Vehicle control devices

Abstract

A vehicle control device is provided, which is a control device capable of rapidly progressing learning related to clutch engagement during engine startup in a vehicle equipped with an engine, an electric motor, and a clutch capable of disconnecting the connection between the engine and the electric motor. A learning control unit (98) pre-sets priorities for a plurality of types of learning. When a higher-priority learning unit among the plurality of types of learning units is judged to be in a non-convergent state, the degree of reflection of the learning results of a lower-priority learning unit is reduced compared to a case where the higher-priority learning unit is judged to be in a convergent state. Therefore, even if the higher-priority learning unit is in a non-convergent state, the learning of the lower-priority learning unit is progressed while the degree of reflection of the learning results of the lower-priority learning unit is reduced. This makes it possible to rapidly progress overall learning while reducing the influence of erroneous learning caused by the non-convergent state of the higher-priority learning unit.

Description

Technical Field

[0001] The present invention relates to a control device for a vehicle including an engine, an electric motor, and a clutch capable of disconnecting the engine and the electric motor. Background Art

[0002] A vehicle control device is known, comprising: an engine; an electric motor connected to a power transmission path between the engine and drive wheels for power transmission; a clutch disposed between the engine and the electric motor in the power transmission path, the control state of which is switched by controlling a hydraulic clutch actuator; and a hydraulic control circuit for supplying regulated hydraulic pressure to the clutch actuator. For example, Patent Document 1 discloses a vehicle control device of this type. Patent Document 1 discloses that, when starting the engine, the clutch is engaged while sliding to increase engine speed, and a compensation torque is output from the electric motor to cancel the deceleration torque generated by the clutch engagement; timing learning is performed to correct the offset between the clutch transmission torque and the compensation torque generation timing; after the offset in the generation timing converges, magnitude learning is performed to correct the offset between the clutch transmission torque and the compensation torque; and after the offset in magnitude converges, fast filling time learning is performed to correct the time for fast filling of the clutch packing by temporarily increasing the hydraulic pressure command value at the start of clutch engagement.

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Laid-Open No. 2018-79876 Summary of the Invention

[0006] However, in the technology described in Patent Document 1, learning is performed in the order of timed learning, size learning, and rapid filling time learning. However, if the convergence of the upper learning is delayed, the convergence of the lower learning is also delayed, and it may take an excessively long time to correct the hydraulic deviation.

[0007] The present invention has been made against the background of the above circumstances, and an object of the present invention is to provide a vehicle control device capable of rapidly progressing learning regarding clutch engagement at engine start.

[0008] The gist of the first invention is to provide a control device for a vehicle, the vehicle comprising: (a) an engine; an electric motor connected to a power transmission path between the engine and drive wheels in a power transmission manner; a clutch provided between the engine and the electric motor in the power transmission path, the control state of which is switched by controlling a hydraulic clutch actuator; and a hydraulic control circuit for supplying a regulated hydraulic pressure to the clutch actuator, wherein the control device is characterized in that it comprises: (b) a starting control unit for controlling the electric motor in such a manner that the output torque of the electric motor is increased by a necessary starting torque amount as a torque required for starting the engine, and for controlling the engine in such a manner that the engine starts running; and (c) a clutch control unit for controlling the clutch in such a manner that the clutch is controlled at the time of starting the engine. In the engagement transition when the clutch state switches from a released state to an engaged state, a starting hydraulic command value for regulating the hydraulic pressure of the clutch actuator in such a manner that the clutch transmits the necessary starting torque is output to the hydraulic control circuit as a hydraulic command value for supplying the hydraulic pressure; and (d) a learning control unit for correcting multiple types of learning representing the correlation between the hydraulic pressure and the hydraulic command value in the engagement transition of the clutch, and (e) the learning control unit pre-setting priorities for the multiple types of learning, and when an upper learning with a higher priority among the multiple types of learning is judged to be in a non-converged state, the degree of reflection of the learning result of the lower learning with a lower priority than the upper learning is reduced compared to the case where the upper learning is judged to be in a converged state.

[0009] The gist of the second invention is that, in the first invention, the learning control unit reflects the correlation between the hydraulic pressure and the hydraulic command value in the lower learning when the upper learning is judged to be in a non-convergent state and the change amount according to the state of the clutch obtained in the learning of the lower learning exceeds a pre-set predetermined range.

[0010] The gist of the third invention is that, in the first invention or the second invention, (a) the plurality of types of learning are respectively implemented according to a plurality of progress stages distinguished based on the control state switched in the engagement transition of the clutch, and (b) the plurality of types of learning include: learning that corrects the correlation between the hydraulic pressure and the hydraulic command value based on a variation obtained from the time point of switching to the learnable progress stage to the time point after a predetermined time has passed from the time point of completion of the progress stage, which changes according to the state of the clutch.

[0011] The gist of the fourth invention is characterized in that, in the third invention, (a) the plurality of types of learning include rapid filling time learning, in which: the execution time of outputting a rapid filling hydraulic instruction value for improving the responsiveness of the hydraulic pressure to the clutch actuator in a manner that allows the clutch to quickly become a component-attached completion state, i.e., the rapid filling time, is learned; and (b) the rapid filling time learning learns the rapid filling time based on the variation obtained during the period from the time point of outputting the rapid filling hydraulic instruction value to the time point of a predetermined time after the time point when the output of the rapid filling hydraulic instruction value is completed.

[0012] The gist of the fifth invention is that, in any one of the first to fourth inventions, the learning control unit updates the correlation between the hydraulic pressure and the hydraulic command value using the learning results of each of the multiple types of learning after the start of the engine on which the learning is performed is completed and when the hydraulic pressure of the clutch actuator supplied to the clutch is below a predetermined value or when the hydraulic command value is not output to the hydraulic control circuit.

[0013] According to the first invention, even if the upper-level learning that is higher than the lower-level learning is in a non-convergent state, by reducing the degree of reflection of the learning results of the lower-level learning while making the learning of the lower-level learning progress, the overall learning can be rapidly progressed while reducing the impact of erroneous learning caused by the non-convergent state of the upper-level learning.

[0014] According to the second invention, when the amount of change obtained in the lower-level learning according to the state of the clutch exceeds a predetermined range, the lower-level learning can be advanced without waiting for the convergence of the upper-level learning, so the overall learning can be advanced rapidly while reducing the impact of erroneous learning caused by the unconverged state of the upper-level learning.

[0015] According to the third invention, multiple types of learning include: learning that corrects the correlation between the hydraulic pressure and the hydraulic command value based on the amount of change in the clutch state change obtained from the time point when the progress stage switched in the clutch engagement transition is switched to the time point when the progress stage is completed after a predetermined time, so that appropriate learning can be performed even in the case of a communication delay between the control device and the linear solenoid valve that controls the hydraulic pressure of the clutch actuator, and a response delay caused by the characteristics of the linear solenoid valve.

[0016] According to the fourth invention, the rapid filling time is learned based on the change obtained during the period from the time point when the rapid filling hydraulic command value is output to the time point when the output is completed and a predetermined time has passed. Therefore, even in the case of a communication delay between the control device and the linear solenoid valve that controls the hydraulic pressure to the clutch actuator, or a response delay caused by the characteristics of the linear solenoid valve, the rapid filling time can be appropriately learned.

[0017] According to the fifth invention, after the engine start is completed and the hydraulic pressure of the clutch actuator of the clutch is below a predetermined value or the hydraulic command value is not output to the hydraulic control circuit, the correlation between the hydraulic pressure and the hydraulic command value is updated using the learning results of multiple types of learning. Therefore, when the load caused by the calculation applied to the control device is small, the correlation between the hydraulic pressure and the hydraulic command value is updated to suppress the impact on other controls. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a diagram illustrating a schematic configuration of a vehicle to which the present invention is applied, and is a diagram illustrating control functions for various controls in the vehicle and main parts of the control system.

[0019] Figure 2 It is a partial cross-sectional view showing an example of the K0 clutch.

[0020] Figure 3 This is a diagram illustrating an example of each stage in the K0 control stage definition.

[0021] Figure 4 This is a diagram showing an example of a timing chart when engine startup control is executed.

[0022] Figure 5 This is an example of a relationship map used to calculate the gain of a learning value in contact-point learning, etc.

[0023] Figure 6 This is an example of a relationship map used to calculate the offset of the learning value in contact point learning, etc.

[0024] Figure 7 This is a flowchart for explaining a main part of the control operation of the electronic control device, that is, a control operation for enabling rapid progress of learning when learning of a clutch command value is executed during engine startup transition.

[0025] Figure 8 This is a flowchart explaining the control operation of the electronic control device when learning the QA time.

[0026] Figure 9This is a flowchart illustrating the control operation of the electronic control device when updating each learning value calculated in each learning control.

[0027] (Explanation of Reference Numerals)

[0028] 10: Vehicle; 12: Engine; 14: Drive wheels; 20: K0 clutch (clutch); 56: Hydraulic control circuit; 90: Electronic control unit (control unit); 92c: Start control unit; 94: Clutch control unit; 98: Learning control unit; 120: Clutch actuator; MG: Motor; CTlrnk0: K0 learning control (learning); CTlrnqa: QA time learning (rapid filling time learning, upper learning, learning); CTlrnpk: Contact point learning (learning); CTlrntk: Transfer torque Torque learning (learning); CTlrntm: wasted time learning (learning); PRk0: K0 hydraulic pressure (hydraulic pressure); Rp: predetermined range; Spk0: hydraulic command value; Spk0ff: hydraulic command value for rapid filling; Spk0cr: hydraulic command value for starting (cranking); Tcrn: required starting torque; TMqa: QA time (rapid filling time); VALlrn: learning value; α: predetermined time; ΔNm: MG rotation variation (variation); ΔTm: MG torque variation (variation). DETAILED DESCRIPTION

[0029] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following embodiments, the drawings are appropriately simplified or deformed, and the dimensional ratios and shapes of each part are not necessarily accurately depicted.

[0030] [Example]

[0031] Figure 1 1 is a diagram illustrating a schematic structure of a vehicle 10 to which the present invention is applied, and is a diagram illustrating control functions for various controls in the vehicle 10 and a main part of the control system. Figure 1 In FIG, the vehicle 10 is a hybrid vehicle including an engine 12 and a motor MG as driving force sources for traveling. The vehicle 10 also includes drive wheels 14 and a power transmission device 16 provided as a power transmission path between the engine 12 and the drive wheels 14.

[0032] The engine 12 is a well-known internal combustion engine such as a gasoline engine or a diesel engine. The engine 12 is controlled by an electronic control unit 90 described later, which controls an engine control unit 50 provided in the vehicle 10 and includes a throttle actuator, a fuel injection device, an ignition device, etc., thereby controlling the engine torque Te, which is the output torque of the engine 12.

[0033] The electric motor MG is a rotating electrical machine that functions as both an engine that generates mechanical power from electrical power and a generator that generates electrical power from mechanical power. It is a so-called motor generator. The electric motor MG is connected to a battery 54 installed in the vehicle 10 via an inverter 52. The electric motor MG is controlled by the electronic control unit 90, described later, to control the inverter 52, thereby controlling the MG torque Tm, which is the output torque of the electric motor MG. For example, when the rotation direction of the electric motor MG is positive, which is the same rotation direction as the engine 12, the MG torque Tm is a power running torque when it is a positive torque on the acceleration side, and a regenerative torque when it is a negative torque on the deceleration side. Specifically, the electric motor MG generates driving power using electricity supplied from the battery 54 via the inverter 52, instead of the engine 12 or in addition to the engine 12. Furthermore, the electric motor MG generates electricity using the power of the engine 12 and the driven force input from the drive wheels 14. The electricity generated by the electric motor MG is stored in the battery 54 via the inverter 52. The battery 54 is an electric storage device that supplies and receives electric power to the electric motor MG. The electric power, unless otherwise specified, also has the same meaning as electric energy. The power, torque, and force, unless otherwise specified, also have the same meaning.

[0034] The power transmission device 16 includes a K0 clutch 20, a torque converter 22, an automatic transmission 24, and the like within a housing 18, which is a non-rotating component mounted on the vehicle body. The K0 clutch 20 is a clutch provided between the engine 12 and the electric motor MG in the power transmission path between the engine 12 and the drive wheels 14. The torque converter 22 is coupled to the engine 12 via the K0 clutch 20. The automatic transmission 24 is coupled to the torque converter 22 and is interposed in the power transmission path between the torque converter 22 and the drive wheels 14. The torque converter 22 and the automatic transmission 24 each constitute a portion of the power transmission path between the engine 12 and the drive wheels 14. Furthermore, the power transmission device 16 includes a screw shaft 28 coupled to a transmission output shaft 26, which is an output rotating component of the automatic transmission 24; a differential gear 30 coupled to the screw shaft 28; and a pair of drive shafts 32 coupled to the differential gear 30. The power transmission device 16 also includes an engine connection shaft 34 that connects the engine 12 and the K0 clutch 20 , a motor connection shaft 36 that connects the K0 clutch 20 and the torque converter 22 , and the like.

[0035] The electric motor MG is connected to the motor coupling shaft 36 within the housing 18 in a power-transmittable manner. The electric motor MG is connected to the power transmission path between the engine 12 and the drive wheels 14, particularly the power transmission path between the K0 clutch 20 and the torque converter 22. That is, the electric motor MG is connected to the torque converter 22 and the automatic transmission 24 in a power-transmittable manner, without passing through the K0 clutch 20. From a different perspective, the torque converter 22 and the automatic transmission 24 each constitute a portion of the power transmission path between the electric motor MG and the drive wheels 14. The torque converter 22 and the automatic transmission 24 each transmit the driving force from the engine 12 and the electric motor MG, respectively, to the drive wheels 14.

[0036] The torque converter 22 includes a pump impeller 22a connected to the motor connecting shaft 36 and a turbine impeller 22b connected to the transmission input shaft 38, which serves as the input rotating member of the automatic transmission 24. The pump impeller 22a is connected to the engine 12 via the K0 clutch 20 and is directly connected to the motor MG. The pump impeller 22a is the input member of the torque converter 22, and the turbine impeller 22b is the output member of the torque converter 22. The motor connecting shaft 36 also serves as the input rotating member of the torque converter 22. The transmission input shaft 38 also serves as the output rotating member of the torque converter 22, being integrally formed with the turbine shaft, which is rotationally driven by the turbine impeller 22b. The torque converter 22 is a fluid-type transmission device that transmits driving force from each driving force source (the engine 12 and the motor MG) to the transmission input shaft 38 via a fluid. The torque converter 22 includes an LU clutch 40 that connects the pump impeller 22a and the turbine impeller 22b. The LU clutch 40 is a direct-connection clutch that couples the input and output rotating members of the torque converter 22 , ie, a well-known lock-up clutch.

[0037] The LU clutch 40 switches its operating state, or control state, by varying the LU torque Tlu, representing the torque capacity of the LU clutch 40, using the regulated hydraulic pressure (LU hydraulic pressure PRlu) supplied from a hydraulic control circuit 56 provided in the vehicle 10. The control states of the LU clutch 40 include a fully released state (where the LU clutch 40 is released), a slipping state (where the LU clutch 40 is engaged with slip), and a fully engaged state (where the LU clutch 40 is engaged). When the LU clutch 40 is fully released, the torque converter 22 achieves a torque amplification effect. Furthermore, when the LU clutch 40 is fully engaged, the torque converter 22 enters a locked state, where the pump impeller 22a and the turbine impeller 22b rotate integrally.

[0038] The automatic transmission 24 is a well-known planetary gear automatic transmission equipped with, for example, one or more planetary gear sets (not shown) and a plurality of engagement devices CB. The engagement devices CB are hydraulic friction engagement devices comprised of, for example, a multi-plate or single-plate clutch or brake pressed by a hydraulic actuator, or a band brake tensioned by a hydraulic actuator. Each engagement device CB changes its CB torque Tcb, representing its torque capacity, using a regulated hydraulic pressure (CB hydraulic pressure PRcb) supplied from the hydraulic control circuit 56, thereby switching between control states such as engaged and released.

[0039] The automatic transmission 24 is a stepped transmission that, when one of the engagement devices CB is engaged, establishes one of multiple speed steps (also called gear steps) with different speed ratios (also called gear ratios) γat (=AT input speed Ni / AT output speed No). The automatic transmission 24 selectively establishes multiple gear steps by switching the gear steps based on the driver's (= driver's) accelerator operation, vehicle speed V, and other factors, using the electronic control unit 90 described later. The AT input speed Ni is the speed of the transmission input shaft 38 and is the input speed of the automatic transmission 24. The AT input speed Ni is also the speed of the output rotating member of the torque converter 22 and is the same value as the turbine speed Nt, which is the output speed of the torque converter 22. The AT input speed Ni can be expressed as the turbine speed Nt. The AT output speed No is the speed of the transmission output shaft 26 and is the output speed of the automatic transmission 24.

[0040] The K0 clutch 20 is a wet or dry friction engagement device, for example, comprised of a multi-plate or single-plate clutch pressed by a hydraulic clutch actuator 120, described later. The K0 clutch 20 is controlled by the clutch actuator 120, described later, via the electronic control unit 90, thereby switching between control states such as an engaged state and a released state. The K0 clutch 20 corresponds to the clutch of the present invention.

[0041] Figure 2 : is a partial cross-sectional view showing an example of the K0 clutch 20. Figure 2 In FIG, the K0 clutch 20 includes a clutch drum 100, a clutch hub 102, a partition plate 104, a friction plate 106, a piston 108, a return spring 110, a spring receiving plate 112, and a snap ring 114. The clutch drum 100 and the clutch hub 102 are arranged on the same axis CS. Figure 2 , the radially outer peripheral portion of the K0 clutch 20 in the upper half of the axis CS is shown. The axis CS is the axis of the engine connecting shaft 34, the motor connecting shaft 36, and the like.

[0042] The clutch drum 100 is connected to, for example, the engine connecting shaft 34 and rotates integrally with it. The clutch hub 102 is connected to, for example, the motor connecting shaft 36 and rotates integrally with it. The outer circumference of the spacer plates 104, which are multiple, generally annular plates, are non-rotatably fitted, or splined, to the inner circumference of the cylindrical portion 100a of the clutch drum 100. Friction plates 106 are interposed between the spacer plates 104 and have their inner circumferences non-rotatably fitted, or splined, to the outer circumference of the clutch hub 102. The piston 108 has a pressing portion 108a on its outer circumference that extends toward the spacer plates 104 and friction plates 106. A return spring 110 is interposed between the piston 108 and a spring receiving plate 112, urging the piston 108 to abut a portion of the bottom plate 100b of the clutch drum 100. That is, the return spring 110 functions as a spring element that urges the piston 108 to disengage the spacer plate 104 and the friction plate 106. The snap ring 114 is fixed to the cylindrical portion 100a of the clutch drum 100 at a position between the spacer plate 104 and the friction plate 106 and the pressing portion 108a of the piston 108.

[0043] In the K0 clutch 20, an oil chamber 116 is formed between the piston 108 and the bottom plate portion 100b of the clutch drum 100. An oil passage 118 is formed in the clutch drum 100, leading to the oil chamber 116. In the K0 clutch 20, the clutch drum 100, the piston 108, the return spring 110, the spring receiving plate 112, the oil chamber 116, and the like constitute a clutch actuator 120, which is a hydraulic actuator.

[0044] The hydraulic control circuit 56 supplies the regulated K0 hydraulic pressure PRk0 to the clutch actuator 120. In the K0 clutch 20, when the K0 hydraulic pressure PRk0 is supplied from the hydraulic control circuit 56 to the oil chamber 116 via the oil passage 118, the K0 hydraulic pressure PRk0 causes the piston 108 to move toward the partition plate 104 and the friction plate 106 against the biasing force of the return spring 110. The pressing portion 108a of the piston 108 presses the partition plate 104 and the friction plate 106. When the partition plate 104 and the friction plate 106 are pressed, the K0 clutch 20 switches to the engaged state. The K0 clutch 20 uses the K0 hydraulic pressure PRk0 to change the K0 torque Tk0, which represents the torque capacity of the K0 clutch 20, thereby switching the control state. Furthermore, the torque capacity of the engagement device, such as LU torque Tlu, CB torque Tcb, and K0 torque Tk0, corresponds to the maximum torque that the engagement device can transmit, or the maximum transfer torque. In a narrow sense, this is distinct from the transfer torque of the engagement device, which corresponds to the torque actually transferred by the engagement device. In this embodiment, when no particular distinction is made, the transfer torque of the engagement device also refers to the maximum torque that the engagement device can transmit. For example, K0 torque Tk0 is synonymous with the transfer torque of the K0 clutch 20.

[0045] K0 torque Tk0 is determined by factors such as the friction coefficient of the friction material of the friction plate 106 and the K0 hydraulic pressure PRk0. In the K0 clutch 20, when the hydraulic oil OIL is filled into the oil chamber 116 and the gap between the spacer plate 104 and the friction plate 106 is closed by the pressing force of the piston 108 (=PRk0 × piston pressure area) counteracting the biasing force of the return spring 110, i.e., when the pack clearance of the K0 clutch 20 is closed, the so-called pack contact is completed. In this embodiment, the state in which the pack clearance of the K0 clutch 20 is closed is referred to as the pack contact completion state. The K0 clutch 20 generates K0 torque Tk0 by further increasing the K0 hydraulic pressure PRk0 from the pack contact completion state. Specifically, the pack contact completion state of the K0 clutch 20 is when the K0 hydraulic pressure PRk0 is increased from this pack contact completion state, causing the K0 clutch 20 to begin to have torque capacity, i.e., when the K0 torque Tk0 is generated. The K0 hydraulic pressure PRk0 for contacting the components of the K0 clutch 20 is the K0 hydraulic pressure PRk0 for achieving a state in which the piston 108 reaches the stroke end and the K0 torque Tk0 is not generated.

[0046] Return to Figure 1When the K0 clutch 20 is engaged, the pump impeller 22a and the engine 12 rotate integrally via the engine connecting shaft 34. Specifically, the K0 clutch 20 connects the engine 12 and the drive wheels 14 so that power can be transmitted by engagement. On the other hand, when the K0 clutch 20 is released, power transmission between the engine 12 and the pump impeller 22a is disconnected. In other words, the K0 clutch 20 disconnects the connection between the engine 12 and the drive wheels 14 by releasing the clutch. Because the motor MG is connected to the pump impeller 22a, the K0 clutch 20 is provided in the power transmission path between the engine 12 and the motor MG and functions as a clutch that disconnects the power transmission path, that is, disconnects the engine 12 from the motor MG. Specifically, the K0 clutch 20 is a disconnecting clutch that connects the engine 12 and the motor MG by engagement and disconnects the engine 12 and the motor MG by release.

[0047] In the power transmission device 16, when the K0 clutch 20 is engaged, the power output from the engine 12 is transmitted from the engine connecting shaft 34 to the drive wheels 14 in this order, passing through the K0 clutch 20, the motor connecting shaft 36, the torque converter 22, the automatic transmission 24, the screw shaft 28, the differential gear 30, and the drive shaft 32. Furthermore, regardless of the control state of the K0 clutch 20, the power output from the electric motor MG is transmitted from the motor connecting shaft 36 to the drive wheels 14 in this order, passing through the torque converter 22, the automatic transmission 24, the screw shaft 28, the differential gear 30, and the drive shaft 32.

[0048] The vehicle 10 includes a mechanical oil pump (MOP58), an electric oil pump (EOP60), a pump motor 62, and the like. The MOP58 is connected to the pump impeller 22a and is rotationally driven by the driving force source (engine 12, electric motor MG) to discharge hydraulic oil (OIL) used in the power transmission device 16. The pump motor 62 is a motor dedicated to the EOP60 for rotationally driving the EOP60. The EOP60 is rotationally driven by the pump motor 62 to discharge hydraulic oil (OIL). The hydraulic oil (OIL) discharged by the MOP58 and EOP60 is supplied to the hydraulic control circuit 56. The hydraulic control circuit 56 supplies the CB hydraulic pressure PRcb, the K0 hydraulic pressure PRk0, the LU hydraulic pressure PRlu, and the like, whose pressures are adjusted according to the hydraulic oil (OIL) discharged by the MOP58 and / or EOP60.

[0049] The vehicle 10 also includes an electronic control unit 90, which includes control devices for the vehicle 10 related to, for example, starting control of the engine 12. The electronic control unit 90 is configured as a so-called microcomputer including, for example, a CPU, RAM, ROM, and input / output interfaces. The CPU processes signals according to programs pre-stored in the ROM while utilizing the RAM's temporary storage function to execute various controls for the vehicle 10. The electronic control unit 90 is configured to include various computers for engine control, motor control, hydraulic control, and the like, as needed.

[0050] Various signals based on detection values detected by various sensors installed in the vehicle 10 (e.g., the engine speed sensor 70, the turbine speed sensor 72, the output speed sensor 74, the MG speed sensor 76, the accelerator opening sensor 78, the throttle opening sensor 80, the brake switch 82, the battery sensor 84, the fluid temperature sensor 86, etc.) are supplied to the electronic control unit 90 (e.g., the engine speed Ne which is the speed of the engine 12, the turbine speed Nt which is the same value as the AT input speed Ni, the AT output speed No corresponding to the vehicle speed V, the MG speed Nm which is the speed of the electric motor MG, the accelerator opening θacc which is the amount of accelerator operation by the driver indicating the magnitude of the driver's acceleration operation, the throttle opening θth which is the opening of the electronic throttle valve, the brake ON signal Bon which is a signal indicating the state in which the brake pedal for actuating the wheel brakes is operated by the driver, the battery temperature THbat of the battery 54, the battery charge and discharge current Ibat, the battery voltage Vbat, the working fluid temperature THoil which is the temperature of the working oil OIL in the hydraulic control circuit 56, etc.)

[0051] Various command signals (such as the engine control command signal Se for controlling the engine 12, the MG control command signal Sm for controlling the electric motor MG, the CB hydraulic control command signal Scb for controlling the engagement device CB, the K0 hydraulic control command signal Sk0 for controlling the K0 clutch 20, the LU hydraulic control command signal Slu for controlling the LU clutch 40, the EOP control command signal Seop for controlling the EOP60, etc.) are output from the electronic control unit 90 to each device provided in the vehicle 10 (such as the engine control unit 50, the inverter 52, the hydraulic control circuit 56, the pump motor 62, etc.).

[0052] To implement various controls in the vehicle 10 , the electronic control unit 90 includes a hybrid control unit 92 serving as a hybrid control means, a clutch control unit 94 serving as a clutch control means, and a shift control unit 96 serving as a shift control means.

[0053] The hybrid control unit 92 includes the functions of an engine control unit 92a, which serves as an engine control unit for controlling the operation of the engine 12, and a motor control unit 92b, which serves as a motor control unit for controlling the operation of the motor MG via the inverter 52. Through these control functions, hybrid drive control based on the engine 12 and the motor MG is performed.

[0054] The hybrid control unit 92 calculates the driver's drive demand for the vehicle 10 by applying the accelerator opening θacc and the vehicle speed V to, for example, a drive demand map. The drive demand map is a relationship that is experimentally or designed and stored in advance, that is, a predetermined relationship. The drive demand is, for example, the required drive torque Trdem at the drive wheels 14. From a different perspective, the required drive torque Trdem [Nm] is the required drive power Prdem [W] at the current vehicle speed V. As the drive demand, the required drive force Frdem [N] at the drive wheels 14, the required automatic transmission output torque at the transmission output shaft 26, etc. can also be used. In the calculation of the drive demand, the automatic transmission output speed No, etc. can also be used instead of the vehicle speed V.

[0055] The hybrid control unit 92 outputs an engine control command signal Se for controlling the engine 12 and an MG control command signal Sm for controlling the electric motor MG so as to achieve the required drive power Prdem, taking into account transmission losses, auxiliary machine loads, the gear ratio γat of the automatic transmission 24, the chargeable power Win, and the dischargeable power Wout of the battery 54. The engine control command signal Se is, for example, a command value for the power of the engine 12, i.e., the engine power Pe, which is output at the current engine speed Ne for the engine torque Te. The MG control command signal Sm is, for example, a command value for the power consumption Wm of the electric motor MG, which is output at the current MG speed Nm for the MG torque Tm.

[0056] The chargeable power Win of the battery 54 is the maximum power that can be input, which limits the input power of the battery 54 and indicates the input limit of the battery 54. The dischargeable power Wout of the battery 54 is the maximum power that can be output, which limits the output power of the battery 54 and indicates the output limit of the battery 54. The electronic control unit 90 calculates the chargeable power Win and the dischargeable power Wout of the battery 54 based on, for example, the battery temperature THbat and the state of charge value SOC [%] of the battery 54. The state of charge value SOC of the battery 54 is a value indicating the state of charge of the battery 54 and is calculated by the electronic control unit 90 based on, for example, the battery charge and discharge current Ibat and the battery voltage Vbat.

[0057] When the required drive torque Trdem can be supplied solely by the output of the electric motor MG, the hybrid control unit 92 switches the driving mode to the motor driving (EV driving) mode. In the EV driving mode, the hybrid control unit 92 performs EV driving, with the K0 clutch 20 released and the vehicle using only the electric motor MG as the driving force source. On the other hand, when the required drive torque Trdem cannot be supplied without at least the output of the engine 12, the hybrid control unit 92 switches the driving mode to the engine driving mode, or hybrid driving (HV driving) mode. In the HV driving mode, the hybrid control unit 92 performs engine driving, or HV driving, with the K0 clutch 20 engaged and the vehicle using at least the engine 12 as the driving force source. Even when the required drive torque Trdem can be supplied solely by the output of the electric motor MG, the hybrid control unit 92 establishes the HV driving mode if, for example, the state of charge (SOC) of the battery 54 is less than a predetermined engine start threshold, or if the engine 12, etc., needs to be warmed up. The engine start threshold is a predetermined threshold value of the state of charge (SOC) used to determine whether the engine 12 needs to be forcibly started to charge the battery 54. Thus, the hybrid control unit 92 switches between the EV and HV driving modes by automatically stopping the engine 12 during HV driving, restarting the engine 12 after stopping it, or starting the engine 12 during EV driving, based on the required drive torque Trdem and other factors.

[0058] The hybrid control unit 92 further includes a function of a start control unit 92 c as a start control means and a function of a stop control unit 92 d as a stop control means.

[0059] The start control unit 92c determines whether there is a request to start the engine 12. For example, in the EV driving mode, the start control unit 92c determines whether there is a request to start the engine 12 based on whether the required drive torque Trdem exceeds the range that can be supplied by the output of the electric motor MG alone, whether the engine 12 and the like need to be warmed up, or whether the state of charge value SOC of the battery 54 is less than the engine start threshold. The start control unit 92c also determines whether the start control of the engine 12 has been completed.

[0060] The clutch control unit 94 controls the K0 clutch 20 to execute engine 12 startup control. For example, if the startup control unit 92c determines that there is a request to start the engine 12, the clutch control unit 94 outputs a K0 hydraulic pressure control command signal Sk0 to the hydraulic control circuit 56 for controlling the K0 clutch 20, which is in the released state, to the engaged state to obtain K0 torque Tk0. This K0 torque Tk0 is used to transmit the torque required for starting the engine 12, which is a torque that increases the engine speed Ne, to the engine 12. In other words, when the engine 12 is started, the clutch control unit 94 outputs the K0 hydraulic pressure control command signal Sk0 to the hydraulic control circuit 56 for controlling the clutch actuator 120 to switch the control state of the K0 clutch 20 from the released state to the engaged state. In this embodiment, the torque required for starting the engine 12 is referred to as the required startup torque Tcrn.

[0061] The start control unit 92c controls the engine 12 and the electric motor MG to execute start control of the engine 12. For example, when the start control unit 92c determines that there is a request to start the engine 12, it outputs an MG control command signal Sm to the inverter 52, coinciding with the switching of the K0 clutch 20 to the engaged state by the clutch control unit 94. Specifically, the start control unit 92c outputs the MG control command signal Sm to the inverter 52 to control the electric motor MG so that the electric motor MG outputs the required starting torque Tcrn, i.e., so that the MG torque Tm is increased by the amount of the required starting torque Tcrn, when the engine 12 is started.

[0062] Furthermore, when the start control unit 92c determines that there is a request to start the engine 12, it outputs an engine control command signal Se for starting fuel supply, engine ignition, etc., to the engine control device 50 in conjunction with the start of the engine 12 via the K0 clutch 20 and the electric motor MG. In other words, when the engine 12 is started, the start control unit 92c outputs the engine control command signal Se for controlling the engine 12 so that the engine 12 begins to operate to the engine control device 50.

[0063] When the engine 12 is started, a starting reaction torque Trfcr is generated as a reaction torque associated with the engagement of the K0 clutch 20. This starting reaction torque Trfcr generates a deceleration sensation in the vehicle 10 caused by inertia during engine startup during EV driving, i.e., a drop in the drive torque Tr. Therefore, the MG torque Tm that increases toward the required starting torque Tcrn when starting the engine 12 is the MG torque Tm used to cancel the starting reaction torque Trfcr. This is the MG torque Tm used to compensate for the starting reaction torque Trfcr, i.e., the K0 reaction force compensation amount. The required starting torque Tcrn is the K0 torque Tk0 required for starting the engine 12, and is the MG torque Tm required to start the engine 12, flowing from the electric motor MG to the engine 12 via the K0 clutch 20. The required starting torque Tcrn is a predetermined, for example, constant, starting torque Tcr, based on, for example, the specifications of the engine 12 and the method used to start the engine 12.

[0064] When starting the engine 12 during EV driving, the start control unit 92c causes the electric motor MG to output the required starting torque Tcrn, in addition to the MG torque Tm for EV driving, which generates the driving torque Tr. Therefore, during EV driving, the required starting torque Tcrn must be guaranteed to support the engine 12 start. Therefore, the range within which the required driving torque Trdem can be supplied solely by the output of the electric motor MG is the torque range obtained by subtracting the required starting torque Tcrn from the maximum output torque of the electric motor MG. The maximum output torque of the electric motor MG is the maximum MG torque Tm that can be output using the dischargeable power Wout of the battery 54.

[0065] The stop control unit 92d determines whether there is a request to stop the engine 12. For example, in the HV running mode, the stop control unit 92d determines whether there is a request to stop the engine 12 based on whether the required drive torque Trdem is within the range that can be supplied by the output of the electric motor MG alone, the engine 12 and the like do not need to be warmed up, and the state of charge value SOC of the battery 54 is greater than the engine start threshold.

[0066] The stop control unit 92d controls the engine 12 so as to execute stop control of the engine 12. For example, when the stop control unit 92d determines that there is a stop request for the engine 12, it outputs an engine control command signal Se for stopping the supply of fuel to the engine 12 to the engine control device 50. In other words, when the engine 12 is stopped, the stop control unit 92d outputs the engine control command signal Se for controlling the engine 12 to the engine control device 50 so as to stop the operation of the engine 12.

[0067] The clutch control unit 94 controls the K0 clutch 20 when performing engine 12 stop control. For example, when the stop control unit 92d determines that there is a request to stop the engine 12, the clutch control unit 94 outputs a K0 hydraulic pressure control command signal Sk0 to the hydraulic control circuit 56 for controlling the engaged K0 clutch 20 to be released. Specifically, when the clutch control unit 94 determines that there is a request to stop the engine 12, i.e., when there is a request to stop the engine 12, the clutch control unit 94 outputs the K0 hydraulic pressure control command signal Sk0 to the hydraulic control circuit 56 for controlling the clutch actuator 120 to switch the control state of the K0 clutch 20 from the engaged state to the released state.

[0068] The shift control unit 96 uses, for example, a shift map representing a predetermined relationship to determine whether to shift the automatic transmission 24. As needed, the shift control unit 96 outputs a CB hydraulic pressure control command signal Scb to the hydraulic control circuit 56 for executing the shift control of the automatic transmission 24. The shift map is a predetermined relationship that includes shift lines for determining the shifts of the automatic transmission 24 on a two-dimensional coordinate system using, for example, the vehicle speed V and the required drive torque Trdem as variables. The shift map may use, for example, the AT output speed No in place of the vehicle speed V, and, for example, the required drive force Frdem, the accelerator opening θacc, the throttle valve opening θth, and the like in place of the required drive torque Trdem.

[0069] Here, in order to accurately control the control state of the K0 clutch 20 during the startup of the engine 12, a plurality of progression stages are predefined, distinguished according to the control state switched during the engagement transition of the K0 clutch 20 during the startup process of the engine 12. In other words, a K0 control stage definition Dphk0 is defined for controlling the clutch actuator 120.

[0070] Figure 3 This is a diagram illustrating an example of each stage in the K0 control stage definition Dphk0. Figure 3 In the definition of the stage Dphk0 for K0 control, there are defined stages such as "K0 waiting", "quick apply", "constant pressure waiting when components are attached", "K0 start", "quick drain", "constant pressure waiting before re-engagement", "initial rotation synchronization", "middle rotation synchronization", "final rotation synchronization", "engagement transition sweep", "full engagement transition sweep", "full engagement", "backup sweep", and "calculation stop".

[0071] The "K0 waiting" stage is a stage in which control of the K0 clutch 20 is not started but waited for when starting control of the engine 12. If a K0 waiting determination is made when starting control of the engine 12, the system shifts to the "K0 waiting" stage.

[0072] The "Quick Apply" phase is a phase in which a high command value for the K0 hydraulic pressure PRk0 is temporarily applied to quickly complete component contact of the K0 clutch 20, thereby improving the initial responsiveness of the K0 hydraulic pressure PRk0. The phase transitions to the "Quick Apply" phase when there is no K0 Wait determination at the start of engine 12 startup control. Alternatively, the phase transitions from the "K0 Wait" phase to the "Quick Apply" phase when the K0 Wait determination is canceled while waiting for the start of K0 clutch 20 control.

[0073] The command value for the K0 hydraulic pressure PRk0 is a hydraulic pressure command value for supplying the regulated K0 hydraulic pressure PRk0 to the hydraulic control circuit 56. In this embodiment, the command value for the K0 hydraulic pressure PRk0 is referred to as the K0 hydraulic pressure command value Spk0. The K0 hydraulic pressure command value Spk0 is uniquely converted into a command current for the solenoid driver of the K0 clutch 20 solenoid valve, which is provided in the electronic control unit 90. The K0 clutch 20 solenoid valve is a solenoid valve provided in the hydraulic control circuit 56 that outputs the K0 hydraulic pressure PRk0. The K0 hydraulic pressure control command signal Sk0 is the command current for the solenoid driver that drives the K0 clutch 20 solenoid valve, or the drive current or drive voltage supplied by the solenoid driver. In other words, the K0 hydraulic pressure command value Spk0 is converted into the K0 hydraulic pressure control command signal Sk0 and output to the hydraulic control circuit 56. In this embodiment, for convenience, the K0 hydraulic pressure command value Spk0 and the K0 hydraulic pressure control command signal Sk0 are treated as synonymous.

[0074] The "constant pressure waiting when components are abutting" stage is a stage of waiting at a constant pressure to complete the abutment of the components of the K0 clutch 20. When the rapid application is completed, the "rapid application" stage is transitioned to the "constant pressure waiting when components are abutting" stage.

[0075] The "K0 start" phase is a phase in which the engine 12 is started by the K0 clutch 20. When the components of the K0 clutch 20 are completely engaged, the phase transitions from the "constant pressure wait during component engagement" phase to the "K0 start" phase.

[0076] The "rapid depletion" phase is a phase in which a low K0 hydraulic pressure command value Spk0 is temporarily output to rapidly deplete the K0 hydraulic pressure, thereby improving the initial responsiveness of the K0 hydraulic pressure PRk0, while waiting at a predetermined K0 hydraulic pressure PRk0, such as the packing-end pressure PRk0pk, during the subsequent "constant pressure wait before re-engagement" phase. Upon completion of engine 12 startup and a determination that rapid depletion has been implemented, the phase transitions from the "K0 startup" phase to the "rapid depletion" phase.

[0077] The "constant pressure waiting before re-engagement" stage is a stage of waiting at a predetermined K0 torque Tk0 in a manner that does not interfere with the complete explosion of the engine 12. The complete explosion of the engine 12 is, for example, a state in which the autonomous rotation of the engine 12 is stable after the initial explosion when the ignition of the engine 12 starts. Interference that does not interfere with the complete explosion of the engine 12 means that the autonomous rotation of the engine 12 is not hindered. When the start of the engine 12 is completed and there is no judgment on the implementation of rapid consumption, the stage transitions from the "K0 start" stage to the "constant pressure waiting before re-engagement" stage. Alternatively, when rapid consumption is completed, the stage transitions from the "rapid consumption" stage to the "constant pressure waiting before re-engagement" stage.

[0078] The "initial stage of rotation synchronization" is a stage in which the K0 torque Tk0 is controlled to assist the increase of the engine speed Ne in order to quickly synchronize the engine speed Ne and the MG speed Nm. The "initial stage of rotation synchronization" is a stage in which the engine speed Ne is shifted from the "constant pressure waiting before re-engagement" stage when neither the transition conditions to the "end of rotation synchronization" stage nor the transition conditions to the "mid-stage of rotation synchronization" stage are satisfied at the time of the complete explosion notification from the engine control unit 92a. In addition, the engine control unit 92a outputs the complete explosion notification of the engine 12 when, for example, the time elapsed from the time point when the engine speed Ne reaches the predetermined complete explosion speed of the engine 12 exceeds the predetermined complete explosion notification waiting time TMeng (see below). Figure 4 For example, the complete explosion notification waiting time TMeng is predetermined in consideration of the exhaust requirements of the engine 12.

[0079] The "mid-rotation synchronization" stage is a stage in which K0 torque Tk0 is controlled so that the engine 12 reaches an appropriate increase (=Ne - Nm). The "mid-rotation synchronization" stage is transitioned from the "constant pressure waiting before re-engagement" stage when the conditions for transitioning to the "mid-rotation synchronization" stage are met upon notification of complete detonation from the engine control unit 92a. Alternatively, the "mid-rotation synchronization" stage is transitioned from the "initial rotation synchronization" stage when the conditions for transitioning to the "mid-rotation synchronization" stage are met during the execution of the "initial rotation synchronization" stage.

[0080] The "final rotation synchronization" phase is a phase in which K0 torque Tk0 is controlled to synchronize the engine speed Ne and the MG speed Nm. The "final rotation synchronization" phase transitions from the "constant pressure waiting before re-engagement" phase when the transition conditions to the "final rotation synchronization" phase are met upon notification of complete detonation from the engine control unit 92a. Alternatively, the "final rotation synchronization" phase transitions from the "initial rotation synchronization" phase when the transition conditions to the "final rotation synchronization" phase are met during the execution of the "initial rotation synchronization" phase. Alternatively, the "final rotation synchronization" phase transitions from the "mid-rotation synchronization" phase when the transition conditions to the "final rotation synchronization" phase are met during the execution of the "mid-rotation synchronization" phase. Alternatively, the "final rotation synchronization" phase transitions from the "mid-rotation synchronization" phase when the automatic transmission 24 is not in shift control and the predicted inability to synchronize the engine speed Ne and MG speed Nm persists for a period exceeding the forced rotation synchronization transfer determination time during the execution of the "mid-rotation synchronization" phase.

[0081] The "engagement transition sweep" phase gradually increases the K0 torque Tk0 to engage the K0 clutch 20. The "engagement transition sweep" phase is transitioned from the "rotation synchronization end phase" when the rotation synchronization determination is established during the execution of the "rotation synchronization end phase."

[0082] The "Complete Engagement Transfer Sweep" phase is a phase in which K0 torque Tk0 is gradually increased to bring K0 clutch 20 into a fully engaged state. Bringing K0 clutch 20 into a fully engaged state means, for example, increasing K0 torque Tk0 until a state is achieved that ensures the safety factor of engagement of K0 clutch 20. The "Complete Engagement Transfer Sweep" phase transitions from the "Engagement Transfer Sweep" phase if a K0 engagement determination is established during the "Engagement Transfer Sweep" phase. Alternatively, the "Complete Engagement Transfer Sweep" phase transitions from the "Engagement Transfer Sweep" phase if the rotational synchronization of K0 clutch 20 cannot be maintained during the "Engagement Transfer Sweep" phase. Alternatively, the "Complete Engagement Transfer Sweep" phase transitions from the "Engagement Transfer Sweep" phase if the elapsed time from the start of the "Engagement Transfer Sweep" phase exceeds a predetermined forced engagement transfer determination time and the absolute value of K0 differential rotation ΔNk0 is determined to be greater than a predetermined complete engagement transfer sweep forced transfer determination differential rotation. The K0 differential rotation ΔNk0 is the differential rotation speed of the K0 clutch 20 (=Nm−Ne).

[0083] The "fully engaged" stage maintains the fully engaged state of the K0 clutch 20. The "fully engaged" stage transitions from the "fully engaged transfer sweep" stage when a fully engaged determination is established during the "fully engaged transfer sweep" stage. Alternatively, the "fully engaged" stage transitions from the "fully engaged transfer sweep" stage when the elapsed time from the start of the "fully engaged transfer sweep" stage exceeds a predetermined forced fully engaged transfer determination time, and the absolute value of the K0 differential rotation ΔNk0 is determined to be greater than a predetermined fully engaged forced transfer determination differential rotation.

[0084] The "Fully Engaged" stage also transitions from the "Standby Sweep" stage. The "Fully Engaged" stage transitions from the "Standby Sweep" stage when, during the execution of the "Standby Sweep" stage, a fully engaged determination is established and the absolute value of the K0 differential rotation ΔNk0 is less than a predetermined differential rotation for the standby rotation synchronization determination, and such determinations are established consecutively for a predetermined number of standby rotation synchronization determinations or more. Alternatively, during the execution of the "Standby Sweep" stage, the "Fully Engaged" stage transitions from the "Standby Sweep" stage when the time elapsed from the transition to a stage other than the "K0 Wait" stage after the start of engine 12 startup control exceeds a predetermined engine startup control timeout time, and the absolute value of the K0 differential rotation ΔNk0 is determined to be greater than a fully engaged forced transfer determination differential rotation.

[0085] The "backup sweep" phase is a phase in which backup control is performed to gradually increase K0 torque Tk0 and engage K0 clutch 20. To prevent control sticking, during the execution of any of the phases, for example, the "K0 start" phase, the "constant pressure wait before re-engagement" phase, the "initial rotation synchronization" phase, the "mid-stage rotation synchronization" phase, and the "final stage rotation synchronization" phase, the "backup sweep" phase is transitioned from the currently executed phase if it is determined that the elapsed time from the start of the currently executed phase exceeds a predetermined backup transition determination time for the currently executed phase and the K0 differential rotation ΔNk0 is greater than a predetermined backup transition determination differential rotation for the currently executed phase.

[0086] The "calculation stop" phase is a phase in which calculation of the base correction pressure for the K0 hydraulic pressure PRk0 and the required K0 torque Tk0d used in engine 12 startup control is stopped during fail-safe control. This fail-safe control involves switching the oil path within the hydraulic control circuit 56 so that, if a failure occurs, such as a failure in which the K0 clutch 20 solenoid valve fails to output the regulated K0 hydraulic pressure PRk0, the clutch actuator 120 is supplied with a K0 hydraulic pressure PRk0 sufficient to maintain the fully engaged state of the K0 clutch 20, without passing through the K0 clutch 20 solenoid valve. The K0 hydraulic pressure PRk0 sufficient to maintain the fully engaged state is, for example, a source pressure such as the line pressure supplied to the K0 clutch 20 solenoid valve. The base correction pressure is a value obtained by correcting the base pressure of the K0 hydraulic pressure PRk0 used in engine 12 startup control based on, for example, the working fluid temperature THoil. The required K0 torque Tk0d is the K0 torque Tk0 required to switch the K0 clutch 20 to the engaged state in order to start the engine 12 during the start control of the engine 12 .

[0087] The K0 control phase definition Dphk0 is created, for example, to calculate the base correction pressure for the K0 hydraulic pressure PRk0 and the required K0 torque Tk0d used in starting control of the engine 12. The K0 control phase definition Dphk0 defines each phase based on the required state for controlling the K0 clutch 20, such as the desired control state of the K0 hydraulic pressure PRk0 and the K0 torque Tk0. In other words, the K0 control phase definition Dphk0 is defined based on the control requirement for switching the control state of the K0 clutch 20.

[0088] When the engine 12 is started, the clutch control unit 94 controls the clutch actuator 120 so as to switch the control state of the K0 clutch 20 from the released state to the engaged state based on the K0 control phase definition Dphk0.

[0089] When starting the engine 12, the starting control unit 92c controls the electric motor MG and the engine 12 in accordance with the control state of the K0 clutch 20. During engine 12 startup control, the electric motor MG is controlled so that it outputs the required starting torque Tcrn, and the engine 12 is controlled so that the engine 12 begins to operate. Therefore, when starting the engine 12, the starting control unit 92c controls the electric motor MG and the engine 12 according to the phases required for controlling the electric motor MG and the engine 12 in the K0 control phase definition Dphk0. This simplifies control during engine 12 startup.

[0090] Figure 4 1 is a diagram showing an example of a timing chart for executing the start control of the engine 12. Figure 4 In the K0 control phase, "K0 control phase" indicates the transition state of each phase in the K0 control phase definition Dphk0. Furthermore, the K0 hydraulic pressure command value Spk0 is output as the total hydraulic pressure value obtained by adding the hydraulic pressure value obtained by converting the required K0 torque Tk0d into the K0 hydraulic pressure PRk0 to the base correction pressure of the K0 hydraulic pressure PRk0.

[0091] Figure 4 Time t1 indicates the time when a request to start the engine 12 is received during EV driving, when the vehicle is parked in idling mode, or during EV driving, and engine 12 start control begins. After engine 12 start control begins, the "K0 Wait" phase (see time t1 - time t2), the "Quick Apply" phase (see time t2 - time t3), and the "Component Approach Constant Pressure Wait" phase (see time t3 - time t4) are executed. Following component approach control of the K0 clutch 20, the "K0 Start" phase (see time t4 - time t5) is executed.

[0092] exist Figure 4 In the embodiment, during the "constant pressure wait for component contact" phase, a K0 hydraulic pressure PRk0 equivalent to the necessary starting torque Tcrn required during the "K0 start" phase is applied. During the "constant pressure wait for component contact" phase, the actual K0 hydraulic pressure PRk0 does not rise above a value that generates the K0 torque Tk0.

[0093] During the "K0 Start" phase, the actual K0 hydraulic pressure PRk0 rises above the value required to generate K0 torque Tk0. Furthermore, during the "constant pressure wait during component contact" phase, K0 hydraulic pressure PRk0 may be applied to maintain the K0 clutch 20 in the component contact complete state. During the "K0 Start" phase, the electric motor MG outputs an MG torque Tm corresponding to the required K0 torque Tk0d, or the necessary starting torque Tcrn. During the "K0 Start" phase, as the engine speed Ne increases, engine ignition is initiated, causing the initial combustion of the engine 12. Furthermore, when starting with ignition, for example, the initial combustion of the engine 12 occurs approximately simultaneously with the start of the increase in engine speed Ne.

[0094] After the initial explosion of the engine 12, the "K0 start" phase is followed by a "rapid depletion" phase (see time t5 to time t6) and a "constant pressure wait before re-engagement" phase (see time t6 to time t7) to temporarily output a low K0 hydraulic pressure command value Spk0, so as not to interfere with the complete explosion of the engine 12. When the engine control unit 92a outputs a notification of complete engine explosion (see time t7), the "rotation synchronization initial phase" (see time t7 to time t8), the "rotation synchronization mid-phase" phase (see time t8 to time t9), the "rotation synchronization final phase" phase (see time t9 to time t10), and the "engagement transfer sweep ("engagement transfer SW" in the figure)" phase (see time t10 to time t11) are executed to synchronize the rotation of the engine 12 and the electric motor MG. Following the "engagement transfer sweep" phase, the "complete engagement transfer sweep" phase ("complete engagement transfer SW" in the figure) is executed (see time points t11 to t12), gradually increasing the K0 torque Tk0 until a state is established where a safety factor for engagement of the K0 clutch 20 is maintained. Once the K0 torque Tk0 increases until a state where a safety factor for engagement of the K0 clutch 20 is maintained is established, the "complete engagement" phase is executed (see time points t12 to t13), maintaining the complete engagement of the K0 clutch 20. Time point t13 indicates the completion of engine 12 startup control.

[0095] If reference Figure 3 、 Figure 4 During the "K0 start" phase, when the engine 12 is started, the clutch control unit 94 outputs a starting K0 hydraulic pressure command value Spk0 to the hydraulic control circuit 56 as the K0 hydraulic pressure command value Spk0 during the transition from the released state to the engaged state, i.e., during the engagement transition, when the control state of the K0 clutch 20 is switched from the released state to the engaged state. This starting K0 hydraulic pressure command value Spk0 is used to regulate the K0 hydraulic pressure PRk0 supplied to the clutch actuator 120 so that the necessary starting torque Tcrn is transmitted to the K0 clutch 20. In this embodiment, the starting K0 hydraulic pressure command value Spk0 is referred to as the starting hydraulic pressure command value Spk0cr.

[0096] If reference Figure 3 、 Figure 4During the "quick apply" phase, the clutch control unit 94 outputs a quick apply K0 hydraulic pressure command value Spk0 to the hydraulic control circuit 56 as the K0 hydraulic pressure command value Spk0 before outputting the starting hydraulic pressure command value Spk0cr during the "K0 start" phase when the engine 12 is started. This quick apply K0 hydraulic pressure command value Spk0 is used to improve the responsiveness of the K0 hydraulic pressure PRk0 to the clutch actuator 120, thereby quickly bringing the K0 clutch 20 into the component contact completion state. The quick apply phase also involves rapid filling (= rapid filling) of the oil chamber 116 of the clutch actuator 120 with the hydraulic oil OIL. Therefore, the quick apply K0 hydraulic pressure command value Spk0 is also the K0 hydraulic pressure command value Spk0 for rapid filling. In this embodiment, the K0 hydraulic pressure command value Spk0 for rapid filling is referred to as the rapid filling hydraulic pressure command value Spk0ff.

[0097] On the other hand, the K0 hydraulic pressure PRk0 and K0 torque Tk0 may deviate from the K0 hydraulic pressure command value Spk0 due to various factors. Consequently, during the engagement transition of the K0 clutch 20 during engine 12 startup control, there is a possibility that the MG speed Nm and other speeds may deviate from the target speed. Therefore, it is desirable to perform learning control related to the engagement of the K0 clutch 20 during engine startup, for example, learning control that corrects the relationship between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure command value Spk0.

[0098] Therefore, the electronic control device 90 further includes a learning control unit 98 as a learning means in order to realize a control operation for appropriately controlling the engagement of the K0 clutch 20 when the engine 12 is started.

[0099] The learning control unit 98 performs multiple types of learning control to correct the relationship between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure command value Spk0 during the engagement transition of the K0 clutch 20 related to starting the engine 12. Examples of the relationship indicating the correlation between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure command value Spk0 include, for example, the correlation between the actual K0 hydraulic pressure PRk0r, which is the actual value of the K0 hydraulic pressure PRk0, and the K0 hydraulic pressure command value Spk0; the correlation between the actual K0 hydraulic pressure PRk0r and the requested K0 torque Tk0d; the correlation between the actual K0 torque Tk0r, which is the actual value of the K0 torque Tk0, and the K0 hydraulic pressure command value Spk0; and the correlation between the actual K0 torque Tk0r and the requested K0 torque Tk0d. Learning control to correct the relationship indicating the correlation between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure command value Spk0 includes, for example, learning control to correct a deviation of the actual K0 hydraulic pressure PRk0r from the K0 hydraulic pressure command value Spk0. From a different perspective, learning control that corrects the relationship between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure command value Spk0 is learning control that corrects the deviation of the K0 hydraulic pressure command value Spk0 from the actual K0 torque Tk0r to the required K0 torque Tk0d. In this embodiment, this learning control is referred to as K0 learning control CT1rnk0. In this embodiment, unless otherwise specified, the K0 hydraulic pressure PRk0, K0 torque Tk0, and so on represent respective actual values. The deviation of the K0 hydraulic pressure PRk0 and the deviation of the K0 torque Tk0 are synonymous.

[0100] Multiple types of K0 learning control CTlrnk0 are implemented according to stages (progress stages) differentiated by the control state switched during the engagement transition of the K0 clutch 20. These multiple types of K0 learning control CTlrnk0 include, for example, learning control for correcting the rapid application time (=QA time) TMqa, which is the execution period of the rapid application in the "rapid application" phase before the generation of the K0 torque Tk0, namely, rapid filling time learning for correcting the QA time TMqa (rapid filling time). The QA time TMqa (rapid filling time) is the execution period for outputting a rapid filling hydraulic pressure command value Spk0ff to improve the responsiveness of the K0 hydraulic pressure PRk0 of the clutch actuator 120 to the K0 hydraulic pressure command value Spk0 so that the K0 clutch 20 quickly reaches the component contact completion state. The rapid filling time has the same meaning as the QA time TMqa. In this embodiment, rapid filling time learning is referred to as QA time learning CTlrnqa.

[0101] In addition, multiple types of K0 learning control CTlrnk0 include, for example: learning control for correcting the K0 hydraulic pressure PRk0, i.e., the contact completion pressure PRk0pk, during the constant pressure waiting period in the "constant pressure waiting period during component contact" phase after the "quick application" phase just before the K0 torque Tk0 is generated, and contact point learning CTlrnpk for component contact completion hydraulic pressure learning that corrects the K0 hydraulic pressure PRk0, i.e., the contact completion pressure PRk0pk, used to bring the K0 clutch 20 to the component contact completion state.

[0102] Furthermore, in this embodiment, as the K0 hydraulic pressure command value Spk0 during the "constant pressure wait for component abutment" phase, for example, a starting hydraulic pressure command value Spk0cr and a component abutment hydraulic pressure command value Spk0pk are selectively output to the hydraulic control circuit 56 according to the vehicle 10's condition. The component abutment hydraulic pressure command value Spk0pk is used to regulate the K0 hydraulic pressure PRk0 of the clutch actuator 120 for component abutment during startup of the engine 12, prior to output of the starting hydraulic pressure command value Spk0cr during the "K0 start" phase, to maintain the K0 clutch 20 in the component abutment completed state, i.e., to maintain the K0 hydraulic pressure PRk0 at the abutment completion pressure PRk0pk. Contact point learning CTlrnpk is executed when the component abutment hydraulic pressure command value Spk0pk is output during the "constant pressure wait for component abutment" phase, but is not executed when the starting hydraulic pressure command value Spk0cr is output. The component contact hydraulic pressure command value Spk0pk is output when the vehicle 10 is in a condition where, for example, even a delayed engine start is unlikely to cause discomfort to the driver, or when a starting shock is likely to occur. Conditions where even a delayed engine start is unlikely to cause discomfort to the driver include, for example, when warming up the engine 12 is required, and when starting the engine 12 is requested regardless of the driver's operating operation. Conditions where a starting shock is likely to occur include, for example, when starting the engine 12 in coordination with other controls independent of the engine 12 starting control, such as the shift control of the automatic transmission 24.

[0103] Starting hydraulic pressure command value Spk0cr is output when the vehicle 10 is in a condition where, for example, engine start delays are likely to cause driver discomfort or when starting shock is unlikely to occur. Conditions where engine start delays are likely to cause driver discomfort include, for example, when the driver's demand for vehicle 10 driving increases, resulting in a request to start engine 12. Conditions where starting shock is unlikely to occur include, for example, when engine 12 is started without coordination with other controls independent of engine 12 starting control. During the "constant pressure wait for component contact" phase, if component contact of the K0 clutch 20 is determined to be complete after a predetermined constant pressure wait duration has elapsed since the start of the "constant pressure wait for component contact" phase, the phase transitions to the "K0 start" phase. The constant pressure wait duration in the case where component contact hydraulic pressure command value Spk0pk is output is generally set to a value longer than the constant pressure wait duration in the case where starting hydraulic pressure command value Spk0cr is output.

[0104] Furthermore, various types of K0 learning control CTlrnk0 include, for example, learning control that corrects the difference in the rising timing between K0 torque Tk0 and MG torque Tm of the K0 reaction force compensation amount during the "K0 start" phase following the "constant pressure wait for component contact" phase. Specifically, this includes rising timing learning that corrects the offset between the "transmission torque rising time point at which K0 torque Tk0 increases toward the required starting torque Tcrn" and the "motor torque rising time point at which the electric motor MG begins increasing the required starting torque Tcrn." This rising timing learning also includes wasted time learning CTlrntm that corrects the wasted time TMwt from the start of the "K0 start" phase to the time when K0 torque Tk0 increases relative to the starting hydraulic pressure command value Spk0cr.

[0105] In addition, multiple types of K0 learning control CTlrnk0 include: for example, learning control for correcting the deviation of K0 torque Tk0 relative to the starting hydraulic pressure command value Spk0cr in the "K0 start" stage after the generation of K0 torque Tk0, that is, transfer torque learning CTlrntk for correcting the offset between the required starting torque Tcrn and the K0 torque Tk0 generated by the starting hydraulic pressure command value Spk0cr.

[0106] The learning control unit 98 acquires learning parameters PAlrn for performing various types of K0 learning control CTlrnk0 as needed, and calculates and updates the learning value VALlrn based on the acquired learning parameters PAlrn. The learning parameter PAlrn is, for example, a numerical value indicating a phenomenon caused by the non-convergence of the learning value VALlrn in the K0 learning control CTlrnk0. The learning value VALlrn in the QA time learning CTlrnqa is, for example, a correction value for the QA time TMqa. The learning value VALlrn in the contact point learning CTlrnpk is, for example, a correction value for the contact end pressure PRk0pk. The learning value VALlrn in the wasted time learning CTlrntm is, for example, a correction value for the wasted time TMwt. The learning value VALlrn in the transfer torque learning CTlrntk is, for example, the K0 torque Tk0 generated by the starting hydraulic pressure command value Spk0cr, a correction value of the starting hydraulic pressure command value Spk0cr for making the K0 torque Tk0 equal to the necessary starting torque Tcrn, and the like.

[0107] In principle, deviations in the K0 hydraulic pressure PRk0 and K0 torque Tk0 due to non-convergence of the learning value VALlrn, such as during the "rapid application" phase or the "constant pressure wait during component contact" phase before the generation of K0 torque Tk0, resulting in overshoot of the K0 hydraulic pressure PRk0 and generation of K0 torque Tk0, or during the "K0 start" phase, where the magnitude or generation timing of K0 torque Tk0 deviates from the target, manifest as fluctuations such as an increase or decrease in the MG speed Nm. For example, when K0 torque Tk0 is less than the target value or the generation timing of K0 torque Tk0 is delayed from the target, the MG torque Tm flowing to the engine 12 is less than the target. Due to the relatively greater MG torque Tm than the target, i.e., the excess MG torque Tm not flowing to the engine 12, the MG speed Nm increases, and the deviation in K0 torque Tk0 manifests as an increase in the MG speed Nm. The increase in the MG speed Nm is a positive value within the MG rotational fluctuation ΔNm (=Nm - Nmtgt), which is the fluctuation in the MG speed Nm. The aforementioned "Nmtgt" represents the target MG speed Nm. On the other hand, if K0 torque Tk0 is generated before K0 torque Tk0 is generated, or if K0 torque Tk0 exceeds the target value, or if K0 torque Tk0 is generated earlier than the target, a portion of the MG torque Tm of the drive torque Tr flows to the engine 12. Due to the relatively smaller MG torque Tm than the target, i.e., the insufficient MG torque Tm of the drive torque Tr, the MG speed Nm decreases. The deviation in K0 torque Tk0 manifests itself as a decrease in the MG speed Nm. The decrease in the MG speed Nm is represented by the negative value of the MG rotational variation ΔNm. As such, the MG rotational variation ΔNm varies depending on the state of the K0 clutch 20. Therefore, an example of the learned parameter PA1rn is the MG rotational variation ΔNm.

[0108] Alternatively, MG feedback control CTfbm may be performed, for example, to compensate for excess or deficiency of MG torque Tm through feedback control, maintaining MG speed Nm at target MG speed Nmtgt. When MG feedback control CTfbm is performed, the deviation of K0 torque Tk0 is expressed as, for example, the variation of compensated MG torque Tmfb after the excess or deficiency is compensated by MG feedback control CTfbm, namely, MG torque variation ΔTm (= Tmfb - Tmb). "Tmb" is the base MG torque when MG speed Nm does not deviate from target MG speed Nmtgt. Thus, since MG torque variation ΔTm varies depending on the state of K0 clutch 20, MG torque variation ΔTm is also an example of learning parameter PAlrn. Furthermore, learning parameter PALrn is obtained, for example, as the cumulative value or maximum value of variations (MG rotation variation ΔNm, MG torque variation ΔTm) during the learning period.

[0109] For example, when performing QA time learning CTlrnqa, a learning parameter PAlrn corresponding to the MG rotational variation ΔNm or MG torque variation ΔTm of the electric motor MG is acquired from the start of the "quick apply" phase to the time point at which a predetermined time α has elapsed since the completion of the "quick apply" phase. Specifically, the learning control unit 98 acquires the learning parameter PALlrn from the time the rapid filling hydraulic pressure command value Spkoff is output to the time the predetermined time α has elapsed since the output of the rapid filling hydraulic pressure command value Spkoff is completed. The predetermined time α is a value that takes into account, for example, communication delays between the solenoid driver of the electronic control unit 90 and the solenoid valve for the K0 clutch 20, as well as response delays due to the characteristics of the K0 clutch solenoid valve. In this manner, by acquiring the learning parameter PALlrn, i.e., learning the QA time TMqa, during the QA time learning CTlrnqa until the predetermined time α has elapsed since the completion of the "Quick Application" phase, the learning parameter PALlrn used to calculate the QA time TMqa can be appropriately acquired even in the event of the aforementioned communication delay or response delay. Furthermore, the predetermined time α is preferably appropriately changed based on the operating fluid temperature THoil. Specifically, considering that the lower the operating fluid temperature THoil, the lower the hydraulic response, the lower the operating fluid temperature THoil. Therefore, the predetermined time α is set to be longer as the operating fluid temperature THoil decreases.

[0110] Furthermore, the learning control unit 98 calculates a learned value VALlrn by applying a gain to a learning parameter PA1rn corresponding to the MG rotational fluctuation ΔNm or MG torque fluctuation ΔTm obtained during the QA time learning CTlrnqa. A gain G is set based on the gain processing, for example, based on the hydraulic fluid temperature THoil. The learned value VALlrn for the QA time learning CTlrnqa is calculated by multiplying the MG rotational fluctuation ΔNm or MG torque fluctuation ΔTm obtained during the QA time learning CTlrnqa by the gain G (G×ΔNm, G×ΔTm). Similarly, for the contact point learning CTlrnpk, transfer torque learning CTlrntk, and wasted time learning CTlrntm, each learning value VALlrn is calculated by applying a gain to the MG rotational fluctuation ΔNm or MG torque fluctuation ΔTm obtained during each learning period. Then, after the learning value VAL is updated (corrected) to the new learning value VALlrn, when the K0 clutch 20 is engaged next time, the K0 hydraulic command value Spk0 is corrected by adding (adding or subtracting) the new learning value VALlrn calculated using each K0 learning control CTlrnk0 to the basic K0 hydraulic command value Spk0.

[0111] Alternatively, instead of using the gain processing, each learned value VALlrn may be calculated using an offset process. The offset process calculates the learned value VALlrn for each K0 learning control CTlrnko based on a learning parameter PALlrn corresponding to the MG rotational variation ΔNm or MG torque variation ΔTm obtained during each learning period and an offset M. After the learned value VALlrn is updated (corrected) to the new learned value VALlrn, the K0 hydraulic pressure command value Spk0 is corrected by adding (adding or subtracting) the learned value VALlrn calculated using each K0 learning control CTlrnko to the base K0 hydraulic pressure command value Spk0 during the next engagement of the K0 clutch 20. As a result, by updating each learned value VALlrn, the relationship representing the correlation between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure command value Spk0 is corrected to a new relationship.

[0112] In addition, when the engagement transition period of the K0 clutch 20 is switched to the "backup sweep" phase and the K0 torque Tk0 is gradually increased by the backup control, or when the phase is switched to the "calculation stop" phase, the calculation of the learning value VALlrn by applying gain processing or offset processing to the learning parameter PALlrn is not performed.

[0113] However, during the engine 12 startup process, when the K0 clutch 20 is switched to the engaged state, the "quick apply" phase is first executed, generating the K0 hydraulic pressure PRk0. Therefore, when the QA time TMqa is modified by the QA time learning CTlrnqa and the QA time TMqa fails to converge, the K0 hydraulic pressure PRk0 overshoots, potentially affecting the characteristics of, for example, the contact end pressure PRk0pk. Furthermore, when the contact end pressure PRk0pk fails to converge, for example, the K0 hydraulic pressure PRk0 corresponding to the zero point of the K0 torque Tk0 fluctuates, potentially affecting the correlation between the starting hydraulic pressure command value Spk0cr and the K0 torque Tk0. Furthermore, when the learning value VALlrn in the transfer torque learning CTlrntk fails to converge, for example, the starting hydraulic pressure command value Spk0cr fluctuates, potentially affecting the response characteristics of the K0 torque Tk0, namely, the wasted time TMwt. Therefore, depending on the execution order of each of the multiple types of K0 learning control CTlrnk0, it is possible that after the wasted time TMwt is temporarily converged through, for example, the wasted time learning CTlrntm, the QA time TMqa is changed through the QA time learning CTlrnqa, and the wasted time TMwt does not converge again. As a result, the progress of the multiple types of K0 learning control CTlrnk0 related to the engagement of the K0 clutch 20 at engine start is delayed.

[0114] Therefore, a predetermined priority order is set for executing the various types of K0 learning control CTlrnk0 to prevent mutual deviations from influencing each other and causing erroneous learning. Furthermore, the learning value VALlrn is converged sequentially, starting with the highest-priority K0 learning control CTlrnk0, to progress through the various types of K0 learning control CTlrnk0. Considering how changes in the QA time TMqa affect the characteristics of the contact end pressure PRk0pk, for example, the highest priority order for executing the K0 learning control CTlrnk0 is, for example, QA time learning CTlrnqa. The order of priority for converging the learning value VALlrn is, for example, QA time learning CTlrnqa, followed by contact point learning CTlrnpk, transfer torque learning CTlrntk, and wasted time learning CTlrntm. This priority order is determined based on the order of the phases switched during engagement of the K0 clutch 20 and the control state of the K0 clutch 20 during the engagement transition. Furthermore, in this embodiment, when the QA time learning CTlrnqa is prioritized as the higher-order learning, the contact point learning CTlrnpk, the transfer torque learning CTlrntk, and the wasted time learning CTlrntm are prioritized as lower-order learning relative to the QA time learning CTlrnqa, which is the higher-order learning. Furthermore, when the QA time learning CTlrnqa and the contact point learning CTlrnpk are prioritized as the higher-order learning, the transfer torque learning CTlrntk and the wasted time learning CTlrntm are prioritized as lower-order learning relative to these higher-order learning. Furthermore, when the QA time learning CTlrnqa, the contact point learning CTlrnpk, and the transfer torque learning CTlrntk are prioritized as the higher-order learning, the wasted time learning CTlrntm is prioritized as lower-order learning relative to these higher-order learning.

[0115] Here, the priority for converging the learned value VALlrn does not exclude the engagement control of the K0 clutch 20 associated with starting the engine 12. Two or more K0 learning control CTlrnk0s are executed during a single engine start. Specifically, a lower-priority K0 learning control CTlrnk0 is executed regardless of whether the learned value VALlrn in the higher-priority K0 learning control CTlrnk0 has converged. Regarding the priority for converging the learned value VALlrn, if the learned value VALlrn in a higher-priority K0 learning control CTlrnk0 has not converged, the failure of the learned value VALlrn in the lower-priority K0 learning control CTlrnk0 is considered to be included.

[0116] As described above, if the learned value VALlrn of the K0 learning control CTlrnk0 (upper learning) with the highest priority has not converged, executing the K0 learning control CTlrnk0 (lower learning) with a lower priority may cause erroneous learning. Therefore, to avoid erroneous learning, it is best to wait for the learned value VALlrn of the upper learning with the highest priority to converge before executing the lower learning. However, waiting for the learned value VALlrn of the upper learning to converge makes it difficult for the lower learning to progress. In contrast, by executing the lower learning regardless of whether the learned value VALlrn of the upper learning has converged, the progress of the lower learning can be accelerated.

[0117] When executing engagement control of the K0 clutch 20 associated with starting the engine 12, the learning control unit 98 prioritizes the QA time learning CTlrnqa, which has the highest priority, among the multiple types of K0 learning control CTlrnk0. Furthermore, the learning control unit 98 determines whether the correction value of the QA time TMqa, which is the learning value VALlrn calculated through the QA time learning CTlrnqa, has converged. Next, regardless of the determination result of the QA time learning CTlrnqa, the learning control unit 98 calculates the learning values VALlrn for the contact point learning CTlrnpk, the transfer torque learning CTlrntk, and the wasted time learning CTlrntm, which are the multiple types of K0 learning control CTlrnk0, and determines whether each learning value VALlrn has converged.

[0118] Specifically, when the start control unit 92c determines that there is a request to start the engine 12, the learning control unit 98 initially determines whether to implement the K0 learning control CT1rnk0 during the current engine start control. For example, when the vehicle 10 is stable and the solenoid valve for the K0 clutch 20 is determined to be intact, the learning control unit 98 determines to implement the K0 learning control CT1rnk0. The learning control unit 98 determines whether the vehicle 10 is stable based on, for example, the vehicle speed V, the accelerator opening θacc, the gear position of the automatic transmission 24, the MG speed Nm, and the like. For example, when the vehicle 10 is not stable or when the solenoid valve for the K0 clutch 20 is determined to be malfunctioning, the learning control unit 98 determines not to implement the K0 learning control CT1rnk0.

[0119] When determining that the K0 learning control CTlrnk0 is to be performed, the learning control unit 98 preferentially executes the K0 learning control CTlrnk0 having the highest priority, for example, the QA time learning CTlrnqa.

[0120] The learning control unit 98 determines whether the learned value VALlrn in the executed K0 learning control CTlrnk0 has converged. For example, the learning control unit 98 determines whether the QA time learning CTlrnqa is in a convergence state based on whether the value of the learning parameter PAlrn obtained in the executed K0 learning control CTlrnk0 is less than a predetermined change amount K, which is used to determine that the learned value VALlrn in the QA time learning CTlrnqa has converged. The predetermined change amount K is set in advance through experiments or design, and is set to a value that can be used to determine that the deviation of the learning parameter PALlrn has converged.

[0121] Furthermore, even if the learning control unit 98 determines that the learned value VALlrn in the executed K0 learning control CTlrnk0 has not converged, the learning control unit 98 also executes a K0 learning control CTlrnk0 (lower-level learning) with a lower priority than the executed K0 learning control CTlrnk0. In other words, the learning control unit 98 executes all of the multiple types of K0 learning control CTlrnk0 executed during the engagement transition of the K0 clutch 20 related to the startup of the engine 12, regardless of whether the learned value VALlrn in the higher-priority K0 learning control CTlrnk0 (higher-level learning) has converged.

[0122] As described above, by executing multiple types of K0 learning control CTlrnko even though the learning value VALlrn in the K0 learning control CTlrnk0 with the highest priority has not converged, the progress of each learning of the K0 clutch 20 is accelerated. However, on the other hand, the possibility of erroneous learning is increased. In contrast, when the K0 learning control CTlrnk0 with the highest priority is determined to be in a non-convergent state, the learning control unit 98 reduces the gain G or offset M used when calculating the learning value VALlrn for the K0 learning control CTlrnk0 with a lower priority than the upper learning (lower learning) compared to when the upper learning is determined to be converged. In other words, when the K0 learning control CTlrnko (superior learning) with the highest priority is determined to be in a non-convergent state, the degree of reflection of the learning parameter PALlrn corresponding to the learning result obtained in the lower learning with a lower priority than the upper learning is reduced compared to when the upper learning is determined to be in a converged state. Furthermore, the fact that the K0 learning control CTlrnko is in a converged or non-converged state has the same meaning as the fact that the learned value VALlrn in the K0 learning control CTlrnko is in a converged or non-converged state. Furthermore, reducing the gain G or the offset M corresponds to reducing the degree of reflection of the learning result in the present invention.

[0123] If the K0 learning control CTlrnk0 (superior learning) with the highest priority is in a non-convergent state, the K0 learning control CTlrnk0 (subordinate learning) with the highest priority may be affected by the non-convergent state of the superior learning. Therefore, it is best to minimize the correction amount using the learning value VALlrn. Considering this, by reducing the gain G based on the gain processing or the offset M based on the offset processing compared to when the superior K0 learning control CTlrnk0 is determined to be in a converged state, the value of the learning value VALlrn of the subordinate learning is reduced, and the degree of reflection of the learning results of the subordinate learning is reduced.

[0124] In this embodiment, the learning control unit 98 applies gain processing or offset processing to the learning parameter PALlrn obtained in the QA time learning CTlrnqa, which has the highest priority, to calculate a correction value for the QA time TMqa, which is the learned value VALlrn. Furthermore, since the QA time learning CTlrnqa has the highest priority and is not affected by the other K0 learning control CTlrnk0, the gain G obtained by the gain processing and the offset M obtained by the offset processing remain the same regardless of the convergence state. Furthermore, the learning control unit 98 determines whether the QA time learning CTlrnqa is in a convergence state based on whether the value of the QA learning parameter PALlrnqa obtained by the QA time learning CTlrnqa is less than a predetermined variation K.

[0125] Next, the learning control unit 98 calculates a correction value for the contact end pressure PRk0pk, which is the learned value VALlrn, by applying gain processing or offset processing to the learning parameter PALlrn obtained in the contact point learning CTlrnpk, which has a higher priority than the QA time learning CTlrnqa. When applying gain processing or offset processing, the learning control unit 98 changes the gain G used in the gain processing and the offset M used in the offset processing, depending on whether the QA time learning CTlrnqa, which has a higher priority than the contact point learning CTlrnpk, is determined to be in a convergence state.

[0126] Specifically, when the QA time learning CTlrnqa is determined to be in a non-convergent state, the gain G (hereinafter referred to as gain Gpkn) used when calculating the correction value for the contact end pressure PRk0pk, which is the learned value VALlrn, is smaller than the gain G (hereinafter referred to as gain Gpkc) used when the QA time learning CTlrnqa is determined to be in a convergent state. Alternatively, when the QA time learning CTlrnqa is determined to be in a non-convergent state, the offset M (hereinafter referred to as offset Mpkn) used when calculating the correction value for the contact end pressure PRk0pk is smaller than the offset M (hereinafter referred to as offset Mpkc) used when the QA time learning CTlrnqa is determined to be in a convergent state.

[0127] The learning control unit 98 stores a relationship map of the gain G used when calculating the learning value VALlrn through the contact point learning CTlrnpk, and determines the gain G by applying the operating fluid temperature THoil and the like to the relationship map. Figure 5 This is an example of a relationship map between the gains Gpkc and Gpkn used to calculate the learning value VALlrn in the contact point learning CTlrnpk. Figure 5 In the figure, the horizontal axis represents the operating fluid temperature THoil, and the vertical axis represents the gain G. Figure 5 In the figure, the solid line represents the gain Gpkc used when the QA time learning CTlrnqa is in a converged state, and the single-dot chain line represents the gain Gpkn used when the QA time learning CTlrnqa is in a non-converged state. Figure 5 As can be seen from the relationship mapping, the gain Gpkn used when the QA time learning CTlrnqa is not converging is set to a smaller value than the gain Gpkc used when the QA time learning CTlrnqa is converging. Therefore, when the QA time learning CTlrnqa is not converging, the calculated learning value VALlrn is smaller than when the QA time learning CTlrnqa is converging. In other words, when the QA time learning CTlrnqa is not converging, the learning results of the contact point learning CTlrnpk are less reflected than when the QA time learning CTlrnqa is converging.

[0128] Alternatively, the learning control unit 98 stores a map of the offset amount M used when calculating the learning value VALlrn through the contact point learning CTlrnpk, and determines the offset amount M by applying the operating fluid temperature THoil and the like to the map. Figure 6 This is an example of a relationship map between the offsets Mpkc and Mpkn used to calculate the learning value VALlrn in the contact point learning CTlrnpk. Figure 6In the figure, the horizontal axis represents the operating fluid temperature THoil, and the vertical axis represents the offset M. Figure 6 In the figure, the solid line represents the bias Mpkc used when the QA time learning CTlrnqa is in a converged state, and the single-dot chain line represents the bias Mpkn used when the QA time learning CTlrnqa is in a non-converged state. Figure 6 As can be seen from the relationship map, the bias value Mpkn used when the QA time learning CTlrnqa is in a non-convergent state is set to a smaller value than the bias value Mpkc used when the QA time learning CTlrnqa is in a convergent state. Therefore, when the QA time learning CTlrnqa is determined to be in a non-convergent state, the calculated learning value VALlrn is smaller than when the QA time learning CTlrnqa is determined to be in a convergent state. In other words, when the QA time learning CTlrnqa is determined to be in a non-convergent state, the learning result of the contact point learning CTlrnpk is less reflected than when the QA time learning CTlrnqa is determined to be in a convergent state.

[0129] When calculating the learning value VALlrn in the contact point learning CTlrnpk, the learning control unit 98 determines whether the contact point learning CTlrnpk is in a convergence state based on whether the value of the learning parameter PALlrn obtained in the contact point learning CTlrnpk is less than a predetermined change amount K that is used to determine that the learning value VALlrn of the contact point learning CTlrnpk has converged.

[0130] After determining whether the contact point learning CTlrnpk is converging, the learning control unit 98 calculates the learning value VALlrn for the transfer torque learning CTlrntk, which has a lower priority than the contact point learning CTlrnpk. The learning control unit 98 applies gain processing or offset processing to the learning parameter PALlrn obtained in the transfer torque learning CTlrntk to calculate a correction value for the starting hydraulic pressure command value Spk0cr, which serves as the learning value VALlrn. When applying gain processing or offset processing, the learning control unit 98 changes the gain G used in the gain processing and the offset M used in the offset processing based on whether the QA time learning CTlrnqa, which has a higher priority than the transfer torque learning CTlrntk, and the contact point learning CTlrnpk are converging.

[0131] Specifically, when at least one of the QA time learning CTlrnqa and the touchpoint learning CTlrnpk is determined to be in a non-convergent state, the gain G (hereinafter referred to as gain Gtkn) used when calculating the correction value of the starting hydraulic pressure command value Spk0cr as the learning value VALlrn is smaller than the gain G (hereinafter referred to as gain Gtkc) used when both the QA time learning CTlrnqa and the touchpoint learning CTlrnpk are in a convergent state. Alternatively, when at least one of the QA time learning CTlrnqa and the touchpoint learning CTlrnpk is determined to be in a non-convergent state, the offset M (hereinafter referred to as offset Mtkn) used when calculating the correction value of the starting hydraulic pressure command value Spk0cr as the learning value VALlrn is smaller than the offset M (hereinafter referred to as offset Mtkc) used when both the QA time learning CTlrnqa and the touchpoint learning CTlrnpk are in a convergent state.

[0132] The learning control unit 98 stores, for example, a relationship map of the gain G used when calculating the learning value VALlrn in the transfer torque learning CTlrntk, and determines the gain G by applying the operating fluid temperature THoil and the like to the relationship map. The relationship map of the gain G used in the transfer torque learning CTlrntk is also referred to as Figure 5 The relationship map shown shows the same tendency. Specifically, the gain Gtkn used when at least one of the QA time learning CTlrnqa and the touch point learning CTlrnpk is not converging is set to a smaller value than the gain Gtkc used when both the QA time learning CTlrnqa and the touch point learning CTlrnpk are converging. Consequently, when at least one of the QA time learning CTlrnqa and the touch point learning CTlrnpk is not converging, the calculated learning value VALlrn is smaller than when both the QA time learning CTlrnqa and the touch point learning CTlrnpk are converging. In other words, when at least one of the QA time learning CTlrnqa and the touch point learning CTlrnpk is not converging, the learning results of the transfer torque learning CTlrntk are less reflected than when both the QA time learning CTlrnqa and the touch point learning CTlrnpk are converging.

[0133] Alternatively, the learning control unit 98 stores a relationship map of the offset M used when calculating the learning value VALlrn in the transfer torque learning CTlrntk, and determines the offset M by applying the operating fluid temperature THoil and the like to the relationship map of the offset M. The relationship map used in the transfer torque learning CTlrntk is also referred to as Figure 6The relationship map shown shows the same tendency. Specifically, the offset Mtkn used when at least one of the QA time learning CTlrnqa and the touch point learning CTlrnpk is not converging is set to a smaller value than the offset Mtkc used when both the QA time learning CTlrnqa and the touch point learning CTlrnpk are converging. Consequently, when at least one of the QA time learning CTlrnqa and the touch point learning CTlrnpk is not converging, the calculated learning value VALlrn is smaller than when both are converging. In other words, when at least one of the QA time learning CTlrnqa and the touch point learning CTlrnpk is not converging, the learning results of the transfer torque learning CTlrntk are less reflected than when both are converging.

[0134] When calculating the learning value VALlrn in the transfer torque learning CTlrntk, the learning control unit 98 determines whether the transfer torque learning CTlrntk is in a convergence state based on whether the value of the learning parameter PALlrn obtained in the transfer torque learning CTlrntk is less than a predetermined change amount K determined in advance for determining that the learning value VALlrn has converged.

[0135] After determining whether the transfer torque learning CTlrntk is converging, the learning control unit 98 calculates the learning value VALlrn for the wasted time learning CTlrntm, which has a lower priority than the transfer torque learning CTlrntk. The learning control unit 98 applies gain processing or offset processing to the learning parameter PALlrn obtained in the wasted time learning CTlrntm to calculate a correction value for the wasted time TMwt as the learning value VALlrn. When applying gain processing or offset processing, the learning control unit 98 changes the gain G used in the gain processing and the offset M used in the offset processing based on whether the QA time learning CTlrnqa, contact point learning CTlrnpk, and transfer torque learning CTlrntk, which have a higher priority than the wasted time learning CTlrntm, are converging.

[0136] Specifically, when at least one of the QA time learning CTlrnqa, contact point learning CTlrnpk, and transfer torque learning CTlrntk (hereinafter collectively referred to as "superordinate learning") is determined to be in a non-convergent state, the gain G (hereinafter referred to as "gain Gtmn") used when calculating the correction value for the wasted time TMwt of the learning value VALlrn is smaller than the gain G (hereinafter referred to as "gain Gtmc") used when all the superior learnings are in a convergent state. Alternatively, when at least one of the superior learnings is determined to be in a non-convergent state, the offset M (hereinafter referred to as "offset Mtmn") used when calculating the correction value for the wasted time TMwt of the learning value VALlrn is smaller than the offset M (hereinafter referred to as "offset Mtmc") used when all the superior learnings are in a convergent state.

[0137] The learning control unit 98 stores a relationship map of the gain G used when calculating the learning value VALlrn in the wasted time learning CTlrntm, and determines the gain G by applying the operating fluid temperature THoil and the like to the relationship map. The relationship map used in the wasted time learning CTlrntm also becomes the relationship map used to obtain the learning value VALlrn. Figure 5 The relationship map for the gain G shown in the figure shows the same tendency. Specifically, the gain Gtmn used when at least one of the upper-level learning methods is not converging is set to a smaller value than the gain Gtmc used when all the upper-level learning methods are converging. Consequently, when at least one of the upper-level learning methods is not converging, the calculated learning value VALlrn is smaller than when all the upper-level learning methods are converging. In other words, when at least one of the upper-level learning methods is not converging, the learning results of the wasted time learning CTlrntm are less reflected than when all the upper-level learning methods are converging.

[0138] Alternatively, the learning control unit 98 stores a relationship map of the offset M used in calculating the learning value VALlrn in the wasted time learning CTlrntm, and determines the offset M by applying the operating fluid temperature THoil and the like to the relationship map. The relationship map used in the wasted time learning CTlrntm also becomes a function of the relationship map used to obtain the learning value VALlrn. Figure 6The relationship between the bias M shown in the figure shows the same tendency. Specifically, the bias Mtmn used when at least one of the upper-level learning processes is not converging is set to a smaller value than the bias Mtmc used when all the upper-level learning processes are converging. Consequently, when at least one of the state learning processes is not converging, the calculated learning value VALlrn is smaller than when all the state learning processes are converging. In other words, when at least one of the upper-level learning processes is not converging, the learning results of the wasted time learning CTlrntm are less reflected than when all the upper-level learning processes are converging.

[0139] When calculating the learning value VALlrn in the wasted time learning CTlrntm, the learning control unit 98 determines whether the wasted time learning CTlrntm is in a convergence state based on whether the value of the learning parameter PALlrn obtained in the wasted time learning CTlrntm is less than a predetermined change K that is determined in advance for judging that the learning value VALlrn has converged.

[0140] Next, when updating (correcting) each learned value VALlrn calculated through the multiple types of K0 learning control CTlrnk0, the learning control unit 98 determines whether the learning value VALlrn is in an updateable state. Specifically, when updating each learned value VALlrn, the learning control unit 98 determines whether the learning value VALlrn is in an updateable state after the engine 12 is started during the K0 learning control CTlrnk0 and the K0 hydraulic pressure PRk0 supplied to the clutch actuator 120 is less than a predetermined value β, or the K0 hydraulic pressure command value Spk0 is not output to the hydraulic control circuit 56 (i.e., the K0 hydraulic pressure command value Spk0 is zero). The learning control unit 98 updates each learned value VAL1rn when it determines that the K0 hydraulic pressure PRk0 supplied to the clutch actuator 120 is below a predetermined value β, or that the K0 hydraulic pressure command value Spk0 is not being output to the hydraulic control circuit 56, after the engine 12 is started, while performing K0 learning control CT1rnk0. Specifically, when the above conditions are met, the correlation between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure command value Spk0 is updated. After the engine 12 is started, if the K0 hydraulic pressure PRk0 of the clutch actuator 120 is below a predetermined value β close to zero, or if the K0 hydraulic pressure command value Spk0 is not being output, hydraulic control related to the K0 clutch 20 is not performed. In this case, the load on the electronic control unit 90 is low, so the learning value VAL1rn, which is subject to a higher load, can be appropriately updated. That is, by executing the updating of each learning value VALlrn simultaneously with the execution of the hydraulic pressure control of the K0 clutch 20 by the electronic control device 90 , the influence on the hydraulic pressure control of the K0 clutch 20 is suppressed.

[0141] When the learning control unit 98 determines that the learning object is in an updateable state, it sequentially updates the learning values VALlrn, starting with the learned value VALlrn in the QA time learning CTlrnqa, which has the highest priority. Specifically, the learning control unit 98 rewrites each learned value VALlrn stored, for example, in the ROM of the electronic control unit 90 before the execution of the K0 learning control CTlrnko, with the new learned value VALlrn from the current learning.

[0142] Here, if the K0 learning control CTlrnk0 with the highest priority is determined to be in a non-convergent state, the reliability of each learned value VALlrn calculated by each K0 learning control CTlrnk0 is low. Therefore, if the value of the learning parameter PALlrn representing the amount of variation obtained by each K0 learning control CTlrnk0 exceeds a predetermined range Rp when the K0 learning control CTlrnk0 with the highest priority is determined to be in a non-convergent state, the learning control unit 98 updates the learned value VALlrn. Specifically, by rewriting the learned value VALlrn with the new value, the learning result is reflected in the relationship between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure command value Spk0. If the learning parameter PALlrn representing the amount of variation is large, the hydraulic pressure deviation is large, resulting in a large deviation from the target command value, and the reliability of the learned value VALlrn before the update is considered to be low. In this case, even if the K0 learning control CTlrnk0, which has the highest priority, is not converging, updating the learned value VALlrn can reduce hydraulic pressure deviation and accelerate overall learning progress. Furthermore, when the K0 learning control CTlrnk0, which has the highest priority, is not converging, the gain G and offset M are reduced, thereby reducing the impact of mislearning. Furthermore, the predetermined range Rp is determined and stored in advance through experiments or design, and is set to, for example, the lower limit threshold of the amount of change at which a deviation from the target value of the command value being learned becomes significant. Furthermore, the predetermined range Rp is appropriately modified based on the learning target of the K0 learning control CTlrnko. On the other hand, if the value of the learning parameter PALlrn does not exceed the predetermined range Rp, the learned value VALlrn is not updated. This reduces the frequency of mislearning caused by updating the learned value VALlrn, even if the K0 learning control CTlrnko, which has the highest priority, is not converging.

[0143] Figure 7 This flowchart is used to explain the main part of the control operation of the electronic control device 90, that is, the control operation that enables rapid progress of learning when learning the command value of the K0 clutch 20 during the start transition of the engine 12. This flowchart is repeatedly executed during the operation of the vehicle 10.

[0144] exist Figure 7In the process, first, in step S10 corresponding to the control function of the start control unit 92c (steps omitted below), it is determined whether there is a start request for the engine 12. That is, it is determined whether the start control of the engine 12 has started. If the determination of this S10 is negative, this routine ends. If the determination of this S10 is positive, in step S20 corresponding to the control function of the learning control unit 98, it is determined whether the K0 learning control CTlrnk0 is implemented at the time of this engine start. If the determination of this S20 is negative, this routine ends. If the determination of this S30 is positive, in step S30 corresponding to the control function of the learning control unit 98, the K0 learning control CTlrnk0 with the highest priority, that is, the QA time learning CTlrnqa, is preferentially executed.

[0145] use Figure 8 The flowchart of S30 illustrates the QA time learning CTlrnqa. Whenever the QA time learning CTlrnqa is executed, Figure 8 Flowchart of the process.

[0146] exist Figure 8 In S100 corresponding to the control function of the learning control unit 98, when the engine 12 is started, it is determined whether the "quick application" stage has started. If the determination of S100 becomes negative, this routine ends. If the determination of S100 becomes positive, in S110 corresponding to the control function of the learning control unit 98, the learning parameter PALlrn is obtained based on the MG rotation variation ΔNm or MG torque variation ΔTm of the electric motor MG. Next, in S120 corresponding to the control function of the learning control unit 98, it is determined whether a predetermined time α has passed from the completion time point of the "quick application" stage. If the determination of S120 becomes negative, the routine returns to S110 and continues to obtain the learning parameter PALlrn. If the determination of S120 becomes positive, in S130 corresponding to the control function of the learning control unit 98, the learning value VALlrn is calculated by performing gain processing or offset processing on the learning parameter PALlrn obtained in S110. Next, in S140, which corresponds to the control function of the learning control unit 98, it is determined whether the learning parameter PALlrn acquired in S110 is less than the predetermined variation K. If the determination in S140 is affirmative, in S150, which corresponds to the control function of the learning control unit 98, it is determined that the QA time learning CTlrnqa is in a converged state. If the determination in S140 is negative, in S160, which corresponds to the control function of the learning control unit 98, it is determined that the QA time learning CTlrnqa is not in a converged state.

[0147] Return to Figure 7In S30, when the learning value VALlrn is calculated through the QA time learning CTlrnqa, contact point learning CTlrnpk, which has a higher priority than the QA time learning CTlrnqa, is next executed. In S40, which corresponds to the control function of the learning control unit 98, a determination is made as to whether a higher-priority learning process than the current learning process has converged. For example, if only the QA time learning CTlrnqa has been learned, the contact point learning CTlrnpk becomes the current learning process, and the higher-priority learning process becomes the QA time learning CTlrnqa. In this case, in S40, when the learning value VALlrn is calculated through the contact point learning CTlrnpk, a determination is made as to whether the QA time learning CTlrnqa has converged. If the determination in S40 is negative, in S50, which corresponds to the control function of the learning control unit 98, the learning value VALlrn is calculated based on the gain G or offset M set if the higher-priority learning process has not converged. For example, when learning the learned value VALlrn of the contact point learning CTlrnpk, the learned value VALlrn in the contact point learning CTlrnpk is calculated based on the gain Gpkn or offset Mpkn set when the QA time learning CTlrnqa was in a non-convergent state. If S40 is affirmative, in S60 corresponding to the control function of the learning control unit 98, the learned value VALlrn is calculated based on the gain G or offset M set when the learning with the highest priority was in a convergent state. For example, when learning the learned value VALlrn of the contact point learning CTlrnpk, the learned value VALlrn in the contact point learning CTlrnpk is calculated based on the gain Gpkc or offset Mpkc set when the QA time learning CTlrnqa was in a convergent state.

[0148] Next, in S70, which corresponds to the control function of the learning control unit 98, a determination is made as to whether the learned values VALlrn have been calculated for all K0 learning control units CTlrnko. For example, if the learned values VALlrn have been calculated up to the contact point learning CTlrnpk, but the learned values VALlrn have not been calculated for the transfer torque learning CTlrntk and the wasted time learning CTlrntm, the determination in S70 becomes negative. In this case, the process returns to S40, and the learned values VALlrn are also calculated sequentially for the K0 learning control unit CTlrnk0, which has a lower priority. Furthermore, within the K0 learning control unit CTlrnk0, the learned values VALlrn are calculated sequentially, starting with the highest priority learning unit. Thus, once the learned value VALlrn in the contact point learning CTlrn is calculated, the learned value VALlrn in the transfer torque learning CTlrntk is calculated in steps S40 to S60. Then, the learned value VALlrn in the wasted time learning CTlrntm is calculated in steps S40 to S60. If the calculation of the learned value VALlrn in the wasted time learning CTlrntm is completed, the determination in S70 becomes affirmative.

[0149] In S80, which corresponds to the control function of the learning control unit 98, after engine 12 startup control is completed, it is determined whether the K0 hydraulic pressure PRko of the K0 clutch 20 is below a predetermined value β or whether the hydraulic pressure command value Spk0 is not being output to the hydraulic control circuit 56. If the determination in S80 is negative, the system waits for the update of each learned value VALlrn until the hydraulic pressure of the K0 clutch 20 is below the predetermined value β or the hydraulic pressure command value is not being output to the hydraulic control circuit 56. If the determination in S80 is positive, in S90, which corresponds to the control function of the learning control unit 98, the learned value VALlrn calculated through each learning step of the K0 learning control CTlrnk0 is updated. Specifically, the learned value VALlrn stored before the learning is overwritten with the new learned value VALlrn from the current learning.

[0150] use Figure 9 The control operation of S90 is described in the flowchart of FIG. Figure 9 The flowchart is executed when each learning value VALlrn calculated in each K0 learning control CTlrnk0 is updated. In addition, the update of the learning value VALlrn is also performed in order from the learning with the highest priority.

[0151] In S200, which corresponds to the control function of the learning control unit 98, a determination is made as to whether a higher-priority learning process than the current learning process has reached convergence. If the determination in S200 is affirmative, in S210, which corresponds to the control function of the learning control unit 98, the calculated learning value VALlrn is updated (corrected) to the new learning value VALlrn. That is, the learning value VALlrn stored in the electronic control unit 90 is rewritten with the new learning value VALlrn for the current learning process. If the determination in S200 is negative, in S220, which corresponds to the control function of the learning control unit 98, a determination is made as to whether the value of the learning parameter PALlrn corresponding to the variation obtained in the current learning process exceeds a predetermined range Rp. If the determination in S220 is positive, the process proceeds to S210, where the calculated learning value VALlrn is updated (corrected) to the new learning value VALlrn. If the determination in S220 is negative, in S230, which corresponds to the control function of the learning control unit 98, the learning value VAL is not updated. In S240, which corresponds to the control function of the learning control unit 98, a determination is made as to whether the updating of all learned values VALlrn has been completed. If the determination in S240 is negative, the routine returns to S200 and executes the updating of the learned values VALlrn, which are the learning values with the lowest priority. If the determination in S240 is positive, the updating of all learned values VALlrn has been completed, and this routine ends. By updating each learned value VALlrn using each K0 learning control CTlrnko in this manner, the next time the K0 clutch 20 is engaged, the updated learned values VALlrn are added (added to or subtracted from) the K0 hydraulic pressure command value Spk0, which serves as the basis for the K0 clutch 20, to appropriately correct the K0 hydraulic pressure command value Spk0.

[0152] As described above, according to this embodiment, the learning control unit 98 pre-sets priorities for multiple types of learning. When an upper-level learning with a higher priority among multiple types of learning is judged to be in a non-convergent state, the degree of reflection of the learning results of the lower-level learning with a lower priority than the upper-level learning is reduced compared to the case where the upper-level learning is judged to be in a converged state. Therefore, even if the upper-level learning is in a non-convergent state, by reducing the degree of reflection of the learning results of the lower-level learning while making the learning of the lower-level learning progress, the overall learning can be rapidly progressed while reducing the impact of erroneous learning caused by the non-convergent state of the upper-level learning.

[0153] Furthermore, according to this embodiment, when the learning parameter PALlrn corresponding to the variation obtained in the lower-level learning exceeds the predetermined range Rp, the lower-level learning can be advanced without waiting for the upper-level learning to converge. This allows for rapid overall learning while reducing the impact of erroneous learning caused by the upper-level learning not converging. Furthermore, QA time learning CTlrnqa is performed from the time the rapid filling hydraulic pressure command value Spkoff is output until the predetermined time α has elapsed from the time the output is completed. Therefore, even in the event of communication delays between the electronic control unit 90 and the linear solenoid valve controlling the K0 hydraulic pressure PRk0 of the clutch actuator 120, or response delays due to the characteristics of the linear solenoid valve, QA time TMqa can be appropriately learned. In addition, after the start of the engine 12 is completed and when the K0 hydraulic pressure PRk0 of the clutch actuator 120 of the K0 clutch 20 is less than the predetermined value β or when the hydraulic pressure command value Spk0 is not output to the hydraulic control circuit 56, the learning values VALlrn learned by the K0 learning control CTlrnk0 are updated. Therefore, when the load due to the calculation applied to the electronic control unit 90 is small, the learning values VALlrn are updated to suppress the influence on other controls.

[0154] As mentioned above, although the embodiment of the present invention was described in detail based on the drawings, the present invention can also be applied in other aspects.

[0155] For example, in the above embodiment, QA time learning CTlrnqa, contact point learning CTlrnpk, transfer torque learning CTlrntk, and wasted time learning CTlrntm are executed as the K0 learning control CTlrnk0 executed during the startup transition of the engine 12. However, these K0 learning controls CTlrnk0 are merely examples, and the learning targets of the K0 learning controls CTlrnk0 may be appropriately changed. For example, in addition to the above-described K0 learning control CTlrnk0, other K0 learning controls CTlrnk0 may be added. Alternatively, other K0 learning controls CTlrnk0 may be executed in place of the above-described K0 learning control CTlrnk0.

[0156] Furthermore, in the above embodiment, when the K0 learning control CTlrnk0, which has the highest priority, is in a non-convergent state, and when the value of the learning parameter PALlrn exceeds the predetermined range Rp, the learning value VALlrn of the lower-ranked K0 learning control CTlrnk0 is updated. However, each learning value VALlrn may be updated regardless of whether the value of the learning parameter PALlrn exceeds the predetermined range Rp. Even in this case, the gain G and bias M used to calculate the learning value VALlrn are smaller than when the upper-ranked K0 learning control CTlrnk0 is in a converged state. Therefore, even if the upper-ranked K0 learning control CTlrnk0 is in a non-convergent state, the impact of the updated learning value VALlrn is reduced.

[0157] In addition, in the above embodiment, the QA time learning CTlrnqa is learned based on the learning parameter PALlrn obtained during the period from the start of the "quick application" stage to the time point at which the predetermined time α has passed from the completion time point, but within the range that does not cause contradiction, with respect to other K0 learning control CTlrnk0, it can also be learned based on the change amount, that is, the learning parameter PALlrn, obtained during the period from the time point of switching to the learnable stage to the time point at which the predetermined time has passed from the completion time point of the stage, to correct the correlation between the K0 hydraulic pressure PRk0 and the K0 hydraulic pressure instruction value Spk0.

[0158] In addition, in the above embodiment, the learning parameter PAlrn can be a numerical value that represents a phenomenon caused by the failure of the learning value VALlrn in the K0 learning control CTlrnk0 to converge, or a numerical value that represents a phenomenon caused by the deviation of the K0 hydraulic pressure PRk0 from the K0 hydraulic pressure command value Spk0, and is not limited to the MG rotation variation ΔNm or the MG torque variation ΔTm.

[0159] Furthermore, in the above embodiment, as an example of a method for starting the engine 12, a method is described in which the engine 12 is ignited in conjunction with the start of the engine 12 in the transition state from the release state to the engagement state of the K0 clutch 20, and the engine speed Ne of the engine 12 itself is also increased. However, the method is not limited to this embodiment. For example, the method for starting the engine 12 may also include a method in which the engine 12 is started until the K0 clutch 20 is in a fully engaged state or a state close to a fully engaged state, and then the engine 12 is ignited.

[0160] In addition, in the above embodiment, the learning control unit 98 determines whether the learning value VALlrn has converged based on whether the value of the learning parameter PAlrn in the executed K0 learning control CTlrnk0 is less than the predetermined change amount K determined in advance for judging that the learning value VALlrn has converged. However, it is also possible to determine whether the learning value VALlrn has converged based on whether the learning value VALlrn based on the executed K0 learning control CTlrnk0 is less than the predetermined change amount determined in advance for judging that the learning value VALlrn has converged.

[0161] In addition, in the above embodiment, each learning value VALlrn calculated by the K0 learning control CTlrnko is updated separately. During the next engagement control of the K0 clutch 20, the K0 hydraulic command value Spk0 serving as the basis is corrected by adding (adding or subtracting) each learning value VALlrn to the K0 hydraulic command value Spk0 serving as the basis. However, after each learning value VALlrn is calculated by the K0 learning control CTlrnko, the K0 hydraulic command value Spk0 serving as the basis may be updated (corrected) according to each learning value VALlrn.

[0162] Furthermore, in the above embodiment, the gain G and the offset M do not change regardless of whether the QA time learning CTlrnqa, which has the highest priority, is in a converged state or a non-converged state. However, the gain G and the offset M of the QA time learning CTlrnqa may be changed depending on its own converged state or non-converged state. Specifically, the gain G and the offset M set when the QA time learning CTlrnqa is in a non-converged state are smaller than the gain G and the offset M set when the QA time learning CTlrnqa is in a converged state.

[0163] In the above embodiment, a planetary gear type automatic transmission is exemplified as the automatic transmission 24, but the present invention is not limited to this embodiment. The automatic transmission 24 may also be a synchromesh type parallel two-shaft automatic transmission including a known DCT (Dual Clutch Transmission), a known belt-type continuously variable transmission, or the like.

[0164] Furthermore, while the torque converter 22 is used as a fluid transmission device in the above embodiment, the present invention is not limited to this embodiment. For example, a fluid transmission device such as a fluid coupling without a torque amplification function may be used in place of the torque converter 22. Alternatively, the fluid transmission device is not essential and may be replaced with, for example, a clutch for opening the clutch.

[0165] The above is merely one embodiment, and the present invention can be implemented in various ways with modifications and improvements added based on the knowledge of those skilled in the art.

Claims

1. A control device (90) for a vehicle (10), the vehicle (10) comprising: Engine (12); a motor (MG) connected to a power transmission path between the engine (12) and the drive wheels (14) so as to transmit power; A clutch (20) is provided between the engine (12) and the electric motor (MG) in the power transmission path and is controlled by a hydraulic clutch actuator (120) to switch a control state; and The hydraulic control circuit (56) supplies the regulated hydraulic pressure (PRk0) to the clutch actuator (120). The control device (90) is characterized by comprising: a starting control unit (92c) for controlling the electric motor (MG) to increase the output torque of the electric motor (MG) by a necessary starting torque (Tcrn) when the engine (12) is started, and controlling the engine (12) to start the engine (12), wherein the necessary starting torque (Tcrn) is a torque required for starting the engine (12); a clutch control unit (94) which outputs a starting hydraulic pressure command value (Spk0cr) to the hydraulic control circuit (56) as a hydraulic pressure command value (Spk0) for supplying the hydraulic pressure (PRk0) during an engagement transition of switching the control state of the clutch (20) from a released state to an engaged state when the engine (12) is started, wherein the starting hydraulic pressure command value (Spk0cr) is a command value for regulating the hydraulic pressure (PRk0) to the clutch actuator (120) in such a manner that the clutch (20) transmits the necessary starting torque (Tcrn); and A learning control unit (98) performs a plurality of types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm) for correcting the correlation between the hydraulic pressure (PRk0) and the hydraulic pressure command value (Spk0) during the engagement transition of the clutch (20). The learning control unit (98) pre-sets priorities for the multiple types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm), and when a higher-order learning (CTlrnqa) with a higher priority among the multiple types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm) is judged to be in a non-convergent state, the degree of reflection of the learning result of the lower-order learning with a lower priority than the upper-order learning is reduced compared to the case where the upper-order learning is judged to be in a converged state.

2. The control device (90) of the vehicle (10) according to claim 1, characterized in that The learning control unit (98) causes the learning result of the lower-level learning to be reflected in a relationship indicating the correlation between the hydraulic pressure (PRk0) and the hydraulic pressure command value (Spk0) when the upper-level learning is judged to be in a non-convergent state and the change amount (ΔNm, ΔTm) that changes according to the state of the clutch (20) obtained in the learning of the lower-level learning exceeds a pre-set predetermined range (Rp).

3. The control device (90) of the vehicle (10) according to claim 1, characterized in that The plurality of types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm) are respectively performed corresponding to a plurality of progress stages distinguished according to the control state switched in the engagement transition of the clutch (20), The plurality of types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm) include learning that corrects the relationship representing the correlation between the hydraulic pressure (PRk0) and the hydraulic pressure command value (Spk0) based on a variation (ΔNm, ΔTm), the variation (ΔNm, ΔTm) being obtained during a period from a time point when the learnable progression phase is switched to a time point when a predetermined time (α) has passed since the time point when the progression phase is completed, and varying according to the state of the clutch (20).

4. The control device (90) of the vehicle (10) according to claim 3, characterized in that The plurality of types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm) include rapid filling time learning (CTlrnqa) for learning a rapid filling time (TMqa), the rapid filling time (TMqa) being an execution time for outputting a rapid filling hydraulic pressure command value (Spk0ff), the rapid filling hydraulic pressure command value (Spk0ff) being used to improve the responsiveness of the hydraulic pressure (PRk0) to the clutch actuator (120) in such a manner that the clutch (20) quickly reaches a component contact completion state, The rapid filling time learning (CTlrnqa) learns the rapid filling time (TMqa) based on the change (ΔNm, ΔTm) obtained during the period from the time point when the rapid filling hydraulic command value (Spk0ff) is output to the time point when the predetermined time (α) has passed since the output of the rapid filling hydraulic command value (Spk0ff) is completed.

5. The control device (90) of the vehicle (10) according to claim 2, characterized in that The plurality of types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm) are respectively performed corresponding to a plurality of progress stages distinguished according to the control state switched in the engagement transition of the clutch (20), The plurality of types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm) include learning that corrects the relationship representing the correlation between the hydraulic pressure (PRk0) and the hydraulic pressure command value (Spk0) based on a variation (ΔNm, ΔTm), the variation (ΔNm, ΔTm) being obtained during a period from a time point when the learnable progression phase is switched to a time point when a predetermined time (α) has passed since the time point when the progression phase is completed, and varying according to the state of the clutch (20).

6. The control device (90) of the vehicle (10) according to claim 5, characterized in that The plurality of types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm) include rapid filling time learning (CTlrnqa) for learning a rapid filling time (TMqa), the rapid filling time (TMqa) being an execution time for outputting a rapid filling hydraulic pressure command value (Spk0ff), the rapid filling hydraulic pressure command value (Spk0ff) being used to improve the responsiveness of the hydraulic pressure (PRk0) to the clutch actuator (120) in such a manner that the clutch (20) quickly reaches a component contact completion state, The rapid filling time learning (CTlrnqa) learns the rapid filling time (TMqa) based on the change (ΔNm, ΔTm) obtained during the period from the time point when the rapid filling hydraulic command value (Spk0ff) is output to the time point when the predetermined time (α) has passed since the output of the rapid filling hydraulic command value (Spk0ff) is completed.

7. The control device (90) of a vehicle (10) according to any one of claims 1 to 6, characterized in that: After the start of the engine (12) in which the learning is performed is completed, and when the hydraulic pressure (PRk0) supplied to the clutch actuator (120) of the clutch (20) is below a predetermined value or when the hydraulic pressure command value (Spk0) is not output to the hydraulic control circuit (56), the learning control unit (98) updates the relationship representing the correlation between the hydraulic pressure (PRk0) and the hydraulic pressure command value (Spk0) based on the learning results of each of the multiple types of learning (CTlrnk0, CTlrnqa, CTlrnpk, CTlrntk, CTlrntm).

Images