiment are examples and are not limited to these numerical values.

Explanation of Reference Numerals

[0157] 1 LPG engine system (internal combustion engine system) 10 LPG engine (internal combustion engine) 11A, 11B Intake pipes 11C Intake manifold 12A Exhaust manifold 12B Exhaust pipe 15 Intake air flow detection device 16 Accelerator pedal depression amount detection device 21 Air cleaner 22 Throttle valve 23 Pressure detection device 24 Air-fuel ratio related information detection device 25 Rotation detection device 26 Coolant temperature detection device 27 Fuel quantity detection device 28 Forced learning instruction device (forced learning button) 29A Information output device (display) 29B Information output device (speaker) 30 Fuel mixer 31 Stepper motor 32 Needle 33 Fuel pipe 40 Injector 41 Igniter 42 Spark plug 43 Throttle device 45 Load device (alternator, hydraulic pump) 50 LPG regulator 51 Main fuel pipe 52 Auxiliary fuel pipe 53 Fuel tank 60 Control device 61 CPU 61A Fuel replenishment detection unit 61B Forced learning start determination unit 61C Forced warm-up promotion unit 61D Forced learning value calculation and storage unit 61E Forced learning release unit 62 RAM 63 Storage device (ROM) 64 Timer 65 Non-volatile storage device 70 Three-way catalyst

Claims

1. A control device for an internal combustion engine system, wherein the internal combustion engine system includes, an internal combustion engine, an air-fuel ratio related information detection device that detects air-fuel ratio related information related to the air-fuel ratio of the exhaust gas of the internal combustion engine, and the control device adjusts the fuel injection amount to the internal combustion engine by stoichiometric control using air-fuel ratio feedback control based on the operating state including the air-fuel ratio related information in the internal combustion engine, and a throttle device that can adjust a throttle valve of the internal combustion engine from the control device, and has, the control device, has a plurality of learning regions corresponding to the intake air amount of the internal combustion engine, stores learning values of each of the learning regions in a non-volatile storage device, corrects the fuel injection amount to the internal combustion engine according to the operating state using the learning value corresponding to the learning region corresponding to the intake air amount of the internal combustion engine, when a forced learning execution instruction for forcibly calculating the learning value for correcting the fuel injection amount is input from a user, or when a fuel supply detection unit that detects fuel supply determines that fuel has been supplied, a forced learning start determination unit that starts a forced learning mode for forcibly calculating the learning values of each of the learning regions, when the forced learning mode is started, calculates the learning values of each of the learning regions based on the state of the air-fuel ratio feedback control based on the air-fuel ratio related information at the intake air amount of the internal combustion engine corresponding to the corresponding learning region, and stores the calculated learning values of each of the learning regions in a non-volatile storage device, a forced learning value calculation and storage unit, a forced learning release unit that releases the forced learning mode when the calculation and storage of the learning values of each of the learning regions are completed by the forced learning value calculation and storage unit, and has, in the forced learning mode, the control device, In the forced learning value calculation and storage unit, while automatically changing the intake air amount of the internal combustion engine by adjusting the throttle device so as to be each of the learning regions, the learning value of each of the learning regions is calculated and stored. A control device for an internal combustion engine system.

2. A control device for an internal combustion engine system according to Claim 1, wherein The internal combustion engine system has a load device whose load of the internal combustion engine can be adjusted from the control device, The control device has a plurality of the learning regions according to the rotational speed and intake air amount of the internal combustion engine, and stores the learning values of each of the learning regions in a non-volatile storage device, corrects the fuel injection amount to the internal combustion engine according to the operating state using the learning value corresponding to the learning region according to the rotational speed and intake air amount of the internal combustion engine, when starting the forced learning mode, calculates the learning values of each of the learning regions based on the state of the air-fuel ratio feedback control based on the air-fuel ratio related information at the rotational speed and intake air amount of the internal combustion engine that become the corresponding learning regions, and has the forced learning value calculation and storage unit that stores the calculated learning values of each of the learning regions in a non-volatile storage device, In the forced learning mode, the control device in the forced learning value calculation and storage unit, while automatically changing the rotational speed and intake air amount of the internal combustion engine by adjusting the throttle device and the load device so as to be each of the learning regions, calculates and stores the learning values of each of the learning regions. A control device for an internal combustion engine system.

3. A control device for an internal combustion engine system according to Claim 1 or 2, wherein the internal combustion engine system has an accelerator pedal operated by a user, the control device Before starting the forced learning mode, the forced learning value calculation and storage unit selects a plurality of the learning regions corresponding to the accelerator pedal depression amount, which is the depression amount of the user's accelerator pedal, from among the plurality of the learning regions. A control device for an internal combustion engine system.

4. A control device for an internal combustion engine system according to any one of Claims 1 to 3, wherein the control device has a forced warm-up promotion unit that automatically promotes warm-up when it is determined that the warm-up of the internal combustion engine has not been completed before starting the forced learning mode, and when the forced warm-up promotion unit promotes warm-up, when the warm-up is completed, the promotion of warm-up is terminated and the process proceeds to the processing of the forced learning value calculation and storage unit. A control device for an internal combustion engine system.

5. A control device for an internal combustion engine system, wherein the internal combustion engine system includes an internal combustion engine, an air-fuel ratio related information detection device that detects air-fuel ratio related information related to the air-fuel ratio of the exhaust gas of the internal combustion engine, the control device that adjusts the fuel injection amount to the internal combustion engine by stoichiometric control using air-fuel ratio feedback control based on the operating state including the air-fuel ratio related information in the internal combustion engine, an accelerator pedal operated by a user, and a throttle opening adjustment device that adjusts the opening degree of the throttle valve of the internal combustion engine according to the operation amount of the accelerator pedal by the user, and has wherein the control device has a plurality of learning regions corresponding to the intake air amount of the internal combustion engine, stores the learning values of each of the learning regions in a non-volatile storage device, and corrects the fuel injection amount to the internal combustion engine according to the operating state using the learning value corresponding to the learning region corresponding to the intake air amount of the internal combustion engine. When a forced learning execution instruction for forcibly calculating the learning value for correcting the fuel injection amount is input from the user, or when the fuel replenishment detection unit that detects fuel replenishment determines that fuel has been replenished, a forced learning start determination unit that starts a forced learning mode for forcibly calculating the learning value for each of the learning regions, When the forced learning mode is started, the learning value for each of the learning regions is calculated based on the state of the air-fuel ratio feedback control based on the air-fuel ratio related information at the intake air amount of the internal combustion engine corresponding to the corresponding learning region, and the calculated learning value for each of the learning regions is stored in a non-volatile storage device. A forced learning value calculation / storage unit; A forced learning release unit that releases the forced learning mode when the calculation and storage of the learning value for each of the learning regions are completed by the forced learning value calculation / storage unit; having The internal combustion engine system has an information output device capable of outputting an instruction to the user, In the forced learning mode, the control device In the forced learning value calculation / storage unit, an instruction for the operation amount of the accelerator pedal corresponding to the intake air amount of each learning region is output to the user using the information output device so as to be each of the learning regions, and while changing the intake air amount of the internal combustion engine according to the operation amount of the accelerator pedal to the user, calculate and store the learning value for each of the learning regions. A control device for an internal combustion engine system.

6. A control device for an internal combustion engine system according to claim 5, wherein The control device In the forced learning value calculation / storage unit, before starting the forced learning mode, a plurality of learning regions corresponding to the depression amount of the accelerator pedal, which is the depression amount of the user's accelerator pedal, are selected from among the plurality of learning regions. A control device for an internal combustion engine system.

7. A control device for an internal combustion engine system according to claim 5 or 6, wherein The control device is When it is determined that the warm-up of the internal combustion engine has not been completed before starting the reinforcement learning mode, it has a forced warm-up promotion unit that outputs an instruction to promote warm-up to the user using the information output device. When the forced warm-up promotion unit promotes warm-up, when the warm-up is completed, the promotion of warm-up is terminated and the process proceeds to the processing of the forced learning value calculation / storage unit. A control device for an internal combustion engine system.

Images